A radio audience for advertisers is proved with two separate kinds of evidence: proof that the advert aired, and proof that people heard the station when it did. A transmission log shows the first, and a properly weighted audience survey or a meter panel shows the second. No single document shows both.
Key points:
- Airing proof comes from the station's log, an off-air recording or an independent advert monitor.
- Audience proof comes from asking a sample of people, or from a meter panel where one exists.
- A log says nothing about listeners, and a reach figure says nothing about whether a spot played.
- Coverage maps, streaming logs and on-air responses each measure something narrower.
- These measures are never added together.
What does an advertiser want to know?
Three things, in order: did my advert run as booked, who was listening when it ran, and did they notice it?
Stations often answer the first with confidence, then let that answer stand in for the other two. An invoice with a transmission certificate attached feels like proof of results. It proves the station did its part.
That is worth having. It is still one fact of three.
What proves an advert aired?
Three records do. The first is the station's own log, showing what played and when, and the second is an off-air recording of the spot in its break. The third is a report from an independent advert monitor, which logs adverts across stations.
Kenya has that third kind of data. Reelanalytics' first-half 2026 report, as Khusoko reported, counts advertising insertions by station and names the radio station with the most. The CA's April to June 2026 report charts number of spots against audience size for television.
A station should offer its log freely, and an advertiser should spot-check it against an independent source. Ask for time stamps and for any spot that was missed or moved.
What proves people heard it?
Only evidence about people. Two broad methods exist.
The first asks them. A day-after recall survey asks a sample which stations they heard yesterday and when, and GeoPoll describes a mobile panel surveyed through the day, with ratings available next day. Our page on day-after recall sets out how the question works.
The second measures them. A portable people meter records exposure among a panel of carriers, and our page on people meters explains the method. Both approaches report on a sample and carry a margin of error.
Why do you need both?
Because neither can do the other's job. The clearest example is a US study by Arbitron, Media Monitors and Coleman Insights in 2011. To learn what happened during commercial breaks, the authors had to join radio ratings data with commercial occurrence data.
Ratings alone could not say when the adverts ran. Occurrence logs alone could not say who was there. Put side by side, by the minute, they answered the question.
Kenyan advertisers can apply the same logic with a coarser tool. Take the log for when the spot ran, and the survey for who listens in that daypart.
Which evidence shows which?
Here is the full list.
| Evidence | Shows the advert aired? | Shows people heard it? | What it really tells you |
|---|---|---|---|
| Station log | Yes, on the station's word | No | The spot was played out |
| Off-air recording | Yes, for that moment | No | What went out and when |
| Independent advert monitor | Yes, from a third party | No | Insertions by station and time |
| Coverage map | No | No | Who could receive the signal |
| Streaming server log | No | For stream listeners only | Sessions on the stream |
| On-air responses | Only if they cite the advert | Only those who replied | A self-selected group took action |
| Day-after recall survey | No | Station and daypart, as remembered | Reach, by asking people |
| Advert recall question | Indirectly | Those who remember it | The advert stayed in memory |
| People meter panel | No | Exposure among the panel | Reach, measured passively |
Read the table by column. A strong line for airing is an empty line for hearing.
Why doesn't coverage prove reach?
Coverage counts who could tune in. Our page on radio coverage and signal shows where reach stops. A transmitter's population says nothing about the share who chose that station today, and actual radio listenership is a separate question.
Quoting the population under a mast as an audience is the commonest overclaim in a media pack. Buyers discount it, and then they discount everything else on the page.
Why don't on-air responses prove reach?
Because the people who reply choose themselves. A presenter's call for an SMS brings in the most engaged listeners, and a station with a loud fan base can beat a bigger, quieter one. Our page on on-air response sets this out.
Responses do show something useful. They show a creative or an offer moved someone to act. Report them under that heading.
Do streaming logs count?
They count streams. A server can tell you sessions, durations and devices, as our page on streaming analytics explains. That says nothing about the larger audience on FM.
The CA reports that radio stays closely tied to traditional devices, and that 87 per cent of listening is at home. So a streaming total is a small slice of a station's audience, and it cannot be added to a survey reach figure, since the two overlap by an unknown amount.
Does hearing the station mean hearing the advert?
Only partly. Michelon and colleagues warned in the Journal of Advertising Research in 2020 that radio ratings may overestimate real advertising exposure. Their study of 800 Canadian panel members found mechanical avoidance, such as switching or turning off, at about 3 per cent.
That covers hands on the dial. It leaves out minds elsewhere: the same paper cites a television meta-analysis finding 29 per cent of viewers give an advert full attention. For radio, it says, attention loss turns the set into background noise.
So there are layers. An advert can air, a station can have listeners, and the listener can still miss it.
Good reporting names the layer it reaches. Our page on advertising clutter on radio explains what moves that last layer.
What should a reach note contain?
Keep it to one page, and put these on it:
- the population, area and dates surveyed
- how many people were asked, and how
- daily and weekly reach by daypart, with margins of error
- who measured it
- the schedule and the source of the airing record
- what the note does not show
Here is an invented line: "Daily reach among adults in one county: 34 per cent, plus or minus 3.1 points, from 900 interviews." The margin comes from the usual formula at 95 per cent confidence, before any effect of weighting.
Then the arithmetic. If that county has 500,000 adults, 34 per cent is 170,000 people, with a range of about 154,500 to 185,500. Quote the range, and the buyer stops arguing about the point.
What do stations get wrong when proving reach?
They send one number with no source. A figure with no date, method or margin is a claim, and buyers treat it as one.
They also blend sources. A slide that adds recall reach, stream sessions, Facebook followers and SMS replies into a "total audience" tells a buyer the station cannot separate them. Each counts different people in a different way, and many of those people are counted twice.
And they quote rankings as if there were one. Sources differ. Name the source and the date.
What should an advertiser ask for?
Ask these before you book:
- What record will show my advert ran, and when will I get it?
- Who measured your audience, when, and how many people were asked?
- What is your reach in my target counties and dayparts?
- What is the margin of error on that figure?
- What happens if a spot is missed?
Our guide to radio advertising rates in Kenya shows how to turn the answers into a cost per thousand.
Who should own this evidence?
The station should. Audience evidence is a sales asset, and an outlet that owns its data can show it to every buyer. Takwimu Intelligence (TI) holds that a station needs three separate facts about itself: coverage, reach and appreciation, with airing proof kept in its own file beside them.
An advertiser gains too. A clean separation means a dispute about a missed spot never turns into a dispute about the audience.
Read next
- Audience measurement software for Kenyan media: the TI pack page for this work
- Media advertising in Kenya: planning, buying and proving reach: all the pages in this section
- A reach note for an advertiser on TI
- Why on-air responses do not measure reach
- Day-after recall
- 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
TI handles the audience half of the proof, by asking people. Its Reach note for an advertiser states daily, breakfast and drive-time reach and the station people listened to most yesterday, weighted to KNBS census totals at county level with each margin of error shown, and the airing record stays with the station's own log.
To see it on your own station or title, book a walk-through.