Audience measurement is research that estimates how many people, and which people, were exposed to a radio station, TV channel, publication or website during a stated period. Because nobody can observe a whole population, it works from a sample or a panel and scales the result up. Every figure it produces is an estimate, and a sound one states its base, its period and its method.
Key points:
- Audience measurement answers four questions: how many, who, when and for how long.
- It works from a sample, a panel or a census, and each gives a different kind of evidence.
- A measure that buyers and sellers agree to trade on is called a currency.
- It differs from media monitoring, which records what was broadcast or published.
- It tells you about exposure, and you need separate research to learn what the audience thought.
What does audience measurement measure?
Four things. How many people were reached, who they were, when they tuned in, and how long they stayed.
Those four answers feed every metric in the trade. Reach comes from how many, audience profile from who, dayparts from when, and share from how long. Audience metrics explained walks through each one.
Exposure is the key word. A measurement says a person heard or saw something, or says they did, and it stops there.
Why do audiences have to be estimated?
Nobody can watch everyone. Broadcast leaves no record of who received it, because an FM transmitter sends a signal in one direction, and a station can't see a single listener.
Print has the same problem in a different form, because a sold copy doesn't say who read it. Online looks easier, since a server logs each connection, yet a connection is a device and a moment, and it can't say who was holding the device.
So researchers observe a small group chosen to stand in for the rest. The quality of the whole exercise depends on how that group was chosen.
What is the difference between a sample, a panel and a census?
A sample is a fresh group asked once. A panel is a fixed group followed over time. A census counts everything that passes a given point.
| Design | How it works | Example | Strength | Weakness |
|---|---|---|---|---|
| Sample | New people each wave | Ipsos's 2022 establishment survey, face to face in 47 counties | Each wave is a fresh draw | Costs more per wave |
| Panel | The same people over time | Barb's 7,000 UK households | Shows change within people | Panel members can drift from the population |
| Census of connections | Every request to a server | Triton Digital's streaming metrics | Complete for what it covers | Sees devices, and only online |
Many modern systems combine them. Barb joins panel data to census-level online viewing, and RAJAR merges diary samples with a passive phone panel.
What is a currency in media research?
A currency is the audience measure a market agrees to trade on. Buyers and sellers may dislike a figure, but they accept it as the common unit for setting and checking prices.
Currencies usually rest on a joint industry committee, a body funded by the media owners, agencies and advertisers who use the data. Ipsos's 2022 PAMRO paper lists "non-functional or non-existent JICs" among the barriers across sub-Saharan Africa, alongside scarce funding and data piracy.
In Kenya, TIFA Research described the Kenya Audience Research Foundation in 2019 as the industry body managing audience research. In 2022 Ipsos's Kenya head expected its new survey to become "the industry currency". A currency is a matter of agreement. A seller should say whose figure the deal rests on.
What is television audience measurement?
Television audience measurement, or TAM, estimates who watched which channel and programme. In the UK and US it rests on meters on the TV sets of recruited homes, with each person logging in by button. Kenya takes a different route.
Television audience measurement explains how that works, and TV ratings in Kenya covers the Kenyan figures.
What is cross-media measurement?
Cross-media measurement follows the same people across radio, TV, print and online. Its value is the overlap: how many of a TV campaign's viewers also heard it on radio, and how many were reached by one medium alone.
Overlap needs a single source, meaning the same respondents answer for every medium. The regulator's 2019 report describes "Single Data Sourcing" across TV, radio, print and online. Its August 2026 report shows why it matters: in April to June 2026, 29.6% of people used radio, TV and online together, 16.4% used radio alone and 4.1% used online alone.
Adding separate studies can't give you that. Two studies give two totals and no overlap.
What is digital media measurement?
Digital media measurement counts activity on servers, players and platforms: sessions, impressions, plays and hours. It is fast and detailed. It describes devices.
A cumulative count doesn't even add up across periods. Triton Digital states that its cume metrics "cannot be accumulated or summed to create an aggregate". A station with 3,000 unique connections on each of seven days hasn't reached 21,000 people.
Nielsen makes a broader point about big TV datasets: they are large, and they fail to register who is in front of the screen. That is why the industry joins them to panels. Radio streaming analytics shows the same limit for a station's own stream.
What is social media measurement?
Social media measurement reports what a platform tells an account holder: followers, impressions, reach and engagement. The platform defines those terms and audits them itself.
Followers are a list of accounts. They don't show who saw a post this week, and they overlap with your broadcast audience by an unknown amount. Read what reach means in social media before you put a platform's "reach" next to a survey's.
What is audience measurement research used for?
Three groups have long paid for it. TIFA Research summed them up in 2019: advertisers targeting their messages, media owners pricing their inventory, and agencies planning and buying for clients.
Regulators use it too. The Communications Authority's 2023 report says a broadcasting competition study drew on its audience data for free-to-air and pay TV viewing and the market shares of stations.
Programme teams are the most neglected users. A schedule built on evidence needs the same figures as a rate card.
How often is an audience measured?
It varies by system, from overnight to every few years.
| System | Rhythm |
|---|---|
| Barb | Meter data retrieved each night |
| GeoPoll | Surveys every four hours, ratings the next day |
| KARF tracking survey, 2019 | Daily interviews, reported monthly |
| RAJAR | Samples counted per quarter |
| Communications Authority | Quarterly report |
| Establishment surveys in Kenya | 2015, then 2022 |
Speed and depth pull against each other. A fast tracker suits a schedule change, and a slow establishment survey tells you who the population is.
How does audience measurement differ from media monitoring?
Audience measurement asks who was exposed. Media monitoring asks what ran.
A monitoring firm records stations and logs what aired. The regulator's 2019 report describes Reelforge's method: recording servers on TV and radio antennas, with adverts detected by digital fingerprint and presenter mentions entered by analysts. That shows an advert aired, and it says nothing about who heard it. Media monitoring in Kenya sets the two side by side.
What makes an audience figure trustworthy?
Four things: a sample chosen to represent a stated population, a published method, a repeatable question, and a result someone else could check.
Ipsos's 2022 PAMRO paper sets the floor well. Research, it says, must be "scientific, verifiable" and able to answer basic questions on media access and consumption. Size isn't on that list.
What does audience measurement leave out?
It leaves out the reason. A figure can show that a breakfast show lost a third of its audience in a quarter, and it can't say whether the presenter, the adverts or the signal caused that.
Appreciation has to be asked for. So does the reason people leave. See the appreciation index and why audiences leave.
What do people get wrong about audience measurement?
They treat an estimate as a count. A figure of 2.3 million listeners sounds exact, yet it is a sample result multiplied by a population, with a margin on either side.
They also compare numbers from different systems as if they were one ruler. TIFA Research warned in 2019 that "different methodologies yield different results". The sound habit is to hold one method steady, track it over time, and report what it measures in plain words.
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
- How is radio listenership measured?
- Television audience measurement explained
- Audience metrics explained
- 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
Audience measurement earns its keep when one method runs the same way wave after wave. Takwimu Intelligence (TI) runs a cross-media day-after recall questionnaire and a programme panel, so the same respondents answer for radio, TV and news, and reach, overlap and appreciation can be read from one study. Answers are weighted to KNBS census totals at county level.
To see it on your own station or title, book a walk-through.