Open Forum #29 Advancing Digital Inclusion Through Segmented Monitoring

Summary

This discussion focused on advancing digital inclusion through improved segmentation of data collection for better and more targeted decision-making. The panel, moderated by Morten Meyerhoff Nielsen from UNU eGov, brought together experts from UNESCO, Research ICT Africa, the Global Digital Inclusion Partnership (GDIP), and CETIC Brazil to address the challenge that about one-third of the world’s population remains excluded from meaningful internet access.

The panelists emphasized that current frameworks for measuring digital inclusion are often supply-oriented, focusing on basic questions like “have you used the internet in the last 12 months” rather than examining the quality and type of digital activities. Guilherme from UNESCO highlighted their Internet Universality Indicators based on five pillars: rights, openness, accessibility, multi-stakeholderism, and cross-cutting elements like gender. Onica from GDIP stressed the importance of understanding gender-specific barriers through both quantitative and qualitative research, noting that national averages often fail to serve women and marginalized communities effectively.

The discussion revealed that segmented data collection helps identify hidden gaps and context-specific barriers that wouldn’t emerge from general surveys. For example, research in South Africa showed that digital centers built for rural women went unused due to safety concerns and incompatible operating hours with women’s daily routines. Fabio from CETIC Brazil shared how their multi-stakeholder approach to data collection, funded through domain name registry profits, allows for continuous monitoring and policy-relevant insights.

Key challenges identified included ensuring data privacy and dignity for marginalized communities, preventing data from reinforcing existing inequalities, and making research sustainable and accessible. The panelists agreed that effective digital inclusion requires moving beyond individual-focused metrics to understand collective and household-level dynamics, while balancing innovation with rigorous traditional methodologies to create actionable insights for policymakers.

Keypoints

## Major Discussion Points:

– **Need for Better Data Segmentation**: The panel emphasized moving beyond basic “have you used the internet” metrics to more nuanced segmentation by gender, income levels, geography, education, age, and disability status. This granular data is essential for identifying specific barriers and designing targeted interventions rather than one-size-fits-all solutions.

– **Qualitative vs. Quantitative Approaches**: There was strong consensus that quantitative data alone is insufficient. The discussion highlighted the importance of combining statistical data with qualitative research, policy ethnography, and community engagement to understand the lived experiences and contextual barriers that prevent meaningful digital inclusion.

– **Privacy and Ethical Data Collection**: A significant portion focused on balancing the need for detailed segmentation with privacy protection and dignity of marginalized communities. The panel discussed using anonymized data, local participation in data collection, and decolonized research approaches to build trust and ensure ethical practices.

– **Institutional Models and Sustainability**: The conversation explored different funding and organizational models for sustained data collection, including Brazil’s CETIC model funded by domain registry fees, partnerships with telecommunications companies, and the challenge of making segmented data collection financially sustainable long-term.

– **From Data to Policy Action**: The panel addressed the critical gap between collecting good data and translating it into effective policy interventions. They emphasized the need for capacity building among policymakers, accountability frameworks, and ensuring that data insights actually reach and benefit the communities being studied.

## Overall Purpose:

The discussion aimed to explore how improved segmentation of data collection can advance digital inclusion by enabling better-targeted decision-making. The panel sought to identify best practices for collecting, analyzing, and utilizing granular data to address the digital divide more effectively, while ensuring ethical approaches that respect marginalized communities.

## Overall Tone:

The discussion maintained a collaborative and constructive tone throughout, with panelists building on each other’s insights rather than debating. The tone was professional yet passionate, reflecting the participants’ deep commitment to digital inclusion. There was a notable shift toward more cautionary and nuanced thinking as the conversation progressed, particularly around privacy concerns and the potential for data to reinforce existing inequalities. The panel became increasingly focused on practical implementation challenges and sustainability concerns as the discussion evolved.

Speakers

**Speakers from the provided list:**

– **Morten Meyerhoff Nielsen** – Moderator from UNU eGov (United Nations University eGovernance programme)

– **Guilherme Canela De Souza Godoi** – From UNESCO’s Information for All Program (IFAP)

– **Pria Chetty** – From Research ICT Africa, lawyer with expertise in data collection and digital inclusion research

– **Onica Makwakwa** – From Global Digital Inclusion Partnership/Program (GDIP), works on gender-focused digital inclusion research

– **Fabio Senne** – From CETIC Brazil (Brazilian Internet Steering Committee’s Center for Studies on Information and Communication Technologies), focuses on digital inclusion data and statistics

– **Carmen Ferri** – Online moderator from Global Digital Inclusion Program

– **Audience** – Various audience members who asked questions during the session

**Additional speakers:**

– **Kiho Oshima** – Master’s student at University of Bremen in Germany, studying digital media and society

# Advancing Digital Inclusion Through Improved Data Segmentation: A Comprehensive Panel Discussion Report

## Introduction and Context

This panel discussion, moderated by Morten Meyerhoff Nielsen from the United Nations University eGovernance programme (UNU eGov), brought together leading experts to address one of the most pressing challenges in digital development: how to advance digital inclusion through improved segmentation of data collection for better and more targeted decision-making. With approximately one-third of the world’s population remaining excluded from meaningful internet access, the discussion focused on moving beyond traditional binary connectivity measures to develop more nuanced approaches to understanding and addressing digital divides.

The panel featured distinguished speakers from key international organisations: Guilherme Canela De Souza Godoi from UNESCO’s Information for All Program (IFAP), Pria Chetty from Research ICT Africa, Onica Makwakwa from the Global Digital Inclusion Partnership (GDIP), and Fabio Senne from CETIC Brazil. Carmen Ferri served as online moderator, facilitating questions from virtual participants. It should be noted that Guilherme left early in the session, so later discussions primarily involved the remaining three panelists.

## Current Limitations of Digital Inclusion Measurement

### The Inadequacy of Basic Connectivity Metrics

The discussion began with a fundamental critique of existing digital inclusion measurement frameworks. Morten established the central problem by highlighting that current frameworks are predominantly supply-oriented, focusing on basic questions such as “have you used the internet in the last 12 months” rather than examining the quality and type of digital activities that constitute meaningful connectivity.

Fabio Senne provided compelling evidence of this measurement gap through Brazil’s experience. While Brazil reports 90% basic internet access, their meaningful connectivity indicators reveal a starkly different reality: only 22% of the total population meets the criteria for meaningful connectivity when factors such as device availability, connection quality, affordability, and digital skills are considered. This dramatic difference between surface-level statistics and deeper analysis became a cornerstone of the discussion, demonstrating how conventional metrics can mask significant inequalities.

### The Problem with National Averages

Onica Makwakwa articulated a particularly powerful critique of aggregate data approaches, stating that “national averages are just simply not serving women. They are not serving everyone else as well.” She provided concrete examples from South Africa, explaining that any affordability analysis conducted without stratifying income quantiles would produce over-inflated outcomes that fail to represent those at the bottom of income distributions, noting that more than half of the population lives on less than half of the gross national income.

This observation highlighted how statistical methodology itself can perpetuate inequality by masking the experiences of the most vulnerable populations. The critique extended beyond simple statistical concerns to fundamental questions about whose experiences are made visible or invisible through data collection choices.

## Institutional Models and Frameworks

### UNESCO’s Comprehensive Approach

Guilherme Canela De Souza Godoi presented UNESCO’s work through three main areas. First, the Internet Universality Indicators based on five pillars: rights, openness, accessibility, multi-stakeholderism, and cross-cutting elements including gender. Second, mandatory monitoring exercises that provide member states with evidence-based frameworks for policy development. Third, guidance documents that help countries implement comprehensive digital inclusion measurement approaches.

The UNESCO framework represents an attempt to standardise comprehensive digital inclusion measurement while maintaining flexibility for local contexts, moving beyond simple connectivity measures to encompass the broader ecosystem of factors that enable meaningful digital participation.

### Brazil’s Multi-Stakeholder Model

Fabio Senne described CETIC Brazil’s innovative institutional model, which uses a multi-stakeholder approach with expert groups to define measurement priorities and adjust data production to meet decision-maker demands. Significantly, this model is funded through domain name registry profits, providing sustainable financing that enables continuous monitoring and policy-relevant insights.

The Brazilian model demonstrates how institutional design can support comprehensive data collection by creating stable funding mechanisms and ensuring that research priorities align with policy needs. Fabio also mentioned their C-MAT system for testing broadband quality in Brazilian schools, providing real-time connectivity quality monitoring.

### International Recognition and Coordination

Fabio noted that the G20 has recognised the need for segmented monitoring beyond basic connectivity, mentioning reports produced during both Brazilian and South African presidencies that require disaggregation by demographic, economic, and geographic variables. This international recognition suggests growing consensus among major economies about the limitations of traditional connectivity metrics.

## The Case for Segmented Data Collection

### Gender-Disaggregated Analysis

The panel devoted considerable attention to gender-specific barriers that emerge only through segmented data collection. Onica shared research findings showing that while overall figures might suggest minimal gender gaps in basic internet access, segmented analysis reveals significant disparities. Fabio mentioned that Brazil shows a 10% gap between women and men in meaningful connectivity despite similar basic access rates.

More importantly, segmented data collection reveals gender-specific barriers that regular surveys miss entirely. These include affordability constraints that disproportionately affect women, safety concerns about accessing digital services, and monitoring by family members that restricts women’s digital autonomy. Onica provided a striking example of digital centres built for rural women that went unused because they failed to account for women’s daily schedules and safety concerns about walking to the centres.

### Geographic and Socioeconomic Segmentation

The discussion revealed sophisticated thinking about geographic classification that moves beyond simple rural-urban binaries. Onica noted that peri-urban populations often experience challenges similar to rural communities due to urban inequality, requiring more nuanced geographic classification systems.

Fabio added that geographic disaggregation within cities reveals unexpected patterns, providing an example from São Paulo where they found neighborhoods with high connectivity but low socioeconomic status, challenging assumptions about urban digital inclusion. Perhaps most surprisingly, Fabio revealed that most disconnected people are actually located in urban areas due to population concentration, fundamentally challenging conventional wisdom about where digital exclusion occurs.

### Revealing Hidden Barriers Through Qualitative Approaches

The speakers consistently emphasised that quantitative data alone is insufficient for understanding digital inclusion challenges. Onica advocated for qualitative approaches and policy ethnography to surface hidden gaps and context-specific barriers that quantitative surveys cannot capture. Pria described Research ICT Africa’s after-access research, which combines quantitative and qualitative methods to understand usage patterns, digital literacy levels, and trust issues that affect meaningful connectivity.

These qualitative approaches reveal barriers that would never emerge from standard surveys, such as cultural norms around technology use, intergenerational dynamics affecting device access, and community-specific safety concerns that influence digital participation.

## From Individual to Collective Approaches

### Rethinking the Unit of Analysis

Fabio Senne introduced a significant shift by arguing that digital inclusion should be viewed as collective rather than individual challenges. He noted that most research interviews one individual and tries to think about digital inclusion as an individual characteristic, but “most of the problems are collective problems.”

CETIC Brazil has begun calculating ratios of people per device and examining what percentage of household income is required for device access, thinking of “the household as a collective of people rather than individual.” This approach reveals important dynamics like device-sharing patterns and income allocation decisions that individual-focused measures miss entirely.

### Community Networks and Collective Solutions

The collective approach extends beyond households to consider community networks, schools, and libraries as digital inclusion infrastructure. Onica provided an example of successful collective intervention: a tablet per household programme in Uganda that successfully empowered female-led households, with unexpected benefits for children’s education. This example demonstrated how household-level interventions can create ripple effects that benefit multiple family members.

## Methodological Innovation and Alternative Data Sources

### Combining Traditional and Innovative Approaches

The panel explored various approaches to methodological innovation while maintaining rigorous standards. Fabio advocated for combining surveys with geospatial data and other sources, highlighting the potential of citizen-generated data and satellite data as complementary sources that can be combined with traditional surveys.

However, the speakers consistently emphasised the importance of balancing innovation with traditional rigorous methodologies. Pria warned that there are “high levels of interest in the data, but not always in the process to collect the data,” highlighting the risk that methodological shortcuts could undermine data quality and community participation.

### Leveraging Existing Data Infrastructure

Onica identified national census data as an underutilised resource that could include digital-related questions with proper engagement of census bureaus. This approach could leverage existing, well-funded data collection infrastructure that occurs regularly across countries, potentially reducing the burden and cost of separate digital inclusion surveys.

The speakers also discussed the potential for partnerships with telecommunications operators, who hold valuable segmented data that could support policy decisions through partnerships with regulators. However, these partnerships raise questions about data governance and ensuring that commercial interests align with public good objectives.

## Ethical Considerations and Community Participation

### Privacy and Dignity Concerns

A significant portion of the discussion focused on balancing the need for detailed segmentation with privacy protection and dignity of marginalised communities. Carmen Ferri posed a critical question: “How can we ensure that the segmented data collection respects the privacy and dignity of marginalised communities?”

Pria explained that while anonymised data collection for policy purposes typically falls outside personal data protection regulations, it should still follow ethical standards. She advocated for local participation in data collection to address privacy concerns and ensure willing participation from communities. However, she also expressed caution about data aggregation, noting the potential for “massive harm” when data is brought together inappropriately.

### Decolonised Research Approaches

Onica strongly advocated for decolonised approaches to data collection, emphasising the importance of working with local partners rather than having external researchers study communities. She provided a striking critique of problematic research narratives, specifically mentioning studies that compare mobile phone ownership to toothbrush ownership in ways that expose “the ignorance of the researcher themselves” by failing to recognise alternative approaches to dental health.

This critique highlighted how researchers’ cultural blind spots can lead to problematic narratives about the communities they study. Onica consistently emphasised that communities should be involved in leading data collection processes rather than being passive subjects of external research.

### Preventing Data Misuse

Kiho Oshima, a master’s student from the University of Bremen, raised important questions about preventing segmented data from being used to reinforce marginalisation. This concern reflects the fundamental tension between needing data about vulnerable populations to help them while simultaneously protecting them from potential harms.

Onica provided a concrete example of this risk, describing how South Africa’s COVID-19 tracking app led to people receiving political SMS messages, demonstrating how well-intentioned data collection can be misused. The speakers acknowledged this tension but emphasised that ongoing vigilance and community participation in data governance are essential.

## Sustainability and Implementation Challenges

### Funding and Incentive Structures

Pria identified sustainability as a critical challenge, noting the disconnect between high levels of interest in data and limited willingness to invest in rigorous collection processes. She warned that without building compelling incentive structures for private sector participation, “this work will not be sustainable, because it will be replaced by quicker technical measures that don’t necessarily have the rigour attached to it.”

The discussion explored various funding models, with Brazil’s domain registry funding serving as one successful example of sustainable financing. However, most countries lack similar dedicated funding mechanisms, creating ongoing challenges for maintaining comprehensive data collection efforts.

### Building Private Sector Partnerships

Despite typical concerns about private sector data control, there was consensus that partnerships with private sector entities, particularly telecommunications companies, are essential for sustainable and comprehensive data collection. However, building these partnerships requires creating compelling incentive structures that align commercial interests with public good objectives.

## Key Areas of Agreement and Ongoing Tensions

### Methodological Consensus

The speakers demonstrated strong consensus on fundamental methodological issues. All agreed that traditional binary connectivity measures are insufficient and that comprehensive segmented data collection is essential. There was strong alignment on the importance of qualitative approaches and local context in data collection, with consistent emphasis that quantitative data alone cannot capture the complexity of digital inclusion challenges.

### Approaches to Data Integration

The main area of tension emerged around data aggregation and sharing approaches. Pria expressed strong caution about bringing data together, warning about the potential for “massive harm” and vulnerabilities in data lakes, emphasising the need for deliberate and well-intentioned processes with robust accountability mechanisms.

In contrast, Onica advocated more optimistically for mapping and coordinating existing data sources, focusing on the practical benefits of reducing survey burden and improving research efficiency. This disagreement reflects different risk tolerances and approaches to data coordination.

## Practical Recommendations and Next Steps

### Immediate Actions

Morten noted that the panel discussion would be summarised and shared online, with panellists having the opportunity to comment on the draft. Participants were encouraged to reach out to panellists and their organisations for further questions and collaboration.

### Methodological Innovations

The speakers suggested several practical approaches: using layered data approaches where national-level data provides ‘heat maps’ to identify problem areas before diving deeper with contextual analysis; combining multiple data sources rather than relying on single collection methods; and engaging with national census bureaus to include digital-related questions.

### Governance and Ethics

The discussion pointed toward applying data protection standards to non-personal data as a precautionary measure, even when not legally required. There was emphasis on creating data mapping exercises within countries to coordinate existing data sources rather than centralising all data in one location.

## Conclusion

This comprehensive panel discussion revealed both the urgent need for and the significant challenges involved in advancing digital inclusion through improved data segmentation. The speakers demonstrated remarkable consensus on the inadequacy of current measurement approaches and the necessity of moving toward more nuanced, community-participatory, and ethically grounded data collection methods.

The discussion successfully challenged fundamental assumptions about how digital inclusion should be measured and understood, from questioning the value of national averages to advocating for collective rather than individual approaches to analysis. The speakers provided compelling evidence that current frameworks mask significant inequalities and fail to capture the lived experiences of marginalised communities.

However, the discussion also revealed the complexity of implementing better approaches. Questions about sustainability, privacy protection, preventing data misuse, and balancing innovation with rigour remain ongoing challenges. The tension between the need for detailed segmentation and the protection of vulnerable communities represents a continuing challenge that requires sustained attention and innovation.

The panel’s emphasis on decolonised approaches, community participation, and collective solutions suggests that the field is moving toward more equitable and effective approaches to addressing digital divides. However, translating these insights into sustainable, scalable, and ethically sound data collection systems remains the critical challenge for advancing digital inclusion in practice.

Morten Meyerhoff Nielsen: So, welcome everybody. My name is Morten. I’m from UNU eGov. I’ll be moderating the next 19 minutes. We’re just going to wait a couple of minutes because we see that the coffee breaks are still on. But then we’ll line up. Just a little bit of household. We’ll have an online moderator, so please don’t hesitate to put questions or observations into the chat. And Carmen, our colleague, will chime in when she’s prompted with highlighting some of the questions and observations that we can then discuss in this panel. We will have a couple of rounds of discussion. We will start with the panel, obviously, but we’ll subsequently open up the floor both here in Lilleström and online. Okay. some more people coming in, but let’s get cracking. So again, welcome to this open forum. We’ll be looking at ways to advance digital inclusion through improved segmentation of data collection for better and more targeted decision-making. We have a number of excellent panelists, two of which unfortunately will not join us. Helani from Lina Asia is unfortunately stuck somewhere in space because of issues with flights over the Middle East and Waleed Hamdi from the African Union’s Information Systems Department unfortunately has similar challenges with getting from A to B due to some Middle Eastern conflict issues. So they sent their sincere apologies, but they will be commenting on the report and the feedback of this session after the fact. We have an excellent panel otherwise. We have from the far left, we have Guilherme, he’s from UNESCO. Then we have Pria, she’s from Research ICT Africa. We have Onica from the Global Digital Inclusion in Partnership, or is it program? GDIP. GDIP. And then we have also Fabio from CETIC Brazil. My name is Morten, I’ll be basically facilitating. Now, a couple of things that I’d like to start to set the scene. I think we all agree the internet is not new. It’s been around since the previous millennium even, but really took off in the late 1990s, early 2000s. That said, about a third of the world’s population is not yet meaningfully included in the World Wide Web or the opportunities of such. Now, we see that there’s some segmentation differences on that. We see that generally in the Global South. low-income households, rural areas, seniors, people who are in unique situations or have physical disabilities or even gender segmentation are factors in relation to that, and we’ll dive into that. Similarly, we see that most frameworks promoting digital inclusion tend to recognize the problem but not really measuring it. We see a little bit of a hen and an egg situation that in communities that have the biggest potential community of excluded people are the ones that have the weakest data. This is particularly emerging economies, low-income countries in particular. We also see that these frameworks are often looking at annual assessment cycles. We’ve had some earlier workshops this week already with examples of how to increase that segmentation or those cycles to be more active in terms of giving quicker snapshots for decision makers to target the initiatives. But what they have in common is that they’re still very much supply orientated, as in have you used the internet in the last 12 months, yes or no, limited focus on the type of activities, the type of demands that we’re looking for in terms of gauging the inclusion or the degree of people’s use of digital opportunities. So again, without this knowledge, how can we as decision makers from the public sector, the private sector, from civil society or research community propose more targeted initiatives that meaningfully aim to include those who are not yet included? If we don’t know who they are, where they live or their features, how can we develop policy initiatives or charity initiatives or technical initiatives or capacity initiatives to get them included? So this is some of the elements that we are looking at. We will have an active discussion on this in the coming hour or so. But let’s get cracking with some questions to the panel. Guilherme, from the IFAP program at UNESCO, you are developing a set of data segments to monitor not just digital inclusion or exclusion, but also other things. How do UNESCO promote that as a global standard? And what are the type of things that you find is really interesting to compare across different national contexts or socio-economic contexts from your perspective? And do you have an example of how that has led to better policy initiatives?

Guilherme Canela De Souza Godoi: Thank you, Martin. I’m very glad to be here, first and foremost, because I do think these dynamic coalitions under the umbrella of IGF is also a sort of multilateral policy concrete tool for these kind of interactions. If you see the members of these coalitions, some of them are the speakers here, but many others that are not in this room necessarily, they provide different pieces to this puzzle. Therefore, these kind of discussions and also the concrete outputs that this can generate are already something very relevant for this. But on your concrete question, the Information for All Program is an intergovernmental program under the umbrella of the broader UNESCO governing bodies. And, of course, what the program does is to leverage the different aspects of the multilateral policy that is approved by our different governing bodies, our member states. So I could speak here a lot on different elements that could help to respond to your question. Let me take two or three minutes. . So, let’s start with three examples that are in a way complementary. So, the first one that’s probably more well-known is the Internet universality indicators. Something that was approved many years ago by all UNESCO member states have been refined in different moments. I guess Fabio will speak about that. So, this is a concrete set of indicators based on these five pillars, rights, openness, accessibility, multi-stakeholderism, and the acts are very several cross-cutting elements, gender, children, and so on. And so, this is a concrete set of indicators proposed and validated by a multi-lateral organization like UNESCO, but does not necessarily mean that UNESCO needs to apply it. What we are offering is something that then the different actors can use, either if they are governments and they want to use that to prepare and produce better policies, evidence-based, or if they are a civil society that wants to hold governments accountable, to tell them, you are not doing what yourselves are saying for, like, UNESCO in Paris, or in the UN in New York, or Geneva, and so on, and then, in some cases, we are inviting them to apply it. So, this is a concrete set of indicators, and we are using this set of indicators that are provided by our member states to help with the implementation. But the first thing is this, comprehensive set of indicators. I’m using the example of the ROM-X, but there are several other things connected with meaningful connectivity, sorry, for the redundancy, for example, the RAMs, the readiness assessment methodology for the information, the recommendation of the IGFs that are completed in Oslo, but I don’t want to put it as it is. There is this very important historic issue of the RIVE convention connecting all of these issues with a diversity of cultural expression. So, this is one type of logic. The other type, I will use the title is contact resolution strategy, basically. is when we have mandatory monitoring with our member states based on the things they have approved. So, for example, there is a 2003 recommendation on the multilingualism on the cyberspace. We know that multilingualism is also a critical element of meaningful connectivity. So, this recommendation, every four years, the member states need to report back to UNESCO what they are doing. So, it’s not a concrete data set per se, but what we collect from this mandatory exercise can become that and then be used by the different stakeholders as they see. For example, right now we are in the middle of the international decade of indigenous languages. This kind of data we collect through the 2003 recommendation or the World Atlas of Languages, etc., are fundamental for that. There are other recommendations like that. For example, there is one on documental heritage. So, all the issues of preserving digital heritage, every four years, the member states also need to report back on that. And then finally, there is the guidance related to this need to keep monitoring and evaluating and producing, for example, risk assessments and so on. More recently, UNESCO launched this document that I’m sure several of you heard about that is the Guidelines for the Governance of Digital Platforms. That document is suggesting concrete ways for the different stakeholders to produce risk assessments, to look into what’s happening in the digital ecosystem from that perspective of protecting and promoting freedom of expression. So, in a nutshell, and I finish, we have a concrete set of indicators, we have the mandatory monitoring exercises, and we have this guidance for them. I could keep speaking here on different ways that this impacted reality, but I must say that in the the 40 countries that already have implemented their own acts, several of them used this to change legislation, to then fulfill the gaps that were identified by the application of these kind of indicators, for example. The last thing I want to say to all of you is good news and bad news. The bad news is that I need to leave because I need to open another session. The good news is that this will offer more time for more intelligent people than myself to discuss with you. But thank you very much.

Morten Meyerhoff Nielsen: Thank you, Guilherme. And thank you for warning us before we started. Fully understandable that you’re going to be hectic. I’ll take the opportunity to quickly jump to another angle. And that’s for you, Onica. So from the perspective of the GDIP, how can segmented data, particularly related to gender and things like affordability, safety and trust concerns, digital literacy, but also classical literacy, how can we be better at that? And what are you doing at the GDIP in order to promote that segmentation of data collection for better and more targeted decisions? Yeah, good question.

Onica Makwakwa: Thank you very much for that. And good day to everyone here and online. You know, our premier research that we’ve done in this area is the Connected Resilience Report that looks at gendered experiences of women through meaningful connectivity, taking an approach of both quantitative research as well as qualitative and doing some policy ethnography as well to really understand what’s happening with women. This segmented framework actually became a powerful tool for us to be able to detect and address very gender-specific barriers that women are experiencing through meaningful connectivity, such as affordability. and many others. So, I would like to start by saying that we have seen a lot of connectivity, safety concerns, as well as digital literacy issues that came up quite strongly in that report. And, you know, ensuring that the programs that are then implemented truly focus on women and digital technologies are targeting the needs that women have identified to effectively address the inequities that women have identified. And the second area for us that the segmentation actually has shown importance in is that it helps us to surface hidden gaps that we may not have otherwise been fully aware of, especially the qualitative approach. I think it’s helped us really, truly understand what the hidden gaps may be. So, I would like to start by saying that we have seen a lot of connectivity issues, as well as monitoring and surveillance of women’s activities online by, sometimes, family members. That’s something that, when you just do a regular survey, asking people, are you online, are you not online, all you would find out is that, no, I’m not online, but not really understanding what are the drivers behind that. You know, it could be affordability, it could be safety, and more and more, it could be, you know, the accessibility of women, and, you know, what are the other hidden factors that influence their access, as well as being able to identify some very context-specific barriers. You know, we tend to classify, for example, women as, you know, just that one monolithic group, you know, or even urban and rural, but what we are learning also is that there’s an opportunity for women to be able to have access to information, to be able to communicate with their families and those who are offline. But including the ones that are connected, there is a difference in how they are connected. being able to have a segmented approach in understanding the classifications around age, gender, income, rural, peri-urban, you know, we tend to do this rural urban, but peri-urban populations will tend to actually look closer to the experiences of rural communities because of resources and urban inequality that exists, especially when you look at a country like South Africa is a really good example where within the urban sector you can’t just take the population as is, right? And then one other one that I want to highlight is, you know, being able to tailor interventions and resources. You can’t do that without having very specific, you know, information from that particular population. National averages are just simply not serving women. They are not serving everyone else as well. And I’ll give you an example from South Africa in particular where we continue to be the most unequal society with more than half of the population living on less than half of this GNI. So anything around affordability that you are going to do in South Africa without stratifying the income quantiles, you are always going to get an over-inflated outcome that does not fully represent those who are at the bottom of the income quantiles. And women also tend to be very much over-represented in those lower levels of income. You know, so in order for us to be able to recommend gender-specific and gender-responsive policy interventions, it’s important for them to be informed by lived experiences of women. And that segmentation helps us with that. That leads me to a follow-up question if we have a bit of extra time. Have you got any good examples of where either your program has been able to do better recommendations for policymakers or where policymakers have made better decisions, more targeted, pinpointing initiatives based on better segmented data for gender, for instance, or affordability in that context? Yes, certainly. So, one of the things that we did with the Connected Resilience Report is to introduce a method that we termed policy ethnography, where we actually also brought policymakers together to understand how they make decisions and what informs them. So, one particular country, and please allow me to withhold the name of the country, had gone out to build these digital centers in rural areas because women did not have their own personal devices at home and they did not have a way to connect. So, the idea was that these digital centers that were actually funded through the Universal Service and Access Funds would enable women to be able to have access to connectivity. And over time, they realized that women were just simply not going to these centers. And it was actually through assisting them to do stakeholder consultations and to meet with women and be informed by women that we learned several things. One was that the hours that the center was available for just did not work for the women in that community who have to wake up very early, fetch wood, take care of children, get children to school, come back from the market before kids come back from school. But also, the other issue was the issue of safety, for them to then walk in a direction that is not as well populated by people that they would feel comfortable walking past, for them to be able to utilize the center. So, it had very little to do with, you know, You know, whether they have the skills or interest or any of that, it was really predicated around their own safety and their own lived experience navigating that community that had not been factored into this huge investment of building the centers with the aim and purpose of women being able to use the centers to access. So we’ve got quite a few of similar examples in our reports as well that just really shows how when you design for women and with women at the center, you have to actually design with them informing you so that they are part of that solution.

Morten Meyerhoff Nielsen: Excellent. We’ve spoken a bit about the South African context in the last example, so we’ll move to another country that has similar challenges in some ways and similar success stories. But Fabio, CETIC Brazil is increasingly segmenting their data for the Brazilian context. We’re talking about different types of segmentation already, but what are the ones that you’re finding particularly useful for decision makers in Brazil, both at local but also at the regional state or federal level? And is there a difference between the type of segments that those decision makers need in order to do better policy?

Fabio Senne: Thank you, Morten. Thank you very much for the invitation, it’s a pleasure to be with partners in this discussion. Before answering to your question, I think it’s interesting, Omnica just was very comprehensive in making the case of the advantages of having this type of information. I’d like just to describe a little bit the institutional model that we have in Brazil that I think is useful, and I think other governments are having this as a reference, because CETIC is a non-profit. So, we have a model that is based on NIC.PR, which is a non-profit organization that is funded by the .PR domain name registry. So, we have a model that the authority for the .PR, the NIC.PR, is responsible for the service, the public interest services, and this allows us to have a specific center, which is the TIC.BR, focused on producing this type of data and sharing it with policymakers. So, I think first of all, this strategy is allowing us to have the continuity of surveys and other types of research and to make the case of the relevance of this for the government and the society as a whole. And another thing that is in our DNA that I think is very relevant is that we are not only multi-stakeholder in the process of the organization, but we are also multi-stakeholder when we do research. So, every time, and I think this is a useful thing for other experiences, every time you will start an investigation and start a new survey, we invite what we call a group of experts around a multi-stakeholder group of people who will first define what to measure, what are the topics that we need to measure, what are the demand for data that we have in the government, in the private sector, in the other sectors. So, this is useful because we can adjust the data production to the demand of the decision makers, and this is very useful to us. And just to mention that this type of agreement has growing relevance among governments, we have been participating in this. So, we have been in the G20 processes for the past two years, so last year we had the Brazilian presidency of the G20, and we supported, along with ITU, a report that G20 launched, this report that was just connecting the idea of meaningful connectivity and the need for segmented monitoring. So, just to mention, in this report we argued that we cannot only use two axes, but we need to understand connection quality, availability for use, affordability, device, digital skills, and safety and security, so also in the G20 members recognizing that there are a lot of other dimensions that need to be monitored, and more importantly, that these data need to be disaggregated by demographic variables such as age, gender, household size, and others, economic variables like income, employment, status, and others, and geographic disaggregation, because you know that countries are not, are also, the digital inequalities is also expressed in the territory, and you can find differences between, so in this, in this year, in 2025, we also, along with Research ICT Africa, we supported another paper on this discussion on the G20 South African presidency, and stressing a little bit more about the funding issues, how you can, we need also to guarantee that countries have funds to do this type of research. So, we can discuss more after the…

Morten Meyerhoff Nielsen: I think it’s a very interesting… It was just to say that this institutional and consultative part is very important for… I think it’s a very interesting, CETEC Brazil is a very interesting funding model. And it’s a little bit of a sidetrack from the discussion, and we can maybe come back to it. But we see that often telco licensing and the auctioning leads to a logical profit maximization attempt from the government, particularly the Ministry of Finance. And I used to work for the Danish Ministry of Finance, so I have to put that disclaimer in. But it tends to often result in slower rollout of next generation technology. Then after the license is secured, the telcos will want to make a profit. This is logical, but they will then sacrifice either the rollout of that technology or underserve less attractive areas in remote areas, rural communities, urban communities that are not seen as profitable, or the price is transferred to the customer, which is in fact, yes, helping the government to profit maximize, but kills other government objectives and targets for digital inclusion, affordability and reliability. So there’s some interesting elements, and the Universal Access Fund is often seen as an ability to try and then reinvest the profits from the license into that, like in CETIC’s case for research, or in Tanzania, where it’s then to fix gaps in the infrastructure in remote areas or increase the volume of hotspots or free Wi-Fi hotspots, etc., with all the pros and cons they have. So there’s some interesting elements around that, but will data really help us in that regard?

Fabio Senne: That is maybe more of an open question. So a little bit of a sidetrack, but what are the type of tricks you have at CETIC in terms to nuance the data collection? Is there alternative sources, rather than just one? in the classical, we go and collect, we do surveys, etc. Are there any tricks to the trade, so to speak, from your perspective and your experience that could help increase the segmentation? Yes, I think, of course, methodological innovation can do a lot in this type of exercise. It’s not, surveys cannot take care of everything. So we are trying to mix methods, to integrate more geospatial data and other sources of data to combine with surveys. I can give a few examples. For instance, in the field of connectivity in schools, and NIC.br has a system that is called C-MAT, which is a system for a software that you install anywhere you want to test the quality of the broadband of this organization or household or so on. And in agreement with the Ministry of Education, we put this, we installed this software in more than the 70,000 schools in the country, having real time data on the quality of the connectivity. And we can cross this with the survey data that CETIC has also on the what teachers are doing in the same schools and so on. So here’s an example of you can combine different sources of information to provide more granular information. Another example that I like in terms of geographical disaggregation that I think is interesting. We did some, we tend to think that the urban areas are always well connected, but this is not the case. If you take the number of disconnected, for instance, most of them are in urban areas, very close to because the population is also concentrated in these areas. So, we had one study that we did a few years ago, that we could disaggregate, combining different sources of data. The city of São Paulo, which is the largest city in Brazil, we could disaggregate the data, combining socio-demographical and digital inequalities data. And then, for instance, we understood that one particular neighborhood in the city tends to be very high connectivity, but low level of socio-economic status. What’s happening there? So, we can refine more. And we discovered that because there was a road passing through, close to this area, across to different other sources, there was a very, there are a lot of young people living in this area, and young couples, and we could track the differences that we have

Morten Meyerhoff Nielsen: in this particular area because of the data. So, having this type of data, of course, you can lead to policies that are more attached and focused on different perspectives. Just to close the first round before we open up to the floor, also online. So, Priya, at Research ICT Africa, you’re also working with segmented data, obviously. But what are some of the examples that, again, you find particularly useful for your research, but also for policy recommendation that your center is doing on a regular basis? And, again, how does that become helpful in identifying these location or community-specific or user group-specific digital divides? Are they, again, around gender? Is it around income quartiles? Is it education? and all, what do you see from your perspective? Yeah, thank you very much, and thank you for the opportunity to be on this panel.

Pria Chetty: For us, this work is core to our organization, and so we’ve been running for a number of years our after-access research, where we’ve prioritized segmented data. So we collect data directly from individuals and households, and we ask them very specific questions. And I think our findings really reinforce what Onica mentioned about the value of qualitative analysis and what Fabia mentioned about those insights that come from when you are uniquely able to combine the data. But we ask questions around their access. Do you have a smartphone, computer, broadband at home, or affordability? And then we also ask questions about their usage patterns. So what do they use the internet for? Is it social media, work, education, health, government services? And I think this is to combat some of the assumptions that we make about who’s using what. We also ask questions about their preferred platforms. And then we ask questions about the level of digital literacy and skills that we’re dealing with. And so, you know, what are they able to do? And can they send emails, use online banking? And then some questions around trust. And I think it also reveals very, very specific nuances. And then, you know, I think when it comes to the question of barriers, this is where that contextual information really, really becomes valuable. So we ask questions about, you know, why don’t you use the internet anymore? And is it because you don’t have a need for it? Is it about safety concerns? And then, of course, in our data, we have that valuable demographic segmentation. So that’s by gender, age, income level, education, location. Anika mentioned to include the peri-urban category in there, but also disability status and language as well. So this allows us to draw out very specific insights. And the work that we’ve been doing is now absorbed globally into UN reports by the ITU, OECD, and so on. And then more regionally to define indicators that are set for specific targets. So it’s now producing, I think, for the continent, these insights that can never go away. So inputs into the ecosystem that just have to be longstanding. So at the very minimum, we can confirm that data costs are a primary barrier, but we can do that in a granular way. So in South Africa, 70% of our respondents cited affordability issues. When compared to Uganda, it was 61% that cited affordability. And we’ve got now some contrasts between the different countries. But then we also pull out these additional… traditional barriers, and one of them could be even just the lack of perceived need for some of the services that are on offer. And now we can get quite contextual about the association between what’s on offer and the demand. The digital literacy gaps at a granular level, barriers such as electricity access, privacy concerns. And it brings out, I suppose, the multidimensional element that comes from these contextual nuances that isn’t just about the segmentation, in fact, but in fact, these multidimensional qualities and these insights importantly need to inform very specific recommendations. So I suppose the segmentation and the approach and the methodology, that’s a big learning, but then how to present this information in a way in which it can be absorbed and utilized effectively. So we know that education and income, as Onica mentioned, are key drivers of digital access and use. But it means that we need targeted policies to address these. And our recommendations themselves need to be nuanced as to how this will actually take place and who would be the custodians of those kinds of efforts. So it broadens, in fact, our policy engagement audience. So while it’s not strictly, I suppose, data samples, I think there’s also that value, as Onica mentioned, in understanding the lived experiences, particularly in our context where we’re seeing this huge variation. We also need to understand attitudes across the ages and across the different segments, cultural barriers, specific use cases that draw particular people in. And then also what they consider trusted community channels. And are we exploiting that to the extent that we can? We know that cultural and linguistic barriers in schools with Internet connectivity place additional constraints. So while we might have focused on getting schools connected, are we really seeing some of those linguistic barriers and those cultural barriers that are preventing children from being able to meaningfully leverage online services? And we also know that digital exclusion… is now coming out of this data, and I suppose it’s longitudinal value, is compounded when these factors intersect. And I suppose that’s one of the challenges we take into this conversation, and I hope we return to the funding conversation because that’s an important one. But as we progress and as we get more adaptive in this space, I think we need to be able to deal with this data in a way in which we appreciate the value of the intersectional data that’s coming out and the range of inequalities that we’re seeing and how they intersect, especially for women, as Onica mentioned. But now we also know that there are intersections between drivers and barriers. So, for instance, young people in informal settlements access the internet through shared phones because they might be hungry for job

Morten Meyerhoff Nielsen: content or maybe to access bursaries, but they’re also accessing it at much higher costs. And what do we do with that information? So, I suppose, without sounding like a data geek, the data is very telling, but I suppose the challenge for us is what do we do with the data? What I hear, and correct me if I’m wrong, from all of you, including Gillian, and also reflects a little bit of the discussion we have internally at the office, is you can look at data in different levels. The national level data will be basically a heat map, the classical data. Have you used it for X, Y, Z? What’s your feature? What’s your user profile on a high level? Gives us a heat map that then allows us to say, oh, here we don’t have to worry. Everything is bright green. Here there’s something going on, but let’s wait and see. But here’s something flashing red. We need to dive into that, and based on that heat map, you know, okay, it seems to be around a socioeconomic fragment in this geographical region that has these such and others. We look at the context and see what is it really going on. But it means that we can target our decision-maker. So we layer our data and we dive in where we see the red lights flashing sort of speak, but where everything is green, we don’t have to worry. Is that correctly sort of pulled out on sort of the logic in terms of the data segmentation when it comes to the granularity, or did I misunderstand you a little bit? Any thoughts? I would say, Morten, maybe to challenge it, that it’s a

Pria Chetty: dynamic space. And so I would also exercise caution around the green. And as you mentioned, I think when you started the session, you said there’s value in the timing. And at the moment, the cycles, they’re long. And so how long does the green stay green? And what are the variations impacting the green? I mean, we’ve learned lessons coming out of the pandemic. So, you know, you’ve got, yeah, so we have to be cautious with the green. Oh, yeah, for sure. Yeah. I think at best, the segmentation just really allows us to be able to monitor, you know, what’s happening, but also to begin to think about frameworks of accountability, right? Because a lot of these divides happen within the context of a slew of policies that are there to drive inclusion, you know, whether we are doing it using the universal service and access funds, or we have broadband policies that actually are very explicit about closing the digital divides. But, you know, I think that this segmentation really helps us to be able to monitor, are we truly being effective? Are we being targeted in this? And what is the accountability framework after 20 years of this? Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. I think it’s really important to have a broadband policy that says this, and yet the results on the ground look starkly different from that. It’s not a silver bullet, you know, I think as we were talking also about just how do we fund this data collection, I was just thinking about just sort of an underutilized resource that exists in data collection right now, but it’s underutilized because it requires a lot of transformation, and that is national census data. You know, every 10 years, with no fail, most countries find money to collect national census data, but how many of us have engaged with the national census bureaus to get them to transform their line of questioning, to begin to collect even digital-related data? We’re successful with Mozambique, but then once that happens, you have another challenge, so you’ve collected all this incredible data about use of digital technologies within the country, then what? That’s another resource to invest in analyzing the data and being able to make sure that it truly is utilized to inform intervention strategies, to inform policymaking, so it’s a continuous cycle, I think it’s one of those where we have to continue to work while chewing

Morten Meyerhoff Nielsen: gum at the same time. I’m smiling because I was studying in the UK back in 1999-2000, and I was counted twice for the census back then because I happened to have been on two different addresses when they came knocking on the door, and I wasn’t actually formally registered as living in the UK at the time, so, you know, there’s also a bit of data clearance, and being Danish, we haven’t actually had a census for, I think, 50 years because we don’t need to do it. Our population registry is 99.9% proof. So, they do a sampling and they do a direct survey out with all these things on an annual basis. So, again, different context means that your data methodology and your data collection may vary. And I think that’s a key thing. But I agree that censuses can be one tool, but again, it’s a snapshot every 10 years, and the data is already outdated the next day because things change, as you’re saying. Anyway, Fabio, before we open up to and ask Carmen for any reflections online and then to the audience here, any thoughts?

Fabio Senne: No, yes, I agree with this discussion of the cycle. It’s interesting because if you take, there’s a very strong correlation between the GDP of countries and the availability of statistics on digital matters. And, of course, you can ask, is it because the statistics help countries to develop or, on the other side, because they are developed, they have money to fund the statistics? I think it’s both. And what Onica was saying that I think is very key to us is sometimes with more segmentation you can make the green area very not green. And one example is the discussion on meaningful connectivity in Brazil. Because we had, if you take our general figures, we are about to complete 90% of the population having online services or any connection to the internet. So, you can say that we are, the country is in a green line that we only have 10% disconnected. But when you go to the, when you include devices and availability of the connection and affordability and skills and other things, our figure is much worse. We only have 22% of the total population that we consider in our indicator with meaningful connectivity. So, including these hidden gaps, this also works for gender. So, if you take an overall picture, there’s not much difference, there’s not significant difference between women and men in terms of basic access to the Internet. But when we go to the indicators of meaningful connectivity, we see a 10% gap in the country in comparison of women and men in terms of meaningful connectivity. So, this closure in these gaps, I think it’s very important for this type of research. But of course, the connection with good data and good policy is not immediate. So, we understood this in CETIC, so that’s why we do also capacity building along with UNU and other partners. We do capacity building for civil servants and other strategies because we know that it’s not sufficient to have good data and the policies will get better immediately. So, you need to invest also in this connection.

Morten Meyerhoff Nielsen: I think there’s some interesting observations over the years on also statistical feedback that points in different directions. For instance, I think it was about 10 years ago, we were looking at second generation digital divides from a strategical perspective, sort of econometrics and statistical analysis. And in sub-Saharan Africa, there were more people with access to the Internet than access to reliable electricity. And electricity is a precondition for charging your device and your router. So, you know, what does that mean? You need to dive into the context and see how many people have alternative access to electricity. But it’s also quite a positive signal about how inventive people can be in order to have access to electricity. in order to assure they get access to something where they see a perceived value, which is something that we talked about earlier. Anyway, I’d like to just hand over to Carmen, who’s our online moderator. She’s also from the Global Digital Inclusion Program. And Carmen, is there any observations or questions from the online audience that you’d like to bring to our attention? I think Carmen is saying something, but it’s not coming up on the audio. Could we maybe turn up the sound for Carmen online? Hi. Can you hear me now? Yes. Okay, great. Sorry about that.

Carmen Ferri: So, we have one question in the chat, which is, how can we ensure that the segmented data collection respects the privacy and dignity of marginalized communities? Anyone from the panel that’d like to tackle that one?

Pria Chetty: And in the meantime, if there’s any questions for the audience, please raise your hand. So, maybe everyone’s looking at me because they know that I’m a lawyer. But let me start with that. I think that, I mean, there’s many facets to this, and they aren’t just the legal regulatory aspects. And so, at the very outset, I suppose, in defining the data that you’re collecting and understanding the value of the segmentation, it doesn’t necessarily include personally identifiable information. And so, in some ways, you are freed from the data protection legislation or the privacy legislation to the extent that you are not collecting personally identifiable information. So, I think that’s the first cautionary. The element that’s beyond just the legal and regulatory elements and the concerns around privacy and the perceptions around privacy… and the willingness to participate in this. I think our methodological learning has been the value of having local participation in the collection of the data. And so what you need is also buy-in for the process, and I suppose to distinguish good data from bad data, that you’ve got a willing data provider that is willing to give you the data that you need. And so by using local researchers who are able to also deal with language gaps, understand maybe the concerns of the community in participating on such a survey, you’re likely to address not just the privacy concerns

Onica Makwakwa: but the overall concerns about participation in the data collection itself. I would actually also say, just stepping away a little bit from even regulatory and legal issues, I think ethics, coming from a continent that for the most part we feel overly searched, there’s sort of this gaze on Africa in general around our way of living and all of that. It’s really important to make sure that we are working with local partners and not coming in from a global organization to study these people. We’ve got a history and a baggage that comes with that very approach, and so maybe also just assuming a decolonized approach towards collecting the data so that people consent, people understand also how we are going to utilize this data, and you bring them along so that they are part of the program as well and understand why you are collecting the data. I would like to give you an example about the number of people who are connected to the internet vis-à-vis the number of people with electricity. Because I think one of the studies that raised a huge gasp was one that compared to the number of people with mobile phones and toothbrushes. And it’s those type of narratives that, you know, when we step back and look at them, like, what exactly are you trying to argue, you know, in comparing number of mobile phones and number of toothbrushes? Because if anything, it actually exposed the ignorance of the researcher themselves, because there’s other ways of keeping dental health that’s not just only a Western toothbrush and toothpaste kind of methodology, right? So local context is not just kind of coming in and having taken me out for coffee and having a conversation with me over coffee, but it’s also really about allowing me to also within the community lead some of that collecting of research. So we’re not just kind of swooping in and, you know, having the consultants that we have, but, you know, empowering local communities also to be part of that data collection process

Fabio Senne: as well. Okay. Yeah, no, just, I think those questions are discussed a lot in the data community debates we have. UN has a UN data forum, which is a space where those types of topics are discussed. And it’s interesting because there are some trends related to what my colleagues just said. For instance, there is a concept of citizen generated data, which is now trendy in a lot of contexts where in specific contexts, you can have the citizens involved in the process of generating data with more quality, less costs, and so on. So this is a trend. So, one discussion that I think is relevant for this is that when, with the spread of mobile connections and increasing the number of internet users and digital platforms, there was a general expectation, especially among governments, that the problems of data will be solved because everybody is connected to some device and there will be traces, we know where people go, what they do online, and we have plenty of data, in a data-fied society, we have plenty of data, we don’t need monitoring or surveys anymore because the problem is solved. And now the data communities, there’s a pushback in the discussion, okay, we need complementary sources of data, sometimes we did not solve yet the thing of the digital platforms, much of the data are private and not shared for policy purposes, so there’s a discussion on how to access those types of data. Now you have satellites that provide other types of very interesting data and can be combined, so I think the data community is now in a process of, okay, we need to combine different sources to get the better solutions from cities and generate the data to more technological tools that can provide the best information you can for policymaking. I just want to double check, is there anyone in the audience that have any observations or questions that they would like to highlight? If so, please feel free. I see no one moving. Yes, please come up to the microphone. And please introduce yourself.

Morten Meyerhoff Nielsen: Thank you.

Audience: Hi, I’m Kiho Oshima and I’m a master’s student in the University of Bremen in Germany and I’m studying digital media and society. And I was wondering, you mentioned about privacy of citizens and how do you see the consent to the use of data for citizens, so especially those who are marginalized systematically, like maybe there are people who of course would not like to share their data because of the sensitivity, and also there are cases like those data are used to profile them and expect criminals or something like that, and that’s used in a way that reinforces marginalizations again. how maybe would you consider like including digital rights education and also providing like opt-in and opt-out options so that they know that not only their rights but also how to exercise their rights or would that be too big scale or to implement or something like that? And another thing is yeah so how do we prevent the use of segmented data to be interpreted or used in the way that could reinforce? So when we use data how do you prevent and

Morten Meyerhoff Nielsen: those kind of use in a way that would reinforce marginalization? Yes I would like to. Excellent questions I think particularly we can start with the privacy but I particularly like the sort of how do we use data so we don’t reinforce insisting patterns of exclusion for instance is really really interesting. Anyone want to? No I think these are very good questions

Fabio Senne: from the perspective of privacy we have a very interesting survey that we did with individuals in Brazil asking them about what they perceive about their privacy online and data protection and so on. This survey is very interesting because from one side there’s a growing concern among citizens about how the data are collected and uses. The main topic in the case of Brazil is face facial recognition for instance or when there are this type of data collection is considered the more the one that they are more concerned and also health data so health data is ultra sensitive and people feel worried about those two particular types of sensitive data. But it’s also interesting that still there is not much literacy on understanding how digital platforms collect data from people. So, there’s more concern regarding, for instance, financial frauds or something that has to do with payments. Then, when you go to a social media and put your photos, there’s not much understanding of how this model of data collection works. But when it comes to financial data and frauds, they are more concerned because it’s very objective. Very, very interesting. If we’re looking at addressing or collecting data to address the digital divide, do we need that level of that type of personal data as in what have I paid for, what is my health data? No, no. We do not, do we? Yes, I think there are two different. Traditionally, typically, in the data for policies collection, you have anonymized data even for survey or for administrative. You know, for instance, in Denmark, you know that you have so sophisticated administrative data that you even don’t need to ask people things because you have… It becomes a political risk assessment whether or not we want to use it. And there’s a discussion, there are different solutions, whether in some countries you cannot ask for ethnicity issues because there is concern of this can… So, each society will have different solutions. I can speak about the case of Brazil. There is not a distrust, a necessary distrust. When you provide information to the government, for instance, we have a very important social income program that is national, and to be in this program the government needs to know where the poor people are, so there is this trust in some cases you have to provide data. It is also the case of public health, so public health data can ensure that you are focusing on the right person. But there are some discussions on what type of data needs to be collected. Because that is a very interesting topic for discussion on access to actual opportunities through technology,

Morten Meyerhoff Nielsen: access to banking services online, access to the benefit of shopping online, government services online. And that is a very interesting element. I like that question because I think it is important for us, especially in platforms like this, to continue to raise awareness about this as a major concern for digital consumers in particular.

Onica Makwakwa: I think it is an indication of the trust deficit that exists even amongst all of the stakeholders and even an entity like IGF, for example. One example I will give is during COVID-19, most countries adopted COVID-19 tracking apps where you were encouraged to download something on your phone. I just know that for a lot of people in my country personally, we have learned that after we signed up for something related to COVID-19, we started receiving SMSs from a particular political party for the next local election campaigns that happened immediately after COVID-19. So, there really is that bridge of data privacy, right? But what I also want to encourage us as digital rights advocates is that when it comes to data, the offline and the online should not be any different. All of us probably, I don’t think it happened here in Norway, but in most countries, when we visit and you check it at a hotel, including in your own country, the receptionist takes your passport and makes a photocopy of it, and you leave and leave that photocopy of your passport behind when you check out, right? You know, let’s ask this question around data privacy and safety even in these offline instances where a security officer at a building asks for a copy of your passport or your ID number and all of that. All of those are interrelated and they’re not just unique to data, and I think we’ll make better progress in educating people around also protecting themselves in terms of their data. But as far as the data collected for research, GDIP as well uses anonymized data, so there’s really no way to be able to track who or what, but certainly from an exclusion and location point of view, certain communities could come out as vulnerable, right? And I think we need to be honest and open to exploring some of that. You know, the challenge, though, is that if you don’t count it, then it really doesn’t exist and no one has the opportunity to actually address some of these gaps. Priya, you’ve been indicating you have something to say here, so I’ll let you continue on this track. Yes. This is so close to, I suppose, the work that excites me the most, and it’s in the data

Pria Chetty: for good space, and I think your question really, for me, triggered this question about data for good or data for bad and, you know, how do we manage that? And so, you know, we said that this kind of data might fall outside of the data protection kind of regulatory space, but I want to say that even for non-personal data, there is no reason why we can’t. can’t exercise some of those standards that we would apply to personal data. And so, you know, if you’re collecting personal data, you are limited by the purpose for which you collected it, and you need to use it within those confines of why you collected it. You’ve got to uphold certain security standards. You’ve got to exercise a level of restraint and collect only what you need. And so there are valid principles that can be taken into how we do this kind of work, because I, you know, to Fabio’s point about the technology running away from us and citizen-generated data, we have to get very serious about, you know, not just personal data protection, but the protection of data sets, even the aggregated ones, because there are those harms attached to it, and there are those opportunities also linked to it. And so something that we’re working on is just trying to understand, you know, in the way that the data sets are compiled and made available and the value that sits in them, who actually gets value from it? And what about the citizen and the communities that have contributed their data? To what extent can they also then have access to it? So one of the questions we can ask ourselves is that, you know, have we meaningfully engaged with the community when we receive the results to, you know, to talk, to have a conversation with them about what this means and the decision set that they have to improve their digital inclusion, you know, characterization? And so do we take that data back to them, and do we allow them to use it? There may be budding entrepreneurs in the communities, and, you know, once they have access to this data, to know who’s connected and using what, maybe there are, you know, some level of enterprise that can emerge locally. There are many cases like this where we have to force ourselves to think about, you know, when we have the data, what do we do with it? And have we created a pathway to go back to the community and back to the citizen and make sure some of those benefits sit there? I think Annika and Fabio mentioned in some ways that the digital inclusion problem globally in the development world is big business. And so, in some ways, you know, we’ve got a willing buyer for this kind of data. But I think that doesn’t excuse us from the freedom to say,

Audience: how can we use this data responsibly, but also to get value locally from the data. It was actually a very… Do you have a follow-up? It’s really intriguing to me that to balance approaching marginalized groups, and to do that, we need to use the data. But at the same time, they need to be protected because they are marginalized at the same time. And I think I’m really intrigued to see this balance, and how it will be realized, or maybe it’s really difficult. Because even for me, looking at terms and conditions, I would skip or I would not read every sentence. But I also think that those designs can be improved. So, it’s really user-friendly. And I would just see in one site that, okay, this I would want to say yes, but this no. Also for the cookies.

Morten Meyerhoff Nielsen: It’s interesting. It comes back to that layering of data, I think. And there was an interesting example here in Norway a couple of years ago, where the journalists at the national broadcasters went and bought data from the World Wide Web, marketing data from social media, from the banks, etc. And they could basically layer it, and they identified a Ministry of Defense employer to his local commute from his home, his place of work, who he was married to, where his kids get to school, when he dropped off the kids and when the wife dropped off the kids. And it was really interesting in terms of that anonymized data, but when you have a specific objective in mind and you start layering, they could actually identify it. and many others. So, how do we identify those types of patterns from this individual? I mean, it was obviously a journalistic investigation, but it raises some concern even around anonymized data. But how do we take and build that into some of these sort of regional and global frameworks? How much data do we need also to compare ourselves? Country to country or region to region, city to city? So, how do we do that? And how do we do that in a way that we can make sure that the data we have are really key in order to ensure that we get as many people online so they can enjoy the opportunities of the World Wide Web and similar while minimizing the negative impacts of the data collection? Onica?

Onica Makwakwa: I would add to that, I mean, I think this brings us back to for my ears around open data, right? So, I think the question is, how do we leverage open data for a community? So, they may be looking at, you know, beyond access in terms of digital skills, but I might be interested in something closely related to that, maybe women specifically, but they’ve desegregated their data. They’ve collected the data. I think the other question is, how do we leverage existing data sets amongst the people who are in the digital world, and how do we do that in a way that’s, you know, we’ve

Morten Meyerhoff Nielsen: got this fragmentation of desegregation of populations and all these factors so that not everyone has to collect the same data from the same group of people as well? It’s a very interesting challenge. We had this discussion with our colleagues in Brazil when we were doing a study on youth and young adults, so the under-18-year-olds and the young-adults were defined as size 12-year-olds and other under-again It’s 15 to 18 year olds. So, you know, there was some data variation and we see the same actually in some internal research We’re doing on the digital gender divide where even if there is data collection by research teams They they segmented they do excellent work excellent data collection But they forgot to use the same age groups as for instance the National Statistical Data Agency Which means that it makes it data work and the data analysis more time-consuming or less impactful because there were some small mistakes or small missed Opportunities in alignment of data from different sources. So are there some some some national regional or even global sort of Standards or rules of thumb that you would recommend when it comes to segmentation

Fabio Senne: We’ve talked about age and income levels. We’ve talked about activities online Are there any ones that you think are almost like universal that you would would recommend that you look at? What I can comment is that I agree there are different levels of data that that you can And with etc. We try to do both. So For one side, there is a global discussion on minimal standards of data there is the the UN partnership on measuring STT for development with set meaningful rules for And and because of that we can compare data from Brazil and other countries. So we more or less follow this This this international standards and there are debates on What to measure and how to measure including a list of Disaggregations. I’ve also mentioned the G20 case where there are also recommendations for disaggregating the meaningful connectivity data. So, this is at a global level. But, what I think is important is that, different from the past, when we normally have all the data from one country concentrated in one national institute of statistics or institution. Nowadays, there is an ecosystem of data. The data is not only in public settings, but also in civil society, in the private sector. And, coordinating these efforts, I think it’s key. And, also within governance, sometimes we have silos that don’t talk to each other. So, educational data is not linked to financial data and you cannot cross this type of thing. So, more or less, we believe that this is a discussion on an ecosystem that manages data that is not only public, but public and private. It’s key for making solutions. And, of course, the data that one municipality wants for planning urban mobility is totally different from a national ministry of education planning. So, this type of granularity will come depending on the policy need you have.

Morten Meyerhoff Nielsen: Freya, Onica Makwakwa?

Pria Chetty: It’s very clear from the discussion we’re having today that, even though it’s not broad-based, the segmentation work is reaching a particular level of sophistication and maturity. But, as Onica Makwakwa mentioned, we’re not comparing notes. And so, the opportunity, if there was any kind of regional collaboration or there was an initiative at a global level, is to bring the key players together to understand the different methodologies. And, UNESCO has set up some tools and indicator sets that can be leveraged. But, what is it that has come from this qualitative work and from these unique combinations that we mentioned and these intersections that also add to those frameworks that exist? And, how can we develop it so that the data is more reliable and more accessible? I think what we’re being challenged to do as an organization now is make the data more accessible to people who have questions that we never conceived of. And, I think, to also anticipate that they aren’t researchers and they aren’t policy experts, but they will have data. very unique requirements from the data set and how can you create something that is accessible and allows them to use that data in ways that you haven’t conceived of. So imagining that it’s going beyond this community and it’s being used in new ways because that’s what we’re seeing now that we can’t really identify how the data will be used but we want to make it accessible for uses that we didn’t conceive of. I want to exercise, yeah, I suppose some caution I would say in bringing the data together. I think it needs to be very deliberate and very well intentioned because there’s the potential and propensity there for massive harm if the data is, you know, brought together in ways that we lose that element of control or accountability for why we initially brought it and put the data together. And there’s so many examples where, you know, data lakes have gone badly and introduces just a range of vulnerabilities. So it isn’t, I’m not

Morten Meyerhoff Nielsen: sharing an excitement to bring all the data together without thinking through very carefully what is the process that we use and how do we get a little more adaptive, I suppose, in the way that we… We hear some very interesting reflections also from policymakers and civil servants including in regions like East Africa where you have these massive drives for data lakes but without having a data classification scheme that says this is data that you can easily rely on because it’s high quality. This is not as high quality because that defines how will I interpret the results.

Onica Makwakwa: Just one of the exercises we did in Ghana several years ago and that was to create a data and research working group where they mainly focused on kind of mapping where existing ICT data was within the country, so that as researchers, they kind of talk to each other, they know who has what data and what frequency, you know, how frequent is that data updated, including some level of success at bringing some of the operators, the mobile operators, in place. They are obviously also sitting with tons of data that we may or may not find useful, right? But it wasn’t about bringing all the data in one place, but just really having a sense of a mapping of, you know, who has what data and, you know, being able to also negotiate some openness for researchers who may be interested in doing research and looking at particular issues to be able to know, you know, where they can rely on. It also reduces the size of household surveys that you have to do when you realize that you don’t need 40 questions because this particular

Morten Meyerhoff Nielsen: operator, you know, the three operators within the country can tick off maybe 10 of those. I’m loving that because it was actually the topic of some recommendations CETIC and UNUEGOV did in a G20 policy brief last year under the Brazilian presidency, as in, in 155 countries you must present legal identity in order to get a mobile phone or internet connection. Is there ways and models that we can use that and the telco licensing for the telco providers to provide us a snapshot on gender, on age groups, on basic elements that are anonymized to the telco regulator so they get that snapshot but then also provide certain usage data like on certain types of IP addresses like online commerce, banking, again, so you get that data in snapshots but anonymized. to create sort of the initial heat maps, lower the burden of data collection that then allows you to also dive into the context of the green flashing lights but it gives, it’s a new model and and with the telco regulators you can potentially create those type of partnerships by making it part of the operational responsibility to provide certain types of data so decision makers can address the digital divide and spend more time on on diving into the specific challenges in the context rather than these blanket decisions. So they are some potential models at play there. I would just checking the audience to see if there’s another set of questions before I raise the last question to the audience. I’ve seen online, there’s no more questions online but if you were gonna do one recommendations that you would ideally like to see happen in the segmentation of data collection to address the digital divide, to make those better decisions, what would that one recommendation be for the next 12 months?

Fabio Senne: Well I can start, this is a very difficult question but I think there’s one thing that we kind of discussed but I would like to reinforce is that for a long time that’s because we do mostly surveys or they interview one individual so we try to think about digital inclusion as an individual characteristic. So you take gender and age and all these factors and you think we are talking about individuals but most of the problems are collectives, collective problems. So there’s lots of discussion on community networks, how how communities can build innovative models for digital inclusion. Schools are also important in this debate. In Brazil, we have libraries as one of the most spread public infrastructure that also need to be engaged. So, I do think that, try to think more in collective measures or collective exercise to understand better the situation. One small example, in this meaningful connectivity study, apart from having one indicator, does the household have a computer, which is a traditional indicator that everybody measures. We decided to calculate, okay, but if the household has one person and one computer, you are okay, but if you have one computer and ten people living there fighting from the same device, what will be the quality of the use? So, we decided to calculate a ratio of people per device, or you can do this with income, what’s the percentage of the income of the device. So, this type of thing thinks the household as a collective of people rather than individual. So, I think facing the collective challenges is useful for policy in digital inclusion.

Morten Meyerhoff Nielsen: I won’t answer the question, but I’ll just pick up on what you just said around looking at household v. individual.

Onica Makwakwa: And that’s because we actually did an evaluation of a model, a subsidy model of one tablet per household in Uganda, which was very successful. They focused mainly on female-led households, used USF funds to actually… provide a tablet per household. And very interesting stories, please do look it up on our website, on how that actually helped to empower, you know, the least suspected subject for that intervention. And, you know, even though the focus was mainly with the women in the households, it’s the children who actually benefited the most, and it’s education that was highly impacted by that initiative. So, I really like that notion that, you know, it’s really important for us to also just sort of look at communities the way they are organized.

Morten Meyerhoff Nielsen: I mean, there was some interesting discussions between the ITU and the Arab regions technical working group for statistics, where actually the recommendations from the Arab group on the definition of a household got changed by ITU, because it wasn’t, it was based on a traditional nuclear family, you know, global north concept, and not on multi-generational or alternative household structures that you see all over the world, even in the global north. So, it was really, really interesting in that regard. But I’m taking away the time from Priya to either address the same question, or come up with an actionable suggestion for the

Pria Chetty: next 12 months. Yeah, maybe I suppose, maybe put differently, what is the question that keeps me up at night for this kind of work? And I would say it’s about the sustainability of the work, and the adaptation maybe of methodologies, and how we do it. And, I mean, it takes a long time, and it’s quite costly, and to do it well, you’ve got to really have the local participation, and so on. So, how do we build on the current methodologies, and how do we make it more sustainable? And how do we make sure that there’s continued interest in the process to get this data, because there’s high levels of interest in the data, but not always in the process to collect the data. And linked to that sustainability question for me is, then, what is the compelling way? in which we approach the sharing and exchange of data. And as you spoke about the opportunity with the telecoms firms, I would say to you the biggest challenge, because they would probably be doing that analysis already and have all that kind of segmentation there. They have different reasons for it. But to get them to share it as data for good, how do we present a compelling case to them? How do we build those incentive structures? And if we don’t figure that out, I feel that this work will not be sustainable, because it will be replaced by quicker technical measures that don’t necessarily have the rigour attached to it. So yes, we’ve got to balance innovation with some of the traditional methodologies. We’ve got to adapt it. But how are we going to do that to make sure we don’t ever lose these insights?

Morten Meyerhoff Nielsen: Thank you. Just to try and summarise a little bit on the key takeaways that I’ve seen is that there’s a general consensus that we need better segmented data around gender, income levels, location, education and so forth. But there’s also a recognition that will only get us so far. To address the digital divide, we need to then go down and analyse the contextual, which becomes, to paraphrase Onica Makwakwa, is more of a qualitative assessment. It is not a statistical empirical assessment. And there’s also, I think, more or less a consensus on the panel and with the audience that, yes, we need to balance this sort of anonymisation with privacy when we start layering data across different elements. And then lastly, that there are some opportunities for different types of partnerships for data collection, both with the private sector, and many others. We have been working with a number of partners, particularly telcos, but also with local communities and getting them involved in driving the decisions in order to also target the more tailored initiatives that will include them and give them the opportunities for being digitally included and benefit from that. And we’ll also focus on that. We’ve also now managed to offer an effect things in multiple styles with combination divides in there that doesn’t just go by genetic or income level but are cross-cutting – I forgot what you called it, Pria. Inter odpowiagonal, both on the divide and the inclusion. And we’ve now managed to offer a combination of both of those styles. We have a lot of information on there and details. We will be summarising the discussion of this panel and provide it online in the next couple of weeks. Obviously, the panellists will have a chance to also comment on that draft. But we’ll be sharing that. In the meantime, don’t hesitate to reach out to any of us or to our organisations if you have further questions. We hope that still, in the near future, more discussion of airline logistics can take place. Thank you very much, and enjoy the next couple of days of IGF. Thank you. .

Agreement points

Need for segmented data collection beyond basic connectivity metrics

Speakers

– Guilherme Canela De Souza Godoi
– Fabio Senne
– Onica Makwakwa
– Pria Chetty
– Morten Meyerhoff Nielsen

Arguments

UNESCO provides comprehensive indicator sets like Internet Universality Indicators based on five pillars (rights, openness, accessibility, multi-stakeholderism) that member states can use for evidence-based policies

G20 has recognized the need for segmented monitoring beyond basic connectivity, requiring disaggregation by demographic, economic, and geographic variables

Meaningful connectivity indicators show much lower inclusion rates than basic access statistics – Brazil has 90% basic internet access but only 22% meaningful connectivity

Segmented data collection reveals gender-specific barriers like affordability, safety concerns, and monitoring by family members that regular surveys miss

Research ICT Africa’s after-access research combines quantitative and qualitative methods to understand usage patterns, digital literacy levels, and trust issues

Current frameworks are supply-oriented, focusing on basic usage rather than meaningful connectivity and demand-side activities

Summary

All speakers agree that traditional binary connectivity measures (connected/not connected) are insufficient and that comprehensive segmented data collection is essential to understand the true nature of digital inclusion and exclusion

Topics

Development | Digital access

Importance of qualitative approaches and local context in data collection

Speakers

– Onica Makwakwa
– Pria Chetty
– Fabio Senne

Arguments

Qualitative approaches and policy ethnography help surface hidden gaps and context-specific barriers that quantitative data alone cannot capture

Digital centers built for rural women failed because they didn’t account for women’s daily schedules and safety concerns when walking to centers

Local participation in data collection addresses privacy concerns and ensures willing participation from communities

Geographic disaggregation within cities reveals unexpected patterns, such as high connectivity but low socioeconomic status in specific neighborhoods

Summary

Speakers consistently emphasize that quantitative data alone is insufficient and that qualitative methods with strong local participation are essential to understand the real barriers and contexts affecting digital inclusion

Topics

Development | Human rights principles

Need for multi-stakeholder and collaborative approaches to data collection

Speakers

– Guilherme Canela De Souza Godoi
– Fabio Senne
– Onica Makwakwa
– Pria Chetty

Arguments

Brazil’s CETIC model uses multi-stakeholder approach with expert groups to define what to measure, adjusting data production to decision-maker demands

Modern data ecosystems span public, private, and civil society sectors, requiring coordination rather than relying solely on national statistics institutes

Mapping existing data sources within countries reduces survey burden and helps researchers know what data is available from different stakeholders

Data collection sustainability requires building incentive structures for private sector participation and making compelling cases for data sharing

Summary

All speakers agree that effective data collection requires collaboration across sectors and stakeholders, moving beyond traditional single-institution approaches to leverage diverse data sources and expertise

Topics

Development | Legal and regulatory

Recognition of collective rather than individual nature of digital inclusion challenges

Speakers

– Fabio Senne
– Onica Makwakwa

Arguments

Digital inclusion should be viewed as collective rather than individual challenges, considering community networks, schools, and libraries as infrastructure

Household-level analysis reveals important dynamics like device-sharing ratios and income allocation that individual-focused measures miss

One tablet per household programs in Uganda successfully empowered female-led households, with unexpected benefits for children’s education

Communities should be involved in leading data collection processes rather than being passive subjects of external research

Summary

Speakers agree that digital inclusion is fundamentally a collective challenge requiring community-level solutions and household-level analysis rather than focusing solely on individual characteristics

Topics

Development | Infrastructure

Similar viewpoints

Both speakers emphasize the critical importance of gender-disaggregated data and the inadequacy of national averages for understanding women’s experiences with digital technologies, particularly in highly unequal societies

Speakers

– Onica Makwakwa
– Pria Chetty

Arguments

Segmented data collection reveals gender-specific barriers like affordability, safety concerns, and monitoring by family members that regular surveys miss

National averages fail to serve women and marginalized groups, particularly in highly unequal societies like South Africa where income stratification is essential

Research ICT Africa’s after-access research combines quantitative and qualitative methods to understand usage patterns, digital literacy levels, and trust issues

Topics

Gender rights online | Development

Both speakers advocate for methodological innovation that combines traditional surveys with new data sources and technologies, while recognizing the sustainability challenges of comprehensive data collection

Speakers

– Fabio Senne
– Pria Chetty

Arguments

Methodological innovation should combine surveys with geospatial data and other sources, such as real-time connectivity quality monitoring in schools

Citizen-generated data and satellite data provide complementary sources that can be combined with traditional surveys

Data collection sustainability requires building incentive structures for private sector participation and making compelling cases for data sharing

Topics

Development | Infrastructure

Both speakers emphasize the ethical dimensions of data collection, advocating for decolonized approaches that prioritize local participation and community leadership in research processes

Speakers

– Onica Makwakwa
– Pria Chetty

Arguments

Decolonized approaches to data collection are essential, working with local partners rather than external researchers studying communities

Local participation in data collection addresses privacy concerns and ensures willing participation from communities

Anonymized data collection for policy purposes typically falls outside personal data protection regulations but should still follow ethical standards

Topics

Human rights principles | Privacy and data protection

Unexpected consensus

Private sector data partnerships as essential for sustainability

Speakers

– Fabio Senne
– Pria Chetty
– Morten Meyerhoff Nielsen
– Onica Makwakwa

Arguments

Citizen-generated data and satellite data provide complementary sources that can be combined with traditional surveys

Data collection sustainability requires building incentive structures for private sector participation and making compelling cases for data sharing

Telecoms operators hold valuable segmented data that could support policy decisions through partnerships with regulators

Mapping existing data sources within countries reduces survey burden and helps researchers know what data is available from different stakeholders

Explanation

Despite typical concerns about private sector data control and privacy, there was unexpected consensus that partnerships with private sector entities, particularly telecoms, are essential for sustainable and comprehensive data collection. This represents a pragmatic recognition that traditional survey methods alone are insufficient and costly.

Topics

Development | Economic

Limitations of urban-rural binary classifications

Speakers

– Onica Makwakwa
– Fabio Senne

Arguments

Peri-urban populations often experience challenges similar to rural communities due to urban inequality, requiring more nuanced geographic classification beyond rural-urban

Most disconnected people are actually in urban areas due to population concentration, challenging assumptions about urban connectivity

Geographic disaggregation within cities reveals unexpected patterns, such as high connectivity but low socioeconomic status in specific neighborhoods

Explanation

There was unexpected consensus that traditional rural-urban classifications are inadequate for understanding digital divides. This challenges conventional wisdom about where digital exclusion occurs and suggests more nuanced geographic analysis is needed.

Topics

Development | Infrastructure

Overall assessment

Summary

The speakers demonstrated remarkably high consensus on fundamental issues around digital inclusion data collection, including the need for segmented data beyond basic connectivity, the importance of qualitative and local approaches, multi-stakeholder collaboration, and viewing digital inclusion as collective challenges. There was also unexpected agreement on pragmatic issues like private sector partnerships and the limitations of traditional geographic classifications.

Consensus level

Very high consensus with strong alignment on both methodological approaches and policy implications. This suggests a mature field where practitioners have converged on best practices through experience, creating a solid foundation for advancing digital inclusion measurement and policy interventions globally.

Different viewpoints

Data aggregation and sharing approaches

Speakers

– Pria Chetty
– Onica Makwakwa

Arguments

Data collection sustainability requires building incentive structures for private sector participation and making compelling cases for data sharing

Mapping existing data sources within countries reduces survey burden and helps researchers know what data is available from different stakeholders

Summary

Pria Chetty expressed caution about bringing data together due to potential for massive harm and vulnerabilities in data lakes, while Onica Makwakwa advocated for mapping and coordinating existing data sources. Pria emphasized the need for deliberate and well-intentioned processes with accountability, whereas Onica focused on practical coordination benefits.

Topics

Development | Privacy and data protection

Unexpected differences

Risk tolerance for data integration

Speakers

– Pria Chetty
– Onica Makwakwa

Arguments

Data collection sustainability requires building incentive structures for private sector participation and making compelling cases for data sharing

Mapping existing data sources within countries reduces survey burden and helps researchers know what data is available from different stakeholders

Explanation

This disagreement was unexpected because both speakers are from organizations focused on digital inclusion research and might be expected to have similar approaches to data coordination. Pria’s strong caution about data lakes and integration risks contrasted with Onica’s more optimistic view of data mapping and coordination benefits.

Topics

Development | Privacy and data protection

Overall assessment

Summary

The discussion showed remarkable consensus among speakers on the need for better segmented data collection, the importance of qualitative approaches, and the value of local community involvement. The main area of disagreement was around data integration approaches, with differing views on risk tolerance and coordination strategies.

Disagreement level

Very low level of disagreement. The speakers largely complemented each other’s perspectives rather than conflicting. This high level of consensus suggests a mature field where practitioners have converged on core principles, but may indicate a need for more diverse perspectives to challenge existing approaches and drive innovation in digital inclusion data collection methodologies.

Partial agreements

Similar viewpoints

Both speakers emphasize the critical importance of gender-disaggregated data and the inadequacy of national averages for understanding women’s experiences with digital technologies, particularly in highly unequal societies

Speakers

– Onica Makwakwa
– Pria Chetty

Arguments

Segmented data collection reveals gender-specific barriers like affordability, safety concerns, and monitoring by family members that regular surveys miss

National averages fail to serve women and marginalized groups, particularly in highly unequal societies like South Africa where income stratification is essential

Research ICT Africa’s after-access research combines quantitative and qualitative methods to understand usage patterns, digital literacy levels, and trust issues

Topics

Gender rights online | Development

Both speakers advocate for methodological innovation that combines traditional surveys with new data sources and technologies, while recognizing the sustainability challenges of comprehensive data collection

Speakers

– Fabio Senne
– Pria Chetty

Arguments

Methodological innovation should combine surveys with geospatial data and other sources, such as real-time connectivity quality monitoring in schools

Citizen-generated data and satellite data provide complementary sources that can be combined with traditional surveys

Data collection sustainability requires building incentive structures for private sector participation and making compelling cases for data sharing

Topics

Development | Infrastructure

Both speakers emphasize the ethical dimensions of data collection, advocating for decolonized approaches that prioritize local participation and community leadership in research processes

Speakers

– Onica Makwakwa
– Pria Chetty

Arguments

Decolonized approaches to data collection are essential, working with local partners rather than external researchers studying communities

Local participation in data collection addresses privacy concerns and ensures willing participation from communities

Anonymized data collection for policy purposes typically falls outside personal data protection regulations but should still follow ethical standards

Topics

Human rights principles | Privacy and data protection

Key takeaways

Better segmented data collection is essential for addressing digital divides, requiring disaggregation by gender, income, location, education, age, and other demographic variables rather than relying on national averages

Qualitative research and policy ethnography are crucial for understanding hidden barriers and context-specific challenges that quantitative surveys alone cannot capture

Meaningful connectivity indicators reveal much lower digital inclusion rates than basic access statistics – highlighting the gap between having internet access and being able to use it effectively

Current digital inclusion frameworks are predominantly supply-oriented, focusing on basic usage rather than demand-side activities and quality of digital engagement

Local community participation in data collection is essential for addressing privacy concerns, ensuring cultural sensitivity, and obtaining reliable data from marginalized populations

Digital inclusion should be viewed as a collective challenge requiring household and community-level analysis rather than purely individual metrics

Data privacy and ethical considerations must be balanced with the need for segmented data, requiring anonymization, consent mechanisms, and decolonized research approaches

Alternative data sources including telecoms data, census information, geospatial data, and citizen-generated data can complement traditional surveys and reduce collection burden

Resolutions and action items

Panel discussion results will be summarized and shared online within the next couple of weeks, with panelists having opportunity to comment on the draft

Participants encouraged to reach out to panelists and their organizations for further questions and collaboration

Implicit commitment to continue developing segmented data collection methodologies and sharing best practices among organizations represented

Unresolved issues

How to ensure long-term sustainability and funding for comprehensive segmented data collection efforts, particularly in developing countries

How to create compelling incentive structures for private sector (especially telecoms) to share data for public good while maintaining commercial interests

How to prevent segmented data from being used to reinforce existing patterns of marginalization and exclusion

How to balance the need for detailed segmentation with privacy protection, particularly when layering multiple data sources

How to standardize segmentation categories globally while maintaining relevance to local contexts and needs

How to make segmented data more accessible to non-researchers and enable uses that weren’t originally conceived by data collectors

How to coordinate data collection across the ecosystem of public, private, and civil society actors to avoid duplication and gaps

How to adapt methodologies to be more sustainable and cost-effective while maintaining rigor and local participation

Suggested compromises

Using layered data approach where national-level data provides ‘heat maps’ to identify problem areas, then diving deeper with contextual analysis only where red flags appear

Combining multiple data sources (surveys, administrative data, telecoms data, satellite data) rather than relying on single collection methods

Applying data protection standards to non-personal data as a precautionary measure, even when not legally required

Engaging with national census bureaus to include digital-related questions rather than conducting separate comprehensive surveys

Creating data mapping exercises within countries to coordinate existing data sources rather than centralizing all data in one location

Focusing on anonymized, aggregated data for policy purposes while maintaining individual privacy protections

Balancing innovation in data collection methods with traditional rigorous methodologies to ensure reliability

Using household-level indicators alongside individual metrics to capture collective aspects of digital inclusion

But when you go to the, when you include devices and availability of the connection and affordability and skills and other things, our figure is much worse. We only have 22% of the total population that we consider in our indicator with meaningful connectivity… if you take an overall picture, there’s not much difference, there’s not significant difference between women and men in terms of basic access to the Internet. But when we go to the indicators of meaningful connectivity, we see a 10% gap in the country in comparison of women and men in terms of meaningful connectivity.

Speaker

Fabio Senne

Reason

This comment fundamentally challenges the conventional understanding of digital inclusion by revealing the stark difference between basic connectivity statistics and meaningful connectivity. It demonstrates how surface-level data can be misleading and mask significant inequalities.

Impact

This shifted the discussion from focusing on simple access metrics to understanding the complexity of digital inclusion. It reinforced the panel’s central argument about the need for segmented data and influenced subsequent discussions about the limitations of national averages and the importance of looking beyond basic connectivity measures.

National averages are just simply not serving women. They are not serving everyone else as well… anything around affordability that you are going to do in South Africa without stratifying the income quantiles, you are always going to get an over-inflated outcome that does not fully represent those who are at the bottom of the income quantiles.

Speaker

Onica Makwakwa

Reason

This comment powerfully articulates why aggregate data fails marginalized communities and provides a concrete example of how statistical methodology can perpetuate inequality by masking the experiences of the most vulnerable populations.

Impact

This comment became a cornerstone argument for the entire discussion, establishing the fundamental problem with current data collection approaches. It led to deeper exploration of intersectionality and influenced the conversation about policy ethnography and community-centered research approaches.

I think for a long time that’s because we do mostly surveys or they interview one individual so we try to think about digital inclusion as an individual characteristic… but most of the problems are collectives, collective problems… We decided to calculate a ratio of people per device, or you can do this with income, what’s the percentage of the income of the device. So, this type of thing thinks the household as a collective of people rather than individual.

Speaker

Fabio Senne

Reason

This comment introduces a paradigm shift from individual-focused to collective-focused analysis of digital inclusion, challenging the fundamental assumptions underlying most digital divide research and policy interventions.

Impact

This reframed the entire approach to data collection and analysis, leading to discussions about household dynamics, community networks, and collective solutions. It influenced Onica’s follow-up about successful household-based interventions and changed how participants thought about designing both research and policy interventions.

It’s those type of narratives that, you know, when we step back and look at them, like, what exactly are you trying to argue, you know, in comparing number of mobile phones and number of toothbrushes? Because if anything, it actually exposed the ignorance of the researcher themselves, because there’s other ways of keeping dental health that’s not just only a Western toothbrush and toothpaste kind of methodology.

Speaker

Onica Makwakwa

Reason

This comment brilliantly exposes the colonial and culturally biased assumptions embedded in development research, using a specific example to illustrate how researchers’ cultural blind spots can lead to problematic narratives about communities they study.

Impact

This comment elevated the discussion to address power dynamics, colonial legacies, and cultural sensitivity in research. It reinforced the importance of local participation and decolonized approaches, influencing the subsequent conversation about ethics, community engagement, and the need for local researchers to lead data collection processes.

How can we ensure that the segmented data collection respects the privacy and dignity of marginalized communities?

Speaker

Carmen Ferri (online moderator)

Reason

This question introduced a critical tension at the heart of the discussion – the need to collect detailed data about vulnerable populations while simultaneously protecting them from potential harm through that same data collection.

Impact

This question fundamentally shifted the discussion from technical and methodological considerations to ethical ones, sparking a rich conversation about privacy, consent, power dynamics, and the potential for data to reinforce marginalization. It led to discussions about decolonized research approaches and the balance between data utility and community protection.

There’s high levels of interest in the data, but not always in the process to collect the data… How do we build those incentive structures? And if we don’t figure that out, I feel that this work will not be sustainable, because it will be replaced by quicker technical measures that don’t necessarily have the rigour attached to it.

Speaker

Pria Chetty

Reason

This comment identifies a fundamental sustainability challenge in the field – the disconnect between demand for insights and willingness to invest in rigorous data collection processes, highlighting the risk of losing methodological rigor for convenience.

Impact

This comment brought the discussion full circle to practical implementation challenges and long-term sustainability. It influenced the final recommendations and highlighted the need for innovative partnerships and funding models, connecting back to earlier discussions about telco partnerships and alternative data sources.

Overall assessment

These key comments fundamentally shaped the discussion by progressively deepening and complicating the conversation about digital inclusion data. The discussion evolved from technical considerations about data segmentation to profound questions about methodology, ethics, power dynamics, and sustainability. Fabio’s insights about meaningful connectivity versus basic access established the empirical foundation for why current approaches are inadequate. Onica’s comments about national averages and cultural bias elevated the conversation to address systemic inequalities and colonial legacies in research. The privacy question introduced crucial ethical dimensions, while Pria’s sustainability concerns brought practical implementation challenges into focus. Together, these comments transformed what could have been a technical discussion about data collection into a nuanced exploration of how research methodology intersects with social justice, cultural sensitivity, and long-term impact. The comments built upon each other to create a comprehensive critique of current approaches while pointing toward more equitable and sustainable alternatives.

How can we ensure that the segmented data collection respects the privacy and dignity of marginalized communities?

Speaker

Online audience member (via Carmen Ferri)

Explanation

This addresses the critical balance between collecting necessary data to address digital divides while protecting vulnerable populations from potential harm or exploitation through data collection processes.

How can we prevent the use of segmented data to be interpreted or used in ways that could reinforce marginalization?

Speaker

Kiho Oshima (University of Bremen student)

Explanation

This highlights the risk that data intended to help marginalized communities could inadvertently be used to profile or further discriminate against them, requiring careful consideration of data governance and usage protocols.

How do we balance approaching marginalized groups with data collection needs while protecting them because they are marginalized?

Speaker

Kiho Oshima (University of Bremen student)

Explanation

This explores the fundamental tension between needing data about vulnerable populations to help them while simultaneously protecting them from potential harms of data collection and usage.

How can we build compelling incentive structures for private sector data sharing, particularly with telecommunications firms?

Speaker

Pria Chetty

Explanation

This addresses the sustainability challenge of data collection by exploring how to motivate private companies to share valuable data for public good purposes, which could significantly improve data availability and reduce collection costs.

How can we make segmented data collection more sustainable and adapt methodologies for long-term viability?

Speaker

Pria Chetty

Explanation

This focuses on the practical challenge of maintaining high-quality, locally-participatory data collection processes over time, given their high costs and resource requirements.

How can we leverage open data for communities and reduce fragmentation of data collection efforts?

Speaker

Onica Makwakwa

Explanation

This explores opportunities to maximize the utility of existing datasets and avoid duplicative data collection efforts while ensuring communities can benefit from and access relevant data about themselves.

How can we engage with national census bureaus to transform their data collection to include digital-related questions?

Speaker

Onica Makwakwa

Explanation

This identifies an underutilized opportunity to incorporate digital inclusion metrics into existing, well-funded national data collection infrastructure that occurs regularly across countries.

How can we develop regional collaboration frameworks to compare methodologies and improve data reliability and accessibility?

Speaker

Pria Chetty

Explanation

This addresses the need for coordination among organizations doing similar work to share best practices, standardize approaches where appropriate, and make data more accessible to diverse users with varying technical expertise.

How can we create data mapping exercises to identify existing ICT data sources within countries?

Speaker

Onica Makwakwa

Explanation

This suggests a systematic approach to cataloging available data sources, including from mobile operators and other private entities, to reduce survey burden and improve research efficiency.

How can we develop collective measures for digital inclusion rather than focusing solely on individual characteristics?

Speaker

Fabio Senne

Explanation

This challenges the traditional approach of measuring digital inclusion at the individual level and suggests exploring household, community, and collective indicators that better reflect how digital access and use actually occurs in practice.