Eswatini’s online outage interrupted passport and ID services

The ICT ministry aimed to restore services by Wednesday evening. The reports reviewed did not confirm a full recovery.
A signal on the map is a starting point. Useful access also depends on the cost of completing a task, the device, the connection and the service waiting at the other end.

An internet connection becomes useful when someone can finish something: submit an application, receive a customer’s order, check a public notice or send a document. A coverage map cannot tell the whole story of that experience. Neither can the amount of data sold. A phone may show a signal while the task still fails, costs too much or requires skills and equipment its user does not have.
In a July 2025 release, the World Bank reported that nearly 95% of Eswatini’s population was covered by 4G, while about 58% used the internet. It identified data costs as an obstacle. These are dated figures from that release, not a new measurement for November 2026. The gap is a reason to examine useful access; it does not, by itself, explain every person’s circumstances. [1]
Our practical framework has four questions. Is a connection available where the task happens? Is the full cost manageable? Is it reliable enough to complete the task? Can the person use the device and service successfully? These are questions for comparing experiences, not a new official index or a survey result.
The answers can differ within one household. One person may need a short message, another a document upload and another a long video lesson. The same bundle therefore buys different amounts of useful activity. A claim that a service is “cheap” needs a reference point: cheap for which task, over which period and for which budget?
Consider an entirely hypothetical comparison. Package A costs E100 for 2 GB, valid for 30 days. Package B costs E60 for 1 GB, valid for seven days. Assume a user needs exactly 1 GB in each of four consecutive seven-day periods, with no rollover, fees or other restrictions. These are invented packages, not current offers from any operator.
Over that 28-day window, buying Package B once each week costs E240 for 4 GB. Two purchases of Package A cost E200 for 4 GB, assuming the data can be used as needed throughout the window. A lower purchase price has not produced a lower total cost. But the E100 purchase still requires more cash at one moment than the E60 purchase. Total cost and the ability to pay upfront are different constraints.
The example proves no real package is better. Actual comparisons must check validity, permitted uses, coverage, renewal rules and what happens when an allowance runs out. If usage changes, the calculation changes too. A low unit price for data that expires unused may be less helpful than a smaller purchase matched to a specific need.
A second hypothetical example concerns an online form. Suppose it takes E10 of data to complete at home, but a failed attempt leads to an E30 trip to somewhere with another connection. The connection’s useful cost now includes the unsuccessful attempt and the trip, as well as time. Those amounts are illustrative, not measured Eswatini fares or data prices.
For a real comparison, keep a short record of the task, date, place, data used where visible, payment and outcome. Note whether the failure came from the connection, the device or the website. A repeated error from one service is different from every site failing to load. This record cannot diagnose a network by itself, but it can make a complaint or a later comparison more specific.
Imagine a small seller creating a digital catalogue. A customer must be able to view it, ask a question, receive a clear reply and complete an agreed purchase. The seller must then fulfil the order. Better connectivity can help this chain, but it cannot settle stock shortages, unclear delivery arrangements or a price that the customer rejects.
A useful trial would select one task, such as answering existing customers’ enquiries more consistently, and record its results for a fixed period. Count enquiries answered, confirmed orders and completed deliveries separately. A page view is not an order; an order is not money retained after costs. This is an illustrative test design, not a reported result from a local business.
UNDP’s December 2024 digital-readiness publication organised Eswatini’s assessment around six pillars: digital public infrastructure, connectivity, government, regulation, economy and people. That framework is broader than network construction alone. [2]
For a household or firm, the implication is practical: identify the step that actually blocks completion. A larger allowance will not fix a broken upload form. A new application will not automatically help someone who cannot navigate it. A useful service should make its requirements clear, explain errors in understandable language and let users recover from an interrupted process. These are design proposals, not claims that any particular Eswatini service currently fails those tests.
A news page can apply the same reasoning. Put the essential fact, date and source in readable text. Offer a clearly labelled link to the original document. Let the reader choose whether to play a video or load a large file. Explain how to check whether a notice is still valid. These are editorial choices that reduce unnecessary steps; they should be tested with readers rather than treated as proven national preferences.
Accessibility also means avoiding a false promise. A summary should say when an official document could not be opened or a detail remains unverified. A fast page carrying a wrong deadline is not useful access. Saving data and preserving accuracy have to work together.
On 28 May 2026, ESCCOM announced an ICT-sector survey involving selected households, businesses and institutions across all four regions, with collection scheduled through June. Its stated purposes included understanding the sector and supporting affordability, service quality and inclusion. That announcement is not a publication of the survey’s results, and we do not present it as one. [3]
Alongside infrastructure measures, useful questions would be whether an application was completed, a customer received a reply or a connection problem was resolved. Results should specify the period, the people or firms measured and the method. A small voluntary reader poll can help choose the next question to investigate; it cannot establish how the whole country behaves.
The World Bank’s 2025 release proposed telecom-market reforms, fair access to shared infrastructure, stronger digital skills and support for entrepreneurship. Those are policy recommendations from that dated publication, not outcomes we claim have already happened. [1]
The immediate test is smaller and more concrete: can a person complete a useful task at a cost they can sustain? Track where that chain breaks, compare like with like and distinguish an advertised allowance from a result. Coverage matters. Opportunity appears when the connection, the user and the service work together.
Media: World Bank
Prepared with AI assistance; sources linked below.
Published 18 Nov 2026
Suggest a correction
The ICT ministry aimed to restore services by Wednesday evening. The reports reviewed did not confirm a full recovery.

Zulu D quoted E100 for admission, and organisers are inviting families to bring picnic blankets.

The strategy sets 2030 targets for prevention, earlier diagnosis, treatment access and cancer-care capacity.