Designing Swahili-First Interfaces

Start in Kiswahili

Kiswahili and English are both official languages in Tanzania, and Kiswahili is the language most people actually use with one another. A 2022 localization guide from Localization Africa puts it plainly: messages in this market are widely conveyed in Swahili, which is why Vodacom and Airtel already run product names, slogans and service menus in both languages.

Figures reported from the FinScope Tanzania 2023 survey make the design consequence obvious. About 79 percent of people were literate in Kiswahili, while 61 percent could not read or write in English. Thirteen percent had no formal education and 56 percent had primary education. An English-only screen is not a neutral default. It is a locked door.

Translation is the last step

Swahili uses the Latin alphabet and reads left to right, so you do not need a right-to-left layout or a special font. UTF-8 is enough, and ordinary UI typefaces render it. That is the easy part, and it is why teams stop too early.

The hard part is length, tone and the words people trust. Swahili labels are often longer than the English strings they replace, so buttons, tabs and table headers that were drawn around English wrap or truncate. A native speaker has to read the live interface, not a spreadsheet, because a correct dictionary word can still sound like a bank talking down to someone.

A public note on Tanzanian banking gives a concrete example. A transaction screen used “Miamara” where customers expected “Miamala”. One wrong consonant turns a payment label into nonsense. That is a trust failure, not a copy edit.

What published product tests already showed

There is no public census that says forty products were audited, so this note does not pretend there was one. The pattern is already visible in published work.

A usability study of M-Afya, a Swahili USSD health service written with clinicians at Muhimbili, found that clinical words such as insulin, glucose and ovulation were not understood by the women the service was for. The same study found that a “Rudi” (back) option existed and people still could not tell which key recovered them from a mistake. More than 12,000 women subscribed in the first year after the content was rewritten for that audience. The lesson is to test comprehension of the words, and to make the way back obvious.

A concept test around Deezer in Tanzania found that 21 of 30 people struggled to register because the only paths were email or Facebook, and the page was in English. Many people in that test did not have an email address. The practical fix recorded there was a phone-number login and a Swahili interface, plus local music at the front of the catalogue.

A working order

Write the first version of each flow in Kiswahili, then add English. Budget the layout for the longer string. Sign people in with a phone number. Replace jargon with the word a customer would say at a counter. Put the recovery action where the error appears. Read every string out loud with someone who did not write it, on a Tecno, Infinix or Galaxy A-series phone, not only on the laptop used to design it.