Most “Kenya UX research” is a round of sessions in Nairobi with people who resemble the team. That can be useful. It is not a picture of the country, and it will not tell you whether to cut a flow, change a promise, or leave cash in the product.
Write the decision in one line before you write a screener. “Do first-time borrowers understand what happens if they miss a week?” is research. “Learn about our users in Kenya” is a workshop. The first line tells you who must be in the sample, how long the session is, and when you are allowed to stop.
Kenyan product decisions that actually move the design are rarely about whether people have smartphones. They are about trust in a new name, who else touches the phone, what happens when the bundle ends, and which step still has to work in cash or with an agent. If the study cannot change one of those, it is too early to recruit.
Six sharp interviews that kill a flow are a better programme than a 40-person project that restates the brief. Field the next method only when the last one left a question you cannot answer by arguing.
People will complete a task in your prototype to be polite. The research is whatever they do on a normal Tuesday when you are not on the call.
A salaried worker in Kilimani can install an app alone. A contribution to a chama, a school-fee payment, or a shop’s supplier payment often needs a second person. Interview the one who moves the money, not the one who downloaded it for them.
Receipts in a drawer, a till number on the wall, a float counted at close. If your flow assumes the phone is the whole system, ask them to walk the last real transaction, including the parts your product will never see.
Pending, reversed, deducted, or “call the agent”. Those are different products. A usability test can show the break. An interview tells you which failure they have already learned to fear. Do the interview first if you do not know the fear.
You do not need all three. You need to know which question you are on. Recruitment for each is a separate brief — mixing “discover” and “watch them tap” in one session makes both worse.
Eight people who did the job in the last month. No prototype. You are collecting the sequence, the words, and the person who actually decides. Video interviews from KES 3,500 for a typical 45-minute consumer session.
Only after you know the job. Five moderated usability tests on the step most likely to die on a mid-range Android. Not a tour of the whole app.
When interviews disagree and you need a count — not a national poll. A short survey using their words. If you cannot yet say the question in one sentence, you are not ready to survey.
Run Nairobi if that is where you will launch, and label the readout that way. Add Mombasa, Kisumu or a county town when distance, cash, or a local routine is part of the product: a port, a lakeside market, a town where the agent is the branch. “Urban Kenya” as a single segment hides the difference you needed the study for.
The interface can stay in English. The session should not police language. Ask people to explain the task the way they would to a sibling. Quotes in Kiswahili or Sheng are data. A translator on the team, or a moderator who can follow, matters more than a perfectly English screener.
Households share phones. Shops hand the phone to whoever is free. If your participant is demonstrating someone else’s life, you will design for the most fluent person in the room. Ask who last completed the task, and recruit that person. The practical version of this, for businesses, is how to recruit SME owners.
Afritesting matches on location, work and behaviour, then screens without showing the qualifying answer. You bring the discussion guide. For common profiles, matching is aimed at 48 hours; a specialised audience takes longer. If someone is not who the screener said, they are replaced during the early service at no extra recruitment fee.
If you are still choosing between a global panel and a local recruit, start with user testing for Kenya. This page is the plan. That one is the supply.
Six to eight interviews with people who did the job in the last month, in the places you will actually launch. Use them to learn the workaround. Only then test the riskiest screen. A survey comes after you know which words people use.
Nairobi is enough if you are honest that the findings are about Nairobi. It is not a stand-in for Mombasa, Kisumu, a county town, or a household where the phone is shared. Add a second place when the behaviour is likely to change with distance, cash, or who controls the handset.
The interface may be English. The session should follow the language the participant uses for that task. Forcing English produces cleaner quotes and worse decisions. Ask them to speak the way they would explain it to a relative.
No. Afritesting recruits, screens and schedules participants in Kenya, and replaces people who do not match. Your team runs the interview or test on Google Meet, Zoom, Teams or your prototype.
Put that in the brief, with who must have done the task and where. We’ll match participants, and you run the sessions.