A usability test that passes on a MacBook and office wifi has not been run in Kenya. The failures that matter happen on a mid-range Android, a Safaricom or Airtel bundle, and a step that leaves your interface for M-Pesa, an OTP, or the camera.
You are not testing whether the file looks modern. You are testing whether a specific person can finish one job without you rescuing them. In Kenya that job is often “pay”, “send”, “prove who I am”, or “find the till number”, done while the network is ordinary rather than kind.
Afritesting recruits the people and the phones. You bring the Figma link, the staging build, or the production task, and you watch. We do not moderate. For a common consumer profile, unmoderated tests start at KES 2,000 a person. A moderated 45-minute session is KES 3,500 — the same as an interview — because someone still has to be screened, booked and replaced if they do not show.
If the prototype skips the M-Pesa prompt, the ID photo, or the “pending” state, you have tested the easy half. Put the ugly step in, even as a screenshot, or wait until a build can survive it.
Same price as a 45-minute interview, because the work is the same: screen, schedule, sit with them, replace a no-show. Use it when the flow hits M-Pesa, an OTP, a camera, or a prototype that still lies.
Join on their phone. Your laptop recording of a mirrored desktop is a different device.
They do it on their own time. Eight tests are typically about KES 16,000, incentive included. Use it to compare two labels or confirm a fix. Do not use it to discover why a payment felt unsafe — you will get a shrug and a dropped recording.
If the build needs you to explain “ignore that error”, it is not ready to be unmoderated.
“Tap Pay” tells them the answer. A situation does not. Four that usually earn the session:
Recruit people who have done a cousin of the task in the last month, on the phone they will use in the test. Staff who work on payments will perform the happy path and teach you nothing. If the product is a wallet or a loan, pair this with the audience cuts in fintech user research — “has M-Pesa” is not a usability sample.
Five moderated sessions on one flow will show the breaks. They will not tell you whether the market wants the product. That is a research plan, or a later survey, not a longer usability day. For why a global testing tool’s Kenya seats miss this, see recruiting for product research.
Unmoderated tests start at KES 2,000 per participant. Eight tests are typically about KES 16,000. A moderated live test is KES 3,500 per participant for a typical 45-minute consumer session, the same as a video interview, because the recruitment and scheduling work is the same. Incentives are included. Harder audiences cost more, and you see the quote first.
Moderate the first tests, and anything that leaves your product for M-Pesa, an OTP, or a photo of an ID. Unmoderated is for a short, stable task you can score without watching the network fail in real time.
They should. A test on a lab iPhone or office wifi hides the Tecno or Samsung A-series, the bundle, and the notification shade your users actually have. Ask them to join the call from that phone.
No. A Figma prototype is enough for labels, order and comprehension. It is not enough for the M-Pesa handoff, camera permissions, or what happens when data drops. Test those on a build, even a rough one.
Name the task, the phone and whether you need to watch live. We’ll find the participants and price the study before you pay.