How SpinCounts estimates audience, and how to read the number

Every airplay report on SpinCounts now shows Estimated Audience Impressions: the total of an Estimated Audience calculated for each play. This article explains what those numbers are, how they are built, and how to read them.

What it is

Estimated Audience is a modeled estimate of how many people may have been exposed to a song when it played on a station. Like any audience estimate where a direct count is not available, it is built from public data plus clearly stated assumptions. It is not a rating, it is not a count of unique listeners, and it is not data from a ratings company.

The four ingredients

1. Where the station is. We use the population of the station’s city, from a public geographic dataset, and the station’s time zone.
2. How many of those people are listening to radio. This is an assumption: the share of a population tuned in to radio at a given moment.
3. How large the station is in its market. Also an assumption. Stations are grouped by size, from small community stations to major market leaders, and each group holds a different share of the local radio audience. Satellite channels have no home city, so they use the service’s subscriber base instead, grouped by how popular the channel is.
4. When the song played. Listening changes through the day, so each play is placed in one of five parts of the day in the station’s own local time, with daylight saving applied: morning drive, midday, afternoon drive, evening and overnight. The two drive times carry the most listening, midday a normal amount, the evening less, and overnight the least.

Multiply the four together and round to a whole number, and you have the Estimated Audience for that play.

A worked example

Imagine a station in a city of 100,000 people. Suppose, purely to show the arithmetic, that a quarter of them are listening to radio, the station holds 5% of that listening, and morning drive counts at 1.5 times an average hour:

100,000 × 0.25 × 0.05 × 1.5 = 1,875

These numbers are an illustration. They are not our settings and not any real station.

Reading the totals

When you add up many plays, the total is called Estimated Audience Impressions. One person can hear a song several times, and every exposure counts, so impressions can include repeat exposure to the same listener. That is why we never call the total listeners, reach or unique listeners.

What it is not

• It is not verified audience and not a count of unique listeners.
• It is not Nielsen or Mediabase data.
• It never affects the charts. Charts rank songs by verified spins only.

Limits we want you to know about

• No location, no estimate. If we do not know where a station is, the report shows a dash. It never shows a zero. When only some plays in a report have an estimate, the report says how many (“based on X of Y plays”).
• Very small results. For a station in a very small town the estimate for each play can round down to nothing. Reports then show a dash with the note “Too small to estimate” instead of a zero.
• Population figures come from a public dataset, and how a city’s population is defined varies from country to country.
• It is a model. Two stations of the same size in the same city get the same share, even if one is much more popular than the other.
• Online-only stations are modeled the same way as stations on the dial.

How we keep it honest

• Estimates are stored when the play happens. If we refine our assumptions later, new plays use the new values and earlier plays are not rewritten.
• The method is documented on our Methodology page, and we will say so there when it changes.
• Use Estimated Audience to compare songs, stations and weeks. Please do not quote it as a verified audience figure.

City population and time-zone data © GeoNames, licensed under CC BY 4.0.