Skip to content

guide

How to run an audience survey in Kenya

Eight steps from the decision a station has to make to a report an advertiser can check, with the error that costs most at each step.

Takwimu editorial team, Editorial team9 min read

An audience survey in Kenya is a sample of people, chosen to stand for a population, who are asked what they listened to, watched or read and what they thought of it. You run one in eight steps: fix the decision, define the population, size the sample, choose the mode, write the questions, run the fieldwork, weight the answers and report them with their margin of error.

Key points:

  • A survey gives two of the three facts a station needs: who tuned in and what they thought. Coverage comes from signal data.
  • Sample the whole population of your area. A list of your own listeners cannot give you reach or share.
  • Ask about yesterday, in the order people remember a day, and ask opinions after behaviour.
  • Weight to KNBS census totals at county level and print a margin of error beside every figure.
  • Keep streaming logs, on-air replies and advert logs in their own columns. Adding them to survey figures counts the same person twice.

What is an audience survey?

An audience survey is sample research about media use. It questions a group chosen to stand for a population, then projects the answers to that population.

Kenyans search for it under many names: listenership survey, viewership survey, readership survey, radio audience survey. The method does not change. Only the medium in the questions does.

Its strength is the denominator. A survey also hears from people who did not tune in, so it can state a percentage of a population.

What three facts does a station need?

A station needs three separate facts: who could receive it, who tuned in, and what they thought. These are coverage, reach and appreciation.

Coverage comes from your transmitters and radio coverage and signal data. Reach and appreciation come from asking people. Every source below answers one narrow question.

Source What it measures What it cannot tell you
Recall survey People who say they tuned in, as a share of a population Listening they forgot
Streaming log Connections to your server Who they were, or anyone on FM
On-air response People who chose to reply Anyone who listened and stayed silent
Advert log Spots that ran Who heard them

The numbers in that table never add up. One listener can appear in all four, and most of your audience appears in only one. Our page on on-air responses and reach shows why a full phone line says little about size.

What decision should the survey change?

Write the decision down first. Move the breakfast presenter or keep them. Open a sales desk in Kisumu or stay in Nairobi. Defend a rate to an agency.

The decision sets everything after it. A rate defence needs a population an agency recognises, and a schedule change needs daypart detail and reasons for leaving.

Surveys that begin with "let us learn about our audience" tend to end in a report nobody uses. Name the person who will act on the result and the date they will act.

Who is your population?

State the population in one sentence, with a place and an age. "Adults aged 18 and over in Kisumu, Siaya and Homa Bay" is a population. "Our listeners" is a guess.

Include people who never tune in to you. If you interview only fans, you learn their profile, and your share comes out close to 100%.

Age needs care. The tracking survey supplied to the Communications Authority in 2019 covered people aged 15 and over, yet Article 260 of the Constitution treats anyone under 18 as a child, and section 33 of the Data Protection Act requires a parent's or guardian's consent to process a child's data. Start at 18 unless the decision needs younger people and you have a consent route.

How many interviews do you need?

Set the number from the smallest group you will report. At a result near 50% and 95% confidence, 400 interviews give about ±4.9 points and 1,000 give about ±3.1 points, before weighting.

A station that wants figures for three counties needs a usable sample in each. The full arithmetic is in sample size for a media survey.

Phone or face-to-face?

Choose by who you cannot afford to miss. KNBS and the Communications Authority found that 53.7% of people aged 3 and above owned a mobile phone in 2023/24, at 64.6% in urban areas and 48.6% in rural areas.

So a phone-only design starts with a gap, and the gap is wider in the countryside. Rural stations feel it most. A sound pattern is a face-to-face benchmark with phone waves after it. Compare the two in phone survey vs face to face.

What should the questionnaire ask?

Ask about one concrete day. Ask behaviour first, opinions second and demographics last.

Day after recall is the sensible default, since yesterday is still fresh and the interview stays short. A radio diary survey gives a whole week per person at a much higher effort.

Our guides to audience survey questions and the radio listening survey carry wording you can adapt. Pay attention to prompted recall, because a list of names changes the answer, and define your dayparts before fieldwork starts.

How do you stop fieldwork bending the answers?

Run fieldwork across seven full days, so weekends count. Pilot the questionnaire first. Train every interviewer on the same script, and call back a sample of respondents to confirm the interviews took place.

Screen out people in the trade. Nielsen says no one who works in radio or media is included in its sample, and the Media Rating Council's minimum standards ask services to keep station staff and their households out.

Keep your promotions normal while the survey runs. The same standards require a service to publish it when a station uses unusual promotion to distort ratings or urges the public to cooperate with a survey.

How do you weight the answers?

Compare your sample with census totals, then correct the difference. KNBS publishes county totals by sex from the 2019 census, which counted 47,564,296 people.

Weighting repairs imbalance among the people you reached. It cannot find the people you missed. The steps and the arithmetic are in how to weight a survey.

How do you report a wave?

State the method at the top: who was asked, how, when and how many. Say who paid. Show the base and margin of error beside each figure, and say when a change between waves is within the margin.

The Media Rating Council's standards ask for all of this in each report, along with the errors a survey cannot remove. Our page on the audience research report gives a full structure. Handle the people behind the numbers under data protection for surveys in Kenya.

How often should you run one?

Run the same survey at a steady rhythm. National reporting here is quarterly: the Communications Authority's latest report covers Q4 of 2025/26, and RAJAR in the UK builds a sample of over 22,000 adults per quarter.

Most stations do well with two to four waves a year. What matters is sameness: same questions, same mode, same blocks, same population. Change one and you lose your trend.

Between waves, use focus groups for media to find out why a number moved.

Where do Kenya's published figures fit?

National figures set the context. The Communications Authority put seven-day access in Q4 of 2025/26 at 75% for radio, with television at 73% to 74% across the year, and daily use at 52% for television and 49% for radio.

Those two lines show why you must name the period. The same medium falls from 75% to 49% when the question moves from a week to a day. A station's own wave then answers what the national picture cannot: its share, in its own area, by time of day. The wider field is covered in audience measurement in Kenya.

Can a station run its own survey?

A station can run its own wave, provided the method is open. Advertisers will test a self-run number harder, and they should.

Publish the questionnaire, the dates, the sample and the weighting. Let an agency see the base sizes. A modest survey with its method in view is worth more to a buyer than a large one nobody can check, which is the argument in how a station proves its reach.

Where TI fits

An audience survey earns trust when its method is in the open and its three facts stay apart. Takwimu Intelligence (TI) carries a 43-item cross-media day-after recall questionnaire and a 7-item programme panel, weights answers to KNBS census totals at the level where they exist, in practice the county, and shows a margin of error with each figure. An outlet owns its own data, and respondents' identities stay apart from analysis.

To see it on your own station or title, book a walk-through.

Takwimu Intelligence (TI) is audience measurement software for Kenyan media.

In this section

  1. Day-after recall: how it works and why stations use it

    Asking about yesterday keeps memory short and cost low. It also gives a number that behaves differently from a diary or a meter.

  2. Radio diary surveys: how they work and where they fall short

    A diary gives a whole week per person. It also asks more of people than most Kenyan studies can reasonably expect.

  3. Audience survey questions: what to ask and how

    The order, wording and answer options decide what a survey can prove. Most of the damage is done before the first interview.

  4. Radio listener survey questions, with examples

    Twenty-five questions you can adapt, from whether someone listened yesterday to why they switched away.

  5. Sample size for an audience survey

    How many interviews you need, how much weighting and clustering cost, and how big a change between waves is real.

  6. Phone or face-to-face: choosing how to reach your audience

    The mode decides who you can reach and how they answer. Change it between waves and it can pass for a change in audience.

  7. Weighting an audience survey to the population

    Weighting corrects who answered. It costs precision, it cannot find who you missed, and every step should be written down.

  8. Dayparts in radio and TV: how to define and measure them

    A daypart is a promise about when people are available. Define it around their day, keep it fixed, and never add dayparts together.

  9. Prompted and open recall: why the question decides the answer

    A list of names reminds people of what they would never have said. The two questions measure different things, and both are useful.

  10. How to report an audience wave so it can be trusted

    A report earns trust by showing how it could be wrong. What to include, how to show change, and what to keep off the page.

  11. Data protection for audience research in Kenya

    What the Data Protection Act asks of a station or research firm running a survey, from the first phone number to the final deletion.

  12. Focus groups for radio and TV programmes

    A focus group explains why an audience reacts as it does. It began in radio research, and it cannot tell you how many people agree.

Sources

  1. 1

    Audience Measurement and Broadcasting Industry Trends Report, Q1 2019-2020 · Communications Authority of Kenya, covering July to September 2019The KARF tracking survey method, and viewing in prime time against other blocks. Opened 5 November 2026.

  2. 2

    Kenya: National Laws · Child Rights International NetworkArticle 260 of the Constitution of Kenya on who is a child. Opened 5 November 2026.

  3. 3

    The Data Protection Act, No. 24 of 2019 · Parliament of Kenya, 2019; copy held by CIPIT, Strathmore UniversitySections 2, 18, 25, 26, 28 to 33, 37, 39, 44, 51, 53, 63 and 65, read in the Act as assented. Opened 5 November 2026.

  4. 4

    Analytical Report on ICT, based on the 2023/24 Kenya Housing Survey · Communications Authority of Kenya and KNBS, 2025Mobile phone ownership and use among people aged 3 and above, and household ownership. Opened 5 November 2026.

  5. 5

    Nielsen Audio Measurement: methodology · NielsenDiaries and meters by market size, sample sizes, weighting, and metric definitions. Opened 5 November 2026.

  6. 6

    Minimum Standards for Media Rating Research · Media Rating Council, December 2011Ethical, operational and disclosure standards for rating services. Opened 5 November 2026.

  7. 7

    2019 Kenya Population and Housing Census, Volume I: Population by County and Sub-County · Kenya National Bureau of Statistics, November 2019National and county totals by sex. Opened 5 November 2026.

  8. 8

    Audience Measurement and Industry Trends Report, August 2026 · Communications Authority of Kenya, August 2026Seven-day access, daily use, place of consumption, radio genres and languages, and the evening advertising window for Q4 2025/26. Opened 5 November 2026.

  9. 9

    The RAJAR Methodology · RAJARThe diary sample, passive panel, diary boost panel and calibration. Opened 5 November 2026.

More from the library