Radio station audience data earns money when it settles six decisions: which advertisers to chase, what to charge, what goes on air, why people leave, which presenters to back and whether a change worked. A station that buys data only to print a ranking in a sales deck pays for the survey and collects none of that.
Key points:
- Advertisers pay for proof that their customers listen, so the sales team should carry a short reach note with every rate card.
- A rate holds when you can show what each band delivers and who is in it.
- Check the transmitter before you touch the schedule, because a signal fault looks like a programme fault.
- A single station figure can hide a presenter losing listeners beside one holding them.
- A change has worked only when the next wave shows it moved, on the same questions and the same population.
Why does a radio station need its own audience data?
Your buyer now has an alternative that reports back. The Communications Authority of Kenya (CA) said in August 2026 that digital platforms are taking a growing share of budgets because they provide "targeted reach, measurable engagement and flexible campaign formats".
Radio has the audience to answer. The same report found 75% of people used radio in the past seven days, in April to June 2026. Reach alone no longer closes the deal.
The money is still large, and the argument over it has sharpened. Khusoko's report of Reelanalytics' 2025 study says tracked radio advertising fell 5% to KSh 25.9 billion in 2025 while television fell 22%, mainly as gambling curbs took effect.
The Media Council of Kenya went further in January 2026, writing that advertising and space buying "are almost collapsing" as a revenue model. The CA reads it more calmly, and both readings point the same way: advertising remains, and it goes to whoever can show what it bought.
National figures tell an agency where you stand. They rarely tell a sales manager in Kisii or Nyeri what to say to the hardware chain on Monday morning. For that you need your own market, your own questions and the reasons behind the numbers.
What three facts does a station need?
Keep three facts apart. Coverage is who could receive you.
Reach is who tuned in. Appreciation is what they thought.
Each answers a different business question. Coverage tells an engineer where to spend, reach tells a buyer what the airtime is worth, and appreciation tells a programme controller what to keep. Our page on radio coverage and signal explains why the first never proves the second.
Four sources sit around them. Recall surveys, streaming logs, on-air responses and advert logs measure different things, and they must never be added together.
How do you win and keep advertisers?
The decision is who to call this month and what to put in their hand. Start with who spends.
The CA's figures for April to June 2026 put banking and finance at KSh 1,291 million of KSh 5,073 million in radio advertising, about a quarter. Agriculture and gardening equipment rose from KSh 92 million in July to September 2025 to KSh 494 million.
Then answer the buyer's three questions. Are my customers listening, are they in my county, and are they there at hours I can afford?
Buyers rarely want a full report. On Takwimu Intelligence (TI), the Reach note for an advertiser is the short document that goes out with a rate card, and it exists so the caveats travel with the number. It covers daily radio reach, breakfast, drive time, reach by radio and not TV, the station listened to most, and how respondents compare with the population.
Reached by radio and not by television is the line that takes money from a TV-led plan. The CA found 16.4% of people used radio and no other medium in April to June 2026. Your own wave gives the figure for your county, so a planner can see exactly who the TV schedule misses.
Keep the account the same way. Send each new reach note before the renewal talk begins.
When an advertiser or regulator asks how a figure was produced, TI's service level allows 14 days, and the answer is the sampling design, the achieved sample and the weighting. Our guide on how a station proves its reach to an advertiser covers that conversation.
Which figures should a sales manager ask for first?
Four cuts pay for themselves. Main station by county shows where you are strong.
Main station by age and sex shows who you can sell. Breakfast and drive-time listeners by main station show which hours you own, and reached by radio and not by television by county shows where a TV plan leaves a gap.
Put them on one page. A sales team will use one page and ignore forty.
Turn percentages into people before the meeting. Here is an invented example on a probability sample.
In a county of 1,000,000 adults, 60% listened yesterday and 25% of those named you as their main station, which is about 150,000 adults. A buyer remembers 150,000 far better than 25%, and you should show the margin beside it.
How do you defend the rate card?
The decision is what you will refuse to discount, and why. Discounting is an old pressure in this market.
Business Daily reported in June 2009 that some radio stations had offered discounts above 60% to stay in business. The figure is old. The pressure is familiar.
Evidence changes the argument from price to value. Here is an invented example, with both stations measured in one survey on one question.
| Your station | Rival station | |
|---|---|---|
| 30-second breakfast rate | KSh 20,000 | KSh 10,000 |
| Breakfast listeners who name it as their main station | 100,000 adults | 40,000 adults |
| Cost per thousand | KSh 200 | KSh 250 |
At half your price, the rival costs more per thousand. That is the conversation you want to be having. The formula is on our page on cost per thousand.
Two rules keep the argument safe. On a probability sample every figure carries a margin of error, so show it. On a quota or convenience sample the figures describe respondents and are never ratings, so they stay out of a price negotiation.
Sell what was measured. A daypart figure counts people reached in that band, so call it that, and leave average quarter-hour to the measures built for it.
How do you fix the schedule?
The decision is what moves, what stays and what gets money next quarter. Day-after recall splits yesterday into seven bands, from before 6am to after 10pm. That shows when your market listens, which is often different from when your schedule assumes it does.
TI raises a warning when breakfast is smaller than the afternoon. In almost every radio market breakfast is the largest daypart. So either the show has a problem, or the market's day starts differently from the schedule's assumption, and the second is far more common than most programme controllers expect.
Here is an invented example. In a farming county the band before 6am reaches 18% and the 6am to 10am band reaches 24%, while the county town peaks at 7am. Cross-tabs show the early listeners are rural, so the station starts its breakfast show an hour earlier, and its agrovet sponsor gets the band that holds its customers.
A second warning matters as much. If fewer than a quarter of people can place any programme, every programme figure rests on too small a base, and promotion is a better answer than a schedule change.
Fix the name before you fix the clock. Our page on radio programme scheduling covers the rest of the method.
How do you stop losing listeners?
The decision is what to fix first, and who owns it. So ask the people who left.
The questionnaire asks whether someone switched away part-way through a programme in the last month, and what made them go the last time. Open answers are coded into the nine reasons set out on why audiences leave.
| What the wave shows | What it usually means | Do this first |
|---|---|---|
| More than 20% of leavers blame the signal | Reception, often in one or two counties | Check the transmitter and stream against the county breakdown |
| More than 30% name the adverts | Inventory earning revenue and costing reach | Put sales and programmes in one room and make the trade-off on purpose |
| More than 25% name the presenter | A person or format problem | Read it beside the share held by presenters |
| More than half left a programme this month | The show holds attention for less than its length | Read the reasons first, because adverts and signal cause this pattern too |
Treat a sharp fall in one market as an emergency. TI's service level gives it 48 hours, and the first question is transmission or schedule. Checking a transmitter costs little, while rewriting a schedule to fix a signal fault can cost a year of audience.
Look at who stays as well. A listener held by habit alone has no reason to resist a rival's new show. Held by the presenters is stronger and riskier, because that audience can walk out with a contract.
How do you back the right presenters?
The decision is who gets the contract, the slot and the premium. Presenters are a revenue line. On Berur FM's 2022 rate card a presenter mention cost KSh 40,000 against KSh 24,000 for a 30-second spot at breakfast.
One station figure cannot tell you who is earning that premium. An outlet can be losing audience to one presenter and holding it with another, and a single figure averages the two together.
Use three readings side by side: left because of the presenter, held by the presenters, and the Programme performance report, which shows the programme heard most, its appreciation and how many switched away. The programme panel adds a mark out of ten, whether someone would tell a friend, the part they like and the one thing they would change.
Mind the base. Only people who could place a programme appear in a programme figure, so a show may rest on 150 interviews, and at that size a 40% reading carries a margin of about 8 points.
One wave starts a conversation with a presenter. Two waves can support a contract decision.
Read appreciation correctly as well. It is the mark given by people who chose that programme themselves, so a mark under six out of ten for someone's favourite show is a serious signal. Our page on the appreciation index explains it.
How do you prove a change worked?
The decision is whether to keep it, reverse it or spend more. Run the next wave on the same questions, in the same population, with the same method.
Then read the difference against its margin. If each wave carries 3 points, the gap between them carries about 4, so a 3-point rise proves nothing.
A rise that clears the margin still proves less than it seems. A change earns the credit only where something was held back: one county got the new show and a similar one did not. Everywhere else the figure changed and there was no comparison, and a good report says so in those words.
TI's campaign record keeps the file. It holds what was recommended, what was decided and how it scored, so next year's manager can see what worked without relying on memory.
Know what was on air as well. TI's broadcast monitoring logs programmes by station, air time, presenter and guest, so a breakfast figure reads against the mornings you ran the phone-in. That record stays apart from the survey, because it shows what went out and never who heard it.
Is a survey wave worth the money?
A wave costs money. A sensible station spends it on decisions worth more than that.
Here is an invented test. Take ten annual contracts with a card value of KSh 600,000 each.
If evidence moves your average discount from 40% to 25%, each contract brings KSh 450,000 where it brought KSh 360,000. That is KSh 90,000 a contract and KSh 900,000 across the ten.
Set that beside the cost of a wave on our pricing page. If no decision in the next six months is worth more than the wave, wait. Most stations do well with a wave before a major change and another once it has had time to work.
How should a regional station use this?
A regional station has the easier job. It competes in fewer counties, so one wave can cover its whole market, and a county advertiser cares about exactly that county. The CA said in August 2026 that vernacular and regional broadcasters can strengthen their position by focusing on local news, culture, music, agriculture, business, governance and community issues.
Do not fight the national table. A station on one or two frequencies will never lead it, and it does not need to. It needs to show it holds a county, a language or a farming audience that a national buy reaches poorly.
Answers are weighted at county level, where KNBS census totals exist, so plan the wave around counties. If the budget is thin, spend it on one question.
Should breakfast start earlier? Is the mid-morning presenter worth renewing? One decision settled beats a thick report nobody reads.
How do you get the numbers used inside the station?
A report on a shelf earns nothing. Give each figure an owner: sales holds share and the reach note, programmes holds appreciation and the reasons for leaving, engineering holds the signal line, and finance holds average net rate.
Hold one meeting after each wave. One hour, three decisions, written down with a name and a date. Those decisions go into the campaign record, so the next wave has something to check.
Show presenters their own programme figures, with the margin of error. People argue with a rumour far longer than they argue with a number they can read.
How does this sit beside national ratings?
They do different jobs. National measures from firms such as GeoPoll and Ipsos, and the CA's own reports, give the market a shared picture of who leads. Your own wave gives depth in your own market, with reasons attached.
Keep both, and keep them apart. A figure from one source can be set beside a figure from another only when the population, question, period and sample design match. Our guide to who measures media audiences in Kenya sets out the main sources.
Can on-air replies and stream counts go in the same total?
Keep them out of it. A study can be announced on air, with replies by USSD, SMS, voice or web through the outlet's own gateway account, and a code per station shows which station brought replies. That is useful for promotion, and those replies are self-selected, so they never measure reach.
A stream log counts connections. A phone-in counts keen people.
An advert log records what went out. Add them together and you get a number that means nothing.
What do stations get wrong?
The most expensive error is selling the pool as your own. Daily radio reach belongs to the medium, and a deck that presents it as one station's audience falls apart the first time a planner reads the question.
The second is changing the schedule after one bad wave, before anyone has driven to the transmitter. Each one is avoidable.
The third is buying data once a year for the sales deck. The programme team never sees it, so the decisions that would have earned the money never get made.
What should you do on Monday?
Give each person one job.
| Who | Monday job | Why |
|---|---|---|
| Owner or managing director | Write down the three biggest decisions of the next quarter and what each is worth | It tells you whether a wave pays |
| Sales manager | List the last ten deals lost or discounted, and what the buyer asked that you could not answer | Those questions become your cross-tabs |
| Programme controller | Mark the shows and bands you would change if you had evidence | It sets what the wave must settle |
| Engineer | Check transmitter and stream records by site for the past quarter | It rules out the signal first |
Then agree one market and one date.
Read next
- Audience measurement software for Kenyan media: the TI pack page for this work
- How radio and TV stations grow their business with audience data: all the pages in this section
- How to increase radio advertising revenue
- Rate card for radio stations
- How to grow a radio audience
- 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 station's audience figures should change what it does on Monday, or they are decoration. TI measures audiences by asking people, weights answers to KNBS census totals at county level, and turns a wave into the Audience wave report, the Programme performance report and the Reach note for an advertiser. Each figure shows its margin of error. The station owns its data. Its broadcast monitoring keeps a separate record of what went out on air, so a figure can be read against what ran that week.
To see it on your own station or title, book a walk-through.