Audience measurement in Kenya works mainly by asking people what they listened to, watched and read, then scaling their answers up to the population. The regular Kenyan services that publish their methods are surveys, run by phone, by mobile or face to face. Each measures something slightly different, which is why two sound figures for one station can disagree.
Key points:
- The Communications Authority of Kenya publishes a quarterly audience report. Its August 2026 edition put seven-day radio use at 75% and TV at 74% for April to June 2026.
- GeoPoll and Ipsos sell station-level audience data, and their figures rest on different measures and periods.
- A station needs three separate facts: who could receive it (coverage), who actually tuned in (reach) and what they thought (appreciation).
- Recall surveys, streaming logs, on-air responses and advert logs measure different things, so they must never be added together.
- A figure with no base, period and method can't carry a sale or a media plan.
Why does audience measurement matter in Kenya?
Money moves on these numbers. The Communications Authority's report records KES 14.3 billion of advertising in April to June 2026: KES 7.9 billion on TV, KES 5.1 billion on radio and KES 1.3 billion on print. It counts a further KES 9.12 billion of digital spend for the same quarter.
Each of those shillings was placed on a belief about who was listening or watching. The regulator states the case in its own introduction: "Reliable and independent audience measurement is essential."
The market is also crowded. The Authority counted 265 FM stations and 253 TV stations operational at the end of March 2025. A national top ten leaves out 255 of those FM stations, so most stations have to produce their own evidence.
How is media audience measured?
Five kinds of evidence are in use. Each one counts a different thing.
| Evidence | What it counts | What it can show | What it can't show |
|---|---|---|---|
| Recall survey | What a sample says it heard, watched or read | Reach, share and audience profile, with a margin of error | Exact behaviour minute by minute |
| Meter or phone app panel | Audio a panel member was near | Exposure minute by minute for the population the panel represents | Attention or liking |
| Streaming log | Connections to a server | Sessions and hours online | Anyone on FM, or who sat behind the connection |
| On-air responses | People who chose to reply | How keen some listeners are, and what they say | Audience size |
| Advert log | Spots that aired | Whether and when an advert ran | Who heard it |
Read the table by row. A recall survey can estimate reach across FM and online together, because it asks the person. A streaming log counts hours precisely, but only for people on the stream.
Adding two rows together gives a number that means nothing. The overlap between them is unknown, and they don't share a unit.
What three facts does a station need?
Every station needs three facts, and they are different facts. Coverage is who could receive the signal. Reach is who actually tuned in. Appreciation is what they thought of what they heard.
Coverage comes from transmitter and signal data, reach from measuring people, and appreciation from asking them. Take an invented county station with 1,000,000 adults inside its signal. If a survey finds that 23% tuned in during the past week, its weekly reach is about 230,000, and the other 770,000 are coverage it hasn't turned into listeners.
The common mistake is to sell the first number as if it were the second. Read radio coverage, reach in media and the appreciation index in that order.
What does the Communications Authority report show?
The regulator's report is the national frame. For April to June 2026 it shows that 75% of people used radio in the past seven days, 74% TV, 57% the internet, 17% a newspaper and 6% a magazine. Daily use was 52% for TV, 49% for radio and 44% for social media.
The averages hide the real market. Among LSM 1 to 4, the lowest group on the Living Standards Measure, radio reached 94% and TV 24%. In LSM 12 and above those figures were 68% and 94%, and radio ran from 83% in Western to 45% in North Eastern.
The report gives no station figures. Our guide to the Communications Authority audience measurement report explains what it can carry and how it has changed.
Who measures radio and TV audiences in Kenya?
GeoPoll runs a mobile panel. Its audience measurement page describes surveys sent once every four hours, a method developed with Kantar Media, and ratings in a dashboard the next day.
Ipsos runs a phone tracker and establishment surveys. Its 2022 PAMRO paper described a quarterly Kenyan tracker of 2,500 people by phone, and its 2022 establishment survey targeted 10,000 households across all 47 counties, face to face.
The Media Council of Kenya adds a State of Media survey. Its 2024 report rests on fieldwork from 6 to 13 May 2025, with a target of 3,602 people aged 15 and over in all 47 counties.
On the public record, passive measurement in Kenya is an argument and a trial. Ipsos's Kenya head called it the "gold standard" in July 2022, and TIFA Research tested a phone app on 400 people in Nairobi in April and May 2019. Audience measurement companies in Kenya sets out each player.
Why do station rankings disagree?
Look at one station across three published sources.
| Source | Period | Measure | Citizen | Radio 47 |
|---|---|---|---|---|
| GeoPoll | January to March 2026 | Reach over the quarter | 64.7%, first | 36.1%, tenth |
| Ipsos, as reported by RadioInfo Africa | October to December 2025 | Share of national audience | 17%, first | 9%, second |
| Media Council | Fieldwork May 2025 | "In a typical week, which radio stations do you listen to?" | 29%, first | 9%, eighth |
Nothing here proves any source wrong. Reach counts people, share usually counts time, and a stations-listened-to question counts what people remember naming. A station with a smaller, more loyal audience can sit low on reach and high on share.
Rankings are one source's finding, for one period, on one measure. Quote all three or leave the rank out. Top radio stations in Kenya and the most watched TV station in Kenya show the published figures side by side.
How is radio listenership measured?
Kenya's published radio figures come from recall. GeoPoll asks about the past four hours and reports radio in two-hour blocks, while the tracking survey behind the regulator's 2019 report used telephone interviews and 30-minute blocks.
Abroad, the UK's RAJAR combines diaries with a phone panel, and Nielsen meters the 48 largest US markets. The definition of a listener varies as well. RAJAR needs five minutes in a quarter hour, and Nielsen says moving to three minutes raised measured average audience by 24% on average.
A listener is a definition before it is a number. How radio listenership is measured compares the methods.
How is television measured?
TV measurement abroad rests on metered panels. The UK's Barb panel covers 7,000 households, about 16,000 people, with a meter on each set and a button for each person. The hard part of TV is knowing who is in the room.
Kenya's published TV figures come from surveys. GeoPoll reports TV in 30-minute blocks, which suits half-hour programming and feeds TV ratings in Kenya. For the mechanics, read television audience measurement.
How is print readership measured?
Circulation counts copies. Readership counts people, and one copy can pass through several hands.
Readership is a survey finding. The regulator's 2025/26 tables work from 5.4 million newspaper readers, and 84% of people in April to June 2026 said they rarely or never read a newspaper. Ipsos put print at 6.2 million in 2022, among people aged above 12.
Those two figures use different age floors, so they don't show a clean trend. See newspaper readership in Kenya and newspaper circulation.
How is online audience measured?
Online is the most counted medium and the least understood. A server counts connections, and a platform counts accounts and impressions. Neither knows who the person is. The same goes for big TV datasets, which Nielsen says fail to register who is in front of the screen.
Surveys can count people, at the price of relying on memory. The regulator's 57% internet figure is a survey finding. The Reuters Institute's 2026 Kenya data come from an online survey of mainly English-speaking news users aged 18 to 50, and the report says the findings should not be taken as nationally representative.
Use online news audiences in Kenya and social media audiences in Kenya with those limits in mind.
What do practitioners get wrong?
Four mistakes account for most bad claims.
- Adding sources together. A station combines survey reach, stream counts, Facebook followers and SMS replies into one figure, though the sources overlap and share no unit.
- Selling coverage as reach. Everyone inside the signal is counted as a listener.
- Quoting a rank with no base. "Number one" means nothing until you know among whom, when and on what measure.
- Comparing waves across methods. TIFA Research put it plainly in 2019: "The results for passive should NOT be compared with data from other methodologies."
The fourth applies to any change of mode or question. Keep a method steady, or treat the change as a new series.
What should you ask before trusting a figure?
Six questions settle most arguments.
- Who measured it, and who paid?
- When was the fieldwork?
- Who is the population, and what is the base?
- What is the measure, reach, share or rating, and over what period?
- How were people asked, and what counted as listening or viewing?
- What is the margin of error, and was the sample chosen or did people volunteer?
A seller who can answer all six has a measurement. One who can't has a claim.
What does good practice look like for a station?
It looks dull, and it works. Know your coverage from your signal data. Measure reach in the counties you serve, with a fair sample and the same questions each wave.
Ask what people thought of the programmes that matter. Keep your stream figures, on-air replies and advert logs in their own lines, each with its own label.
Then state every figure with its period, base and margin. A buyer who can check your number will trust the next one, and a programme director can act on a trend in a way nobody can act on a single headline.
Where should you start in this library?
Start with your question. If you want the methods, begin with how media measure reach, and keep audience metrics open beside it for reach, share, ratings and cost.
If you need your own figures, go to how to run an audience survey in Kenya. If you sell or buy airtime, read media advertising in Kenya.
For the market itself, read radio in Kenya, television in Kenya and print and digital audiences in Kenya. For the rules, read broadcasting regulation in Kenya. To see the work done on software, go to audience research software for Kenyan media.
Read next
- Audience measurement software for Kenyan media: the TI pack page for this work
- How media reach is measured in Kenya
- Audience metrics explained
- How to run an audience survey in Kenya
- Takwimu Intelligence (TI) home page
- The TI platform and how we check our numbers
Where TI fits
A station's reach, appreciation and on-air replies belong in separate columns, each with its own method. Takwimu Intelligence (TI) runs cross-media day-after recall surveys and programme panels, weights answers to KNBS census totals at county level, and shows the margin of error on every figure. Its broadcast monitoring is a separate record of what went out on air.
To see it on your own station or title, book a walk-through.