Digital Inclusion Is More Than Access and Usability

A person can reach a digital service, navigate it correctly and submit a transaction yet still be excluded from what the service is meant to provide. Meaningful digital inclusion also requires understanding, real opportunity, agency, workable recourse and outcomes that do not systematically disadvantage people because of their circumstances.

When task success hides service failure

Imagine a parent applying online for a public benefit. She has a smartphone and enough data. The form works with her screen reader. Its questions are clear, and she submits it without an error. A message then says the application was unsuccessful. It gives no intelligible reason, no way to correct the record and no reachable human channel. The deadline passes.

This is an illustration, not a reported case. Yet it exposes a real measurement problem. Every conventional interaction metric may look positive: the service was reached, the form was usable and the transaction was completed. The interface worked. The service did not work for the person.

Digital inclusion is often assessed close to the interface: who is connected, who owns a device, who logs in and who completes a task. Those measures matter. They identify barriers that can exclude people before an interaction begins. But they cannot show, on their own, whether a digitally mediated institution has enabled a person to understand, act, retain control, recover from error or obtain the service that matters to them.

The distinction is consequential. Successful interaction is not the same as successful inclusion. Task completion is evidence relevant to usability. It is not, by itself, evidence that a person has been included on fair and meaningful terms.

Access was an important starting point, not the destination

The early emphasis on the “digital divide” was justified. Without affordable connectivity, a suitable device, accessible content, usable authentication and a viable language or support channel, further questions can be academic. Access remains a necessary condition of digital inclusion.

But research has progressively widened the unit of analysis. A systematic review by Scheerder, van Deursen and van Dijk distinguished inequalities in skills and patterns of use from inequalities in the benefits people obtain online. Van Deursen and Helsper’s study of internet outcomes in the Netherlands found that people with greater social advantage often gained more from similar kinds of internet use. The implication is not that access has been solved. It is that equal access can coexist with unequal returns.

The 2026 review by Anuradha Ganapathy, Richard Heeks and Gianluca Iazzolino brings this shift into ICT4D. Reviewing 77 studies, the authors find that scholarship has moved beyond access-only, technologically deterministic and universalist accounts towards more socially embedded and human-centred approaches. They also identify a “paradigm lag”: research has concentrated more on enabling benefits than on forms of disadvantage produced through digital systems, including surveillance, data extractivism and enforced participation.

That finding supports a wider view of inclusion, but it does not supply the seven-part heuristic proposed below. Nor does it mean that every digital intervention is extractive or coercive. It does mean that being connected to a system cannot automatically be counted as a benefit. As Heeks’s concept of adverse digital incorporation emphasizes, inclusion can occur on terms that allow more powerful actors to capture disproportionate value. A person may be inside the system and still be disadvantaged by it.

Usability solved another layer of the problem

Usability and accessibility address a second indispensable question: can people operate the service effectively and with reasonable effort? If a page cannot be navigated with a keyboard, a form rejects legitimate names, instructions depend on colour, or an authentication journey overwhelms a first-time user, exclusion is already occurring. Good interaction design reduces error, effort and uncertainty. It can be the difference between reaching a service and abandoning it.

Still, an operable interface does not guarantee an informed or effective encounter. A person may correctly press “agree” without understanding how their data will be used. They may complete an eligibility form without understanding which evidence counts, or follow every instruction while an inaccurate administrative record determines the result.

Bangladesh offers a careful example of why these distinctions matter. A 2019 nationwide survey by BRAC Institute of Governance and Development covered 6,500 rural households. Ninety-six per cent reported using a mobile phone, but 59 per cent lacked a smartphone and 37 per cent had internet connectivity. In a practical test, only 13 per cent could complete at least one of three basic information-finding tasks on the Bangla passport website. These are dated, rural household findings, not a description of Bangladesh today. Their value is conceptual: a high level of basic device use did not translate automatically into the skills needed to use a public website.

A 2026 UNDP and a2i study of 354 people with disabilities across 14 Bangladeshi districts found layered barriers: limited access to assistive devices, low self-rated digital skills, inaccessible government websites, and little awareness of accessibility policies or complaint channels. The sample should not be treated as nationally representative. It nevertheless shows why technical conformance, individual capability and institutional redress need to be examined together.

What happens after the user can use the interface?

The capability approach helps explain the next step. In this tradition, development is assessed not only by resources or functions supplied but by people’s substantive opportunities to achieve ways of living they have reason to value. Zheng and Walsham applied this to digital inequality by framing exclusion as capability deprivation. Kleine’s Choice Framework similarly asks how ICT interacts with personal, social and material resources to expand or constrain people’s options.

Interface functionality is therefore not capability. A downloadable certificate is a function. The realistic opportunity to obtain and use it may depend on literacy, time, documents, money, safety, social permission, institutional recognition and human help. These “conversion factors” explain why the same design can create very different opportunities for different people.

Capability also needs to be distinguished from agency. Capability concerns the real set of opportunities available. Agency concerns a person’s ability to form and pursue goals and to influence what happens, individually or with others. A service may expand capability while constraining agency: it might make a payment faster, for example, while making data collection unavoidable or eliminating a workable offline route. Conversely, a nominal choice does not create agency if refusing it means losing an essential service.

Agency is not always individual. Research by Sabira Anwar and Larry Stillman on rural women’s use of smartphones in Bangladesh describes capability development through relationships and collective action. That matters for inclusive design: assistance from family members, community intermediaries and frontline workers can be an enabling social resource. It can also create privacy, dependency or gatekeeping risks. The relevant question is not whether assistance exists, but whether it expands the person’s effective control on acceptable terms.

A broader way to examine digital inclusion

I propose the following heuristic for discussion and further testing:

Access → Usability → Comprehension → Capability → Agency → Recourse → Outcome

This is an applied synthesis, not a validated scale, a universal sequence or a framework derived from Ganapathy, Heeks and Iazzolino. The arrows indicate a useful route for diagnosis: from reaching an interface to experiencing its consequences. In practice, the dimensions overlap, people move back and forth between them, and power and institutional conditions shape every one.

Access: Can the intended person realistically reach the service? Connectivity and devices matter, as do cost, authentication, language, disability access, literacy, privacy, safety and non-digital alternatives. Registration figures should not be mistaken for effective reach.

Usability: Can the person operate it accurately, efficiently and with tolerable effort? Completion rates should be disaggregated alongside abandonment, error, time, accessibility and the burden shifted to users or intermediaries.

Comprehension: Does the person understand what is being asked, what options exist and what material consequences may follow? Comprehension is partly a usability concern, but it deserves separate attention in consequential services because correct operation can occur without informed interaction.

Capability: Does the service create a realistic opportunity to accomplish what matters to the person? This asks whether interface functions become substantive options once resources and structural constraints are considered. It is not a promise that a product team can remove every social disadvantage.

Agency: Can the person meaningfully choose, refuse, modify or influence the process and its use of data? Agency includes individual and collective action. A choice is meaningful only when it can be exercised without an unreasonable penalty.

Recourse: Can an adverse decision, data error or user mistake be understood, corrected and contested through an accessible process? Recourse includes explanation, reversibility, appeal, accountable human assistance and recovery. It is not an optional customer-support layer; it is how agency continues when the ordinary path fails.

Outcome: Did the interaction contribute to the intended real-world result, and are burdens, risks and benefits distributed fairly? Equitable does not mean identical. It means looking for systematic differences in who receives the service, waits longer, pays more, discloses more, bears more verification work or absorbs the cost of failure.

The dimensions are defensible as questions, but the sequence has limits. Comprehension can be treated as part of usability. Agency and capability are mutually reinforcing rather than consecutive stages. Recourse may be needed at any point, not only after agency. Outcomes depend on policy, staffing, infrastructure and social conditions well beyond UX. For those reasons, the heuristic should be used to expose assumptions and investigation gaps, not to generate a single inclusion score.

Why recourse and outcomes belong in the inclusion conversation

The case for recourse is especially strong in public and essential services, where participation may be compulsory in practice and error can affect income, education, health or legal status. A 2026 systematic review by Hegemann and Ritz found that digital public services can reduce administrative burden but can also transfer learning, compliance and psychological costs to citizens. These burdens are produced by technology, organizational arrangements and individual circumstances together.

Holler, Tarshish and Kaplan’s interviews with social-benefit recipients in Israel show what this looks like after a system error. Correcting administrative mistakes created new learning, financial and emotional costs, often requiring persistent human contact. The study is small and context-specific, but it clarifies the principle: if access to an entitlement depends on a digital or administrative record, the ability to repair that record is part of effective inclusion.

India’s welfare systems show the need to analyse trade-offs rather than assign simple verdicts. Muralidharan, Niehaus and Sukhtankar found that stricter biometric identity verification reduced corruption in major welfare programmes, but also caused 1.5 to 2 million legitimate beneficiaries to lose access at some point. Much of the harm arose from transition arrangements. A technically functioning verification regime can therefore improve one public objective while imposing exclusion on eligible people. Design, rollout, exception handling and recourse determine who carries that trade-off.

This is also why non-digital routes matter. The UN’s Universal Digital Public Infrastructure Safeguards Framework says access to essential services should not become digitally mandatory in policy or practice; it calls for accessible complaint and appeal mechanisms and assessment of real-world impact, not only technical adherence. That framework does not validate the heuristic proposed here. It does show that agency, meaningful alternatives and effective remedies are increasingly recognized as governance requirements for inclusive digital infrastructure.

What organizations should measure differently

Organizations do not need to abandon access, accessibility or task metrics. They need to connect them to evidence about consequence. A service dashboard might show who never arrives, who abandons, who completes without understanding important terms, who relies on an intermediary, who has no viable alternative, who attempts a correction, who reaches a human, and who ultimately receives the intended service.

That inquiry should be disaggregated. Aggregate completion can conceal a group that spends twice as long, fails authentication more often, must travel for help or cannot risk an appeal. Qualitative research is equally important because logs rarely reveal coercion, confusion, loss of dignity or the informal work required to make a transaction succeed.

Responsibility also has to follow the service, not stop at the interface. Product and UX teams cannot by themselves correct discriminatory eligibility rules, weak civil-registration systems, staffing shortages or unequal access to income and education. But commissioning organizations can trace where harm is created, assign ownership across policy, operations and technology, preserve alternative channels, and test whether a nominally successful transaction produces a successful service outcome.

The practical change is a shift in the object of evaluation. Do not ask only whether the user completed the journey. Ask whether the institution completed its obligation to the user, on terms the user could understand, influence and challenge.

From digital access to meaningful agency

Putting a service online can widen reach. Making it usable can reduce a major source of exclusion. Neither achievement should be diminished. Yet neither establishes that people have been included in what the service actually distributes: opportunity, security, recognition, resources or voice.

A stronger proposition for inclusive UX is therefore this: inclusive digital services should be judged not only by whether people can complete interactions, but by whether those interactions contribute to meaningful capability, agency, workable recourse and reasonably equitable outcomes. Comprehension connects operation to informed action; power and institutional conditions shape the entire chain.

The seven-layer heuristic offers a way to make that proposition testable, but it remains provisional. Its value will depend on empirical work: studying where the dimensions overlap, which predict outcomes in particular services, and whose account of success determines the measure. The next phase of digital-inclusion practice is not to replace access or usability. It is to follow the interaction far enough to learn whether inclusion actually happened.

Questions for further research

Five questions for empirical investigation

1. Across selected Bangladeshi public services, how often does successful digital submission fail to produce the intended service outcome, for whom, and why?
2. Which forms of explanation and human assistance most improve comprehension and successful correction after an adverse or unclear decision?
3. When do family members, community intermediaries and frontline workers expand users’ agency, and when do they introduce privacy, dependency or gatekeeping risks?
4. Which measures within the seven-layer heuristic predict equitable outcomes, and which dimensions overlap too strongly to justify separate measurement?
5. How do digital-only requirements and authentication rules redistribute time, cost, disclosure and error-recovery burdens across different social groups?

---

Sources and further reading

• Ganapathy, A., Heeks, R. & Iazzolino, G., “Analyzing Inclusion and Inequalities in ICT4D: A Critical Literature Review of Themes and Gaps,” Information Technology for Development 32(2), 176–220: DOI
• Scheerder, A., van Deursen, A. & van Dijk, J., “Determinants of Internet skills, uses and outcomes: A systematic review of the second- and third-level digital divide,” Telematics and Informatics 34(8), 1607–1624: DOI
• van Deursen, A.J.A.M. & Helsper, E.J., “The third-level digital divide: Who benefits most from being online?”: DOI
• Heeks, R., “Digital inequality beyond the digital divide: conceptualizing adverse digital incorporation in the global South,” Information Technology for Development 28(4), 688–704: DOI
• Zheng, Y. & Walsham, G., “Inequality of What? Social Exclusion in the e-Society as Capability Deprivation,” Information Technology & People 21(3), 222–243: DOI
• Kleine, D., “ICT4WHAT? Using the Choice Framework to operationalise the capability approach to development,” Journal of International Development 22(5), 674–692: DOI
• Anwar, S. & Stillman, L., “ICT Use, Collective Agency and Community Transformation in Rural Bangladesh,” Journal of Community Informatics 22(1), 1–22: DOI
• BRAC Institute of Governance and Development, “Digital Literacy and Access to Public Services in Bangladesh”: Report
• UNDP Bangladesh & a2i, “Bridging the Digital Divide: Advancing Accessibility for Persons with Disabilities in Bangladesh’s Digital Ecosystem”: Report
• Hegemann, L. & Ritz, A., “Administrative burden in digital service provision: a systematic literature review and research agenda,” Perspectives on Public Management and Governance: DOI
• Holler, R., Tarshish, N. & Kaplan, S., “‘That’s where the hunt for the correction begins’: Experiences of administrative errors as sites of administrative burden,” International Journal of Social Welfare 33(4), 806–818: DOI
• Muralidharan, K., Niehaus, P. & Sukhtankar, S., “Identity Verification Standards in Welfare Programs: Experimental Evidence from India,” Review of Economics and Statistics 107(2), 372–392: DOI
• United Nations, “Universal Digital Public Infrastructure Safeguards Framework”: Report