Independent digital publicationZimbabwe · South Africa · Wider African region
The Kopje Journal & ReviewForensic analysis for the African perspective.

Audit

Within Signal Range, Still Offline

The larger mobile-connectivity gap is now among people who live under a network but do not use it. Device ownership, recurring costs and practical capabilities explain why another coverage announcement cannot, alone, establish digital inclusion.

The audit question

Within Signal Range, Still Offline

A mobile network can reach a community without putting its residents online. The missing link may be a suitable phone, money for recurring data or the ability to complete an unfamiliar task safely. Coverage is necessary where there is no signal. It is not evidence that every person inside the coverage area can use the service.

GSMA Intelligence's September 2026 connectivity report puts the distinction in numerical terms. At the end of 2025, it estimated that 59% of the global population used mobile internet, while 38% lived within mobile-broadband coverage but did not use it. Only 3% remained outside that coverage. These are the industry's mobile-internet categories, not a measure of all ways of using the internet. GSMA Intelligence, State of Mobile Internet Connectivity 2026.

The larger gap is consequently not a blank patch on a network map. It is the difference between a technically reachable service and a service people can actually use. That changes how a claim of digital progress should be tested.

Count the person, not just the connection

ITU estimates that 76% of the world's population uses the internet in 2026. That is not directly comparable with the GSMA mobile measure: the dates and access categories differ. The figures should not be subtracted to produce an invented count of people using only fixed connections. ITU, Facts and Figures 2026.

A similar caution applies to subscriptions. The World Bank's 2025 Digital Progress and Trends Report explains that subscription totals can include multiple subscriptions held by one user, business connections and connected devices. More subscriptions do not necessarily mean an equal number of additional people online. World Bank, connectivity chapter.

For a public programme, this is more than statistical housekeeping. A connection count answers how many services exist. A user count answers how many people participate. Neither, on its own, establishes whether a person can reliably submit a form, follow a lesson or communicate when needed. Reporting all three as simply “connected” conceals the difference between infrastructure, adoption and usable service.

A phone is not always the required device

The World Bank's Global Findex 2025 release, drawing on 2024 observations, reports that 86% of adults worldwide owned a mobile phone, while 68% owned a smartphone. The denominator is adults, not the whole population. The distinction also means that mobile ownership cannot be treated as universal possession of the same capabilities. World Bank, Global Findex release.

Consider an illustrative service, not a surveyed household: an application requires somebody to upload a document. A basic phone used successfully for calls does not establish that its owner can complete that application. A borrowed smartphone may make the task possible, but only when the owner makes it available. A device-sharing arrangement therefore answers a different question from independent access.

The appropriate test begins with the activity. What equipment does it require? Who controls the equipment? Can the user access it when the deadline arrives? A device distribution programme can address one constraint without resolving the others. Counting handsets delivered is a useful implementation measure, but not a complete measure of the service people gained.

Affordability returns every month

ITU's 2026 price assessment uses a benchmark of 2% of gross national income per person for its broadband baskets. For the mobile basket, 138 of 202 economies with available data met that benchmark. The benchmark compares prices with national income; it does not establish affordability for each low-income household. ITU, affordability assessment.

The arithmetic illustrates the limitation. A hypothetical monthly service costing US$4 is 2% of US$200, but 8% of US$50. Neither figure is a quotation or an estimate of a particular country's household income. It shows why the same price can satisfy a national benchmark while imposing a much larger burden on someone with fewer resources.

A handset purchase and a continuing data bill also belong to different budgets. Removing the purchase barrier does not guarantee money for next month's usage. Conversely, cheaper data is not enough for somebody without a suitable device. Evaluating either intervention alone risks attributing non-use to personal reluctance when the remaining obstacle has not been measured.

For the user, the relevant cost is the package required to perform the task, at the time it must be performed. An affordable small allowance does not establish that a larger recurring requirement is affordable. Comparing current operators' packages or recommending a provider would require dated, like-for-like quotations, which are outside this article's scope.

Going online is not the same as knowing how

ITU measures digital skills through activities people report performing, grouped into communication, problem solving, safety, content creation and information literacy. Its 2026 assessment warns that the evidence is incomplete: only 22 reporting countries supplied overall skill levels across all five areas, and these tended to have high internet use. Their results cannot stand in for every country. ITU, digital-skills assessment.

The conceptual distinction is still useful. Being comfortable with messages does not demonstrate equal confidence in assessing an unfamiliar website, securing an account or submitting an electronic application. Those are different activities. A service can be technically available while remaining difficult to use without help.

Support should therefore be evaluated against the intended task, not just attendance at a training session. Can a participant repeat the task afterwards? Is assistance available when something fails? These are proposed outcome tests, not a claim that a particular programme has failed them. They turn the vague category of skills into something observable without blaming users for a poorly designed service.

A public-service transition makes the distinction particularly consequential. If an application moves online, an existing paper route and a newly available website are not automatically equivalent for every applicant. Evaluating that transition means finding out who can complete the process independently, who needs assisted access and who cannot complete it at all. That does not establish that digital delivery should be abandoned. It identifies the missing test before a coverage statistic is used to justify treating an online route as universally available. Successful digitisation is a claim about users reaching an outcome, not only about an institution launching a website.

Quality remains part of access

The World Bank's 2025 report found that median fixed and mobile download speeds remained below 25 megabits per second in low-income and lower-middle-income country groups in 2023 and 2024. That historical comparison describes uneven service quality; it is not a speed test of a particular village or a universal minimum requirement for every task. World Bank, chapter 2, figure 2.7.

A useful local assessment would match service performance to the activity being promised. It would distinguish somebody who has never used the service from somebody who uses it intermittently, and both from somebody unable to finish a task because the connection fails. Combining them into one offline category would obscure which intervention could help.

The evidence supports a layered definition of progress. Extend coverage where it is absent; measure device access, recurring affordability and practical capability where coverage already exists; then test whether the resulting service works for its intended purpose. A population inside signal range is not yet a population able to participate. The meaningful result is the task a person can complete, not merely the network that passes nearby.

Evidence & accountability

Sources & editorial status

  1. GSMA Intelligence, State of Mobile Internet Connectivity 2026 ↗Source
  2. ITU, Facts and Figures 2026 ↗Source
  3. World Bank, connectivity chapter ↗Source
  4. World Bank, Global Findex release ↗Source
  5. ITU, affordability assessment ↗Source
  6. ITU, digital-skills assessment ↗Source
  7. World Bank, chapter 2, figure 2.7 ↗Source
Source note

By The Kopje Journal & Review. Original editorial research; sources linked in the article.

Send a correction or source ↗

Cookie choices

You can read every audit with optional media switched off. Your choice applies to this browser for up to 180 days.

Essential website functions

Security, Publisher sign-in and remembering your privacy choice.

Always active

No KJR advertising or analytics tags run on this version of the website. Social links and Stripe checkout open only when you follow a link.

Cookie policy · Privacy statement · European & UK rights