The mobile-internet market is a usage gap, not a coverage map

GSMA's 2026 mobile report places Africa's mobile-internet usage gap at 63% of the population and its coverage gap at 9%. The first group lives within mobile-broadband reach without using the service. The second lacks coverage.

Those groups need different capital and policy. A new radio can help the uncovered location. It cannot make a handset affordable to a covered household.

Coverage is not adoption.

Preserve the denominators

The percentages describe shares of population under GSMA's method, not shares of subscriptions or adults. One person can hold several SIMs; a household can share a device. Subscriber counts should not be subtracted from population without adjustment.

Coverage is a modeled geographic and population measure with technology and signal assumptions. It does not guarantee indoor service, capacity or consistent speed at every point. Field experience can differ.

Usage needs a recent threshold. A person who used mobile internet once and an active daily user create different demand. Read the report definition before comparing with another survey.

Regional percentages hide country and local variation. ITU's work highlights income, gender and urban-rural gaps. A national product plan needs national and subnational evidence.

Branch one: no compatible device

The household may own a basic phone, share a smartphone or use an old device with limited storage and battery. Retail price includes import cost, currency, tax, distribution, finance and margin.

Device financing lowers upfront cash and adds instalments, eligibility and repossession or lock risk. Compare total paid and contract terms. A cheap monthly amount can extend beyond the device's expected useful life.

Refurbished devices can broaden access and need quality, warranty, battery and security support. Unsupported operating systems can expose users and prevent applications from running.

A policy lowering device tax should track retail price and volume after existing stock clears. The tax change may be absorbed elsewhere in the chain.

Branch two: recurring data is unaffordable

A household chooses data beside food, transport, power and rent. A low unit price can remain unaffordable when income is irregular or packages expire before use.

Measure the cost of a useful basket, not the smallest bundle. The basket should fit actual services, including updates and failed sessions. Promotions can obscure the standard price.

Zero-rated services reduce data cost for selected content and can influence competition and user choice. Evaluate access benefit and market effect under local rules.

Applications can reduce burden through compression, smaller updates, offline work and clear data estimates. Product teams control some affordability.

Branch three: service has no trusted use

People adopt when a service solves a job safely enough. Local content, language, digital skill and trust influence the decision. “Awareness” is too broad to diagnose a failed registration.

Observe the journey: finding the service, installing, granting permissions, creating an account, recovering access, paying and getting support. Each exit has a different remedy.

Fraud can make abstention rational. Strong authentication, confirmation, complaints and accountable providers are adoption infrastructure. Marketing cannot compensate for unresolved loss.

Government and employer services can create reasons to connect and can exclude people when offline alternatives disappear too quickly.

Branch four: network experience is weak

Living within coverage does not guarantee capacity at the time a user can afford data. Congestion, latency and outages can make a paid session fail.

Measure completion of common tasks by location, device and time. Peak download speed is insufficient. Failed mobile-money, video or form sessions carry different cost.

Indoor reach and power interruption matter. A network site with backup can stay live while the user's phone is uncharged. Community charging adds money and travel.

Operators should combine radio data with customer reports without exposing individual behaviour. A usage gap partly caused by quality needs capacity or operations, not another awareness campaign.

The operator's household model

Build a set of household cases rather than one average. Record income timing, device ownership, shared use, charging, desired tasks, data purchase and trust concerns. Do not treat cases as national prevalence without representative sampling.

For each case, identify the binding constraint and the cost to change it. A device subsidy for a household constrained by data may produce an inactive phone. A cheaper bundle for a person who fears fraud may go unused.

Repeat the interviews after an intervention and keep households that did not adopt in the sample. Ask what they spent, which tasks they completed, who else used the device and why sessions stopped. Self-reported affordability, transaction records and network activity answer different questions; compare them without presenting any one as the complete household budget.

Commercial offers can combine device, data and service while increasing contract complexity. Disclose total cost and cancellation. Test whether support can explain it.

Track 30- and 90-day active use after acquisition. Connections sold are an input; sustained useful activity is closer to the outcome.

Coverage still matters

Nine percent outside mobile-broadband coverage represents a substantial population. Remote deployment can require lower-cost spectrum, sharing, subsidies, satellite or other technologies under local conditions.

Universal-service funding should publish project cost, new coverage, service quality, adoption and ongoing viability. A live site with few paying users may need explicit long-term support.

The usage and coverage agendas can compete for capital. Policy should state distribution goals and assess both, not assume the larger percentage automatically receives every next dollar.

Africa's mobile opportunity is often described through people still to be connected. GSMA's split refines the phrase. Most of the measured gap is already under a signal; the missing link sits in a household decision.

Sources

• GSMA, 16 June 2026: The Mobile Economy Africa 2026
• International Telecommunication Union, April 2025: Measuring digital development in Africa