Radio listenership is measured in three main ways: by asking a sample what they listened to, by having people keep a diary, or by metering what a panel hears. In Kenya the regular published figures come from recall surveys by phone and mobile. Each method defines a listener differently, so the method has to travel with the number.
Key points:
- Kenyan listenership figures on record come from recall: telephone interviews in the tracking survey the regulator used in 2019, and mobile surveys at GeoPoll.
- Diaries remain the core of the UK's RAJAR survey, with 15,000 people filling in a seven-day diary each quarter.
- Meters measure the 48 largest US markets, and a phone app panel supplements RAJAR.
- The rule for what counts as listening moves the result: five minutes at RAJAR, eight at Australia's GfK ratings, three at Nielsen's meters.
- A station's stream logs and on-air replies measure other things, and they don't belong in a listenership figure.
How is radio listenership measured in Kenya?
The clearest published account is in the regulator's report for July to September 2019. The Kenya Audience Research Foundation ran a fixed panel with a target of 3,000 interviews a day nationally, among people aged 15 and over. Panellists were recruited face to face, interviewed by phone, and reported in 30-minute blocks each month.
GeoPoll takes a mobile route. It recruits panellists and sends surveys by SMS and mobile app several times a day, asking about the previous four hours, with radio usually reported in two-hour blocks.
Ipsos uses the phone. Its 2022 PAMRO paper describes a quarterly Kenyan tracker of 2,500 people, built on weekly phone panels after a move away from a self-filling mobile app.
The history matters. TIFA Research traced Kenyan practice from paper diaries through mobile apps to day-after recall, and Ipsos told PAMRO that self-filling e-diaries failed in Kenya. Its own tracker has since moved to interviewers on the phone.
What is day-after recall?
Day-after recall asks people what they listened to yesterday, usually part of the day at a time, with a prompted list of stations. It is cheap to run, it covers a whole week across a sample, and memory of yesterday is fairly fresh.
The weak spot is memory. A person remembers the station they chose, and forgets the one playing in the matatu. Day-after recall explains how to run it.
What is four-hour recall?
Four-hour recall shortens the memory gap. GeoPoll assigns respondents at random to one of four dispatch times, 10:00, 14:00, 18:00 and 22:00, and asks only about the four hours before.
The trade-off is detail by time. Radio is mostly reported in eight two-hour blocks between 06:00 and 22:00. That makes it hard to split a show that ends at 09:00 from the one before it, though it gives fast, frequent data.
How does a radio diary survey work?
A diary asks each person to log their listening for a set period. RAJAR's 15,000 quarterly diarists record seven days, online, on mobile or on paper. Australia's GfK Radio Ratings draw on more than 60,000 people aged 10 and over each year, each taking part for one week.
In the US, Nielsen uses diaries outside its metered markets, with address-based samples of about 1,000 to 6,000 people per market. These are large, long-running operations. Kenya tried diaries and moved on.
Radio diary surveys covers where they work and where they fall short.
How do radio meters work?
A meter logs the audio around a person with no questionnaire at all. There are two families.
The Portable People Meter is a device the panel member wears, which detects an inaudible code added to each station's audio. MediaCell, the app used on RAJAR's panel of about 5,000, takes short samples of ambient sound from a smartphone and matches them against what every station was airing at that moment.
Both need a recruited panel, and both record exposure. People meters explains the devices and how the station is identified.
What does a radio listenership survey ask?
A sound listenership survey moves from the general to the specific. Here is the usual order.
- Did you listen to any radio yesterday?
- Which stations, from a prompted list?
- In which parts of the day?
- Where were you, and on what device?
- How many days in a typical week do you listen to each station?
Order matters. Ask about a favourite station first and people answer for their habits, and you lose yesterday. Radio listener survey questions has worked examples.
What counts as listening?
Each system sets a threshold. The threshold decides the number.
| System | What counts | Unit reported |
|---|---|---|
| RAJAR | At least 5 minutes within a quarter hour | Quarter hours |
| GfK Radio Ratings, Australia | 8 minutes or more in a quarter hour | Quarter hours |
| Nielsen PPM, US | 3 minutes or more | Quarter hours |
| KARF tracking survey, Kenya, 2019 | What the respondent reports | 30-minute blocks |
| GeoPoll, Kenya, 2022 | What the respondent reports | 2-hour blocks |
Nielsen shows how much this matters. It says 45% of metered listening occasions run under five minutes, and that lowering the qualifier to three minutes in January 2025 raised measured average audience by 24% on average. Listeners didn't change that month. The ruler did.
Who does each method leave out?
Every mode has a blind spot. An app panel needs a smartphone, and Ipsos put smartphone ownership at 51% in 2022, up from 19% in 2015, with basic phones holding at 50%.
A phone survey needs a working number and a willing person. Ipsos notes that refusal to take calls from strangers affects telephone interviews in markets such as Ghana. Face-to-face work reaches further and costs more.
That is why an establishment survey done face to face matters so much. It tells a phone or mobile tracker who it has to represent. Phone or face-to-face covers the choice.
Why do different methods give different numbers?
Memory favours habit. A recall survey tends to credit the station people think of as theirs, and passive methods catch the rest.
TIFA Research's Nairobi test in April and May 2019 logged an average of 5.2 radio stations a day per person, across 400 people and 75 radio stations. TIFA's own conclusion was that passive methods always capture more channels, and that passive results "should NOT be compared with data from other methodologies."
RAJAR handles this by calibration. Its model matches the panel's listening levels to those of the 15,000 diarists, so one established measure governs the other.
How is a listenership figure turned into a number of people?
Weighting and scaling. The survey finds a percentage, the sample is weighted to match the population by age, sex and place, and the percentage is multiplied by the population.
So the population figure matters as much as the survey. The regulator's August 2026 report prints a base of 32.23 million, and Ipsos in 2022 worked from about 34 million people above the age of 12. A radio total of 24.4 million and one of 27.3 million can't be compared until you line those bases up. Check the base before you compare.
Is radio listenership declining?
On the regulator's figures it has fallen, and the record needs care. Its 2023 report says radio stood above 90% in 2014 and 77% in 2023, and the August 2026 report shows 75% for April to June 2026.
The catch is that the source changed. Figures to 2019 came from KARF, and the 2023 one from the Authority's own national survey. Radio listenership in Kenya takes this further.
Can a station measure its own listenership?
It can commission or run a survey. It can't read listenership off its own systems.
A streaming server counts connections, and a phone line counts volunteers. Read why on-air responses do not measure reach and actual radio listenership before quoting either.
How often should listenership be measured?
Often enough to see a schedule change work. For most stations that means two to four waves a year, run the same way each time.
Keep the timing steady. A wave during an election month or a school holiday shouldn't be read against a quiet month without saying so.
Which method should a Kenyan station use?
Use the method you can repeat. For most stations that is a day-after recall survey in the counties you cover, weighted to KNBS totals and run the same way every wave.
Practitioners tend to chase the newest method and lose their trend. A modest survey run four times with the same questions tells a programme director more than one impressive study. When a new method does arrive, run it alongside the old one for a wave, and report the two as separate series.
Read next
- Audience measurement software for Kenyan media: the TI pack page for this work
- How media reach is measured in Kenya: all the pages in this section
- Day-after recall
- Radio diary surveys
- Radio listenership in Kenya
- Takwimu Intelligence (TI) home page
- Audience measurement in Kenya: the guide to this library
- The TI platform and how we check our numbers
Where TI fits
A listenership figure is only as good as the question behind it, asked the same way each time. Takwimu Intelligence (TI) asks about radio, TV and news yesterday, with a prompted outlet list and time blocks, and reports station share, daily and weekly reach, reach by daypart and days a week listened. Each figure shows its margin of error.
To see it on your own station or title, book a walk-through.