Skip to content

guide

Why on-air responses do not measure reach

A radio call-in response rate needs the size of the audience as its denominator, and that is the number a station is trying to find. On-air replies count volunteers, and they can't give an audience size.

Takwimu editorial team, Editorial team9 min read

A radio call-in response rate can't be worked out by a station, because the rate needs the size of the audience as its denominator, and that is the number the station is trying to find. On-air replies count volunteers. They tell you how many people chose to respond, and no arithmetic can turn that into an audience size.

Key points:

  • Replies equal audience multiplied by the reply rate, and a station knows only the replies.
  • In a 2013 household survey in two Kenyan constituencies, about one person in five had ever taken part in an interactive show.
  • Cost and the expectation of not getting through were the main barriers, so replies lean towards those who can pay and persist.
  • A margin of error doesn't apply to self-selected replies, and a bigger count doesn't fix that.
  • On-air replies are good for engagement, ideas and testing a promotion. They belong in their own line.

What is a radio call-in response rate?

It is the share of an audience that responds to an on-air request, by phone, SMS, USSD, WhatsApp or web. The word rate implies a division: replies over listeners.

A station has the top of that fraction. It doesn't have the bottom. Without an independent audience figure, a response rate can't exist, however many replies come in.

Why can't replies give an audience size?

Because one equation has two unknowns. Replies equal audience multiplied by reply rate, and the station can see only the replies.

Take an invented breakfast show that gets 1,200 SMS replies. Here is what that could mean.

If this share of listeners replied The audience would be
1 in 50 60,000
1 in 200 240,000
1 in 1,000 1,200,000

Each row fits the facts. The reply rate is the missing number, and the only way to learn it is to know the audience already, in which case you no longer need the replies.

Who replies to radio stations in Kenya?

The strongest evidence comes from a University of Cambridge and University of Nairobi study, a random household survey in May 2013 of 760 people in Ruaraka in Nairobi and Seme in Kisumu. Its authors make no claim that the two constituencies represent the country.

Listening to interactive shows was high, at 80% to 90%. The gap is wide. Taking part wasn't common: about one person in five had ever done so, 19% in Ruaraka and 21% in Seme. That is ever, in a lifetime of listening, and any single show draws a tiny fraction of that.

Among those who listened to interactive shows in Kenya, 27.2% of men and 17.9% of women had taken part. The study found male and more educated listeners more likely to engage, and among women, those taking part tended to be younger, more educated and wealthier.

Why do volunteers differ from listeners?

Because replying costs something, and listening doesn't. In the same study, 60.3% of Kenyan listeners named cost as what stopped them, and 33.5% expected not to get through.

So the people you hear from are those who can spare the airtime and who believe they will be heard. The channel shapes them too. Calling was the main route at 71.0% of those taking part, and SMS was far more common in urban Ruaraka, at 60.3%, than in rural Seme, at 37.7%.

A station that reads its SMS inbox as its audience is reading a narrower, better-off group. Decisions built on that group will drift away from the rest.

Does a bigger response mean a bigger audience?

It can mean many other things. The topic, the prize, the presenter's push and the price of a message all move replies, with the audience unchanged.

The Kenyan study found that participation varied more by type of show than listening did. A politics phone-in and a music request hour can have the same audience and very different inboxes.

One person can also reply many times. A competition rewards exactly that, so a rise in replies might be 40 keen people. Without counting unique numbers, a station can't even say how many people replied.

What reply rate should a station expect?

A fair benchmark doesn't exist. A "normal" response rate would need a measured audience behind every show, and it would change with topic, prize, price and hour.

A station that knows its reach from a survey can work out its own. Take an invented case: a survey estimates 240,000 weekly listeners, a campaign draws 1,200 replies, and the rate is 0.5%.

That rate is true of that prompt on that show. It doesn't transfer to the next campaign, so a station can't run the sum backwards to find a later audience.

What do survey researchers call this?

They have a name for it, and it isn't flattering. The Encyclopedia of Survey Research Methods calls call-in polls "an unscientific attempt to measure public preferences" and says they "should not be referred to as polls".

The same reference describes the self-selected listener opinion poll, or SLOP, as an unscientific poll conducted by broadcast media. These aren't insults. They describe how the respondents got there: nobody chose them, so nobody can say whom they stand for.

Does a margin of error apply to on-air replies?

It can't. AAPOR says the margin of sampling error applies only to probability-based surveys, where participants have a known and non-zero chance of being included, and that it does not apply to opt-in polls.

A volunteer's chance of being included is unknown. So 10,000 replies carry no more scientific weight on audience size than 100, and quoting a percentage from them with a margin would mislead.

Size is the trap here. A big inbox feels like proof, and scale doesn't cure self-selection.

How does the reply channel change who responds?

Each channel has a different door. USSD and SMS work on a basic phone, while WhatsApp and web need a smartphone and data.

Ipsos put basic phone ownership at 50% and smartphone ownership at 51% in 2022, so the choice of channel decides who can reply at all. Voice suits people who would rather speak than type, and the Kenyan study found calling the most common route, especially in rural areas.

Offering several channels widens the door. It doesn't turn volunteers into a sample.

Can a station code show which station is working?

A code tells you which station brought a reply. That has real use in a multi-station campaign, because you can see which station's audience moved to respond.

It remains a count of responses. Station A bringing twice the replies of Station B doesn't show it has twice the audience, because presenters read the message differently and audiences differ in their willingness to text. A code measures response by station, which is a useful thing with its own name.

What are on-air responses good for?

A good deal, as long as you call them what they are.

  • Showing that a segment moves people to act.
  • Collecting story ideas, complaints and voices for programmes.
  • Comparing two promotions on the same show.
  • Finding people willing to join a listener panel or focus group.
  • Giving a sponsor evidence of response, labelled as response.

For the deeper reasons people stay or go, use focus groups and proper radio listener survey questions. For sponsors, see sponsorship evaluation.

How should a station report replies?

Name the channel, the count and the period. Stop there. Here is an invented example: "The Monday phone-in received 1,200 SMS from 870 unique numbers in October."

That is exact, and useful. What a station shouldn't write is "1,200 replies, showing an audience of hundreds of thousands". It should also keep percentages from replies out of any claim about listeners in general.

Practitioners tend to fold replies into a sales deck because they are the one number the station owns. Advertisers are getting wiser, and one inflated claim taints every other figure in the deck. How a station proves its reach shows what to offer them.

Can on-air replies sit beside a proper survey?

They can, in separate lines. A survey gives reach and actual radio listenership with a margin of error. Replies give colour, ideas and evidence of response.

Announcing a study on air is a fine way to invite people to give a view, as long as the result is reported as the view of those who replied. Stations covering sensitive topics should take special care, since a loud inbox can look like public opinion when it isn't. Media and election violence shows why that matters in a tense season.

Where TI fits

Replies and reach belong in separate columns. On Takwimu Intelligence (TI), 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 showing which station brought them. Those replies are self-selected, so TI never reports them as reach; reach comes from a weighted recall survey with its margin of error shown.

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

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

Sources

  1. 1

    Interactive media audiences in Africa: a comparison of four constituencies in Kenya and Zambia · Lopes, Mudhai and others, PiMA Working Paper 4, University of Cambridge, April 2015Random household survey, May 2013: 760 respondents in Ruaraka and Seme, participation, channels and barriers. Opened 5 November 2026.

  2. 2

    Call-in polls, Encyclopedia of Survey Research Methods · SAGEDefinition and data quality. Opened 5 November 2026.

  3. 3

    Self-selected listener opinion poll (SLOP), Encyclopedia of Survey Research Methods · SAGEDefinition. Opened 5 November 2026.

  4. 4

    Margin of Sampling Error/Credibility Interval · American Association for Public Opinion ResearchWhen a margin of sampling error applies. Opened 5 November 2026.

  5. 5

    Democracy in media research: power of the markets (PAMRO 2022) · Ipsos, Barnabas Muya, 7 November 2022The Kenyan tracker design, barriers to measurement in sub-Saharan Africa, and joint industry committees. Opened 5 November 2026.

More from the library

  • Can a radio station know its actual listenership?

    Actual radio listenership can only be estimated, because FM leaves no record of who received it. A fair survey gives a figure with a margin of error, and that range is what a station should quote.

  • What is audience measurement?

    Audience measurement estimates how many people, and which people, were exposed to a station, channel, title or site in a stated period. It works from samples and panels, so every figure is an estimate.

  • How is radio listenership measured?

    Radio listenership is measured by recall surveys, diaries or meters. In Kenya the regular published figures come from phone and mobile recall, and the rule for what counts as listening changes the answer.