Burning Question: Impressions

[Hefty Hanna]

I have been tracking songs on Mediabase ever since 2005. I know everything inside and out, but there was always one thing that has always caught my attention that I have never been able to quite figure out, and that is how exactly a radio impression works.

Example: 'I'll Be Lovin' U Long Time' has 14.466 million impressions this week.

Does that mean exactly 14,466,000 people listened to the song?

If so, how do they measure that?

Does that mean 14,466,000 people listened to the whole song?
Does listening to half of the song count towards a whole impression? Half an impression?
What if I'm scanning through the channel, stop at 'I Kissed A Girl', and turn it off 3 seconds later. Do I contribute towards Katy's impression count?

Thanks!

- B4L

[Nostalgia Kills]

Jul 10, 2008 12:55:28 GMT -5 Hefty Hanna said:

I have been tracking songs on Mediabase ever since 2005. I know everything inside and out, but there was always one thing that has always caught my attention that I have never been able to quite figure out, and that is how exactly a radio impression works.

Example: 'I'll Be Lovin' U Long Time' has 14.466 million impressions this week.

Does that mean exactly 14,466,000 people listened to the song?

If so, how do they measure that?

Does that mean 14,466,000 people listened to the whole song?
Does listening to half of the song count towards a whole impression? Half an impression?
What if I'm scanning through the channel, stop at 'I Kissed A Girl', and turn it off 3 seconds later. Do I contribute towards Katy's impression count?

Thanks!

- B4L

It means 14,466,000 listened to the song.

However, the methodology is very unscientific since they base the number on quarter hour audience data from Arbitron.

So scanning through the channel makes no difference, since it was already predetermined from the last ratings book how many people were listening to the station in that 15 minute period when the song came on.

[Hefty Hanna]

I still don't get that? :(

[Nostalgia Kills]

Do you get how ratings work?

Except for the PPM (Portable People Meter) in a few markets, ratings are done by paper diary. From those diaries people fill out, Arbitron calculates (among other things) how many people are listening to a certain station in a 15 minute interval. So any calculation of an audience impression would be based on the diaries from the previous ratings period.

Let's say as an example, Arbitron calculated from the previous Spring ratings period, 750,000 people listen to Z100 New York from 6am to 6:15am on Thursdays. "I Kissed A Girl" played at 6:00am today ( July 10th). Cross referencing the exact time of airplay from Mediabase and multiplying that by the 750,000 people that listen in that time period, you get 750,000 audience impressions. Suppose Arbitron calculated from the Spring ratings that 400,000 people listen to Z from 3:15pm to 3:30pm on Thursdays. Mediabase detects that Katy is played at 3:26pm. Multiply that detection by 400,000 listeners and you get 400,000 audience impressions. By adding just those 2 detection alone, Katy has 1.15 million audience impressions. And if you add all the other times the song has played on the station as well as other stations with their own unique Arbitron listener data, you get a total impression for a song in a particular format in the U.S.

The audience impression is only as good as the last ratings period (I believe BDS only uses Spring and Fall Arbitron ratings to calculate impressions, not sure about Mediabase) data.

So the process is very unscientific. And of course, the Arbitron/paper diary ratings method has been attacked by critics ever since its inception, and rightly so especially in this day and age.

I hope that helps!

[John77]

^^ the above is the reason why I always would rather term it "estimated audience" then "audience impressions" as these are truly only estimates. FYI, some stations are only surveyed every 6 months and the sample sizes can be ridiculously small in the small-medium markets.

[atlantaboy]

^Yeah I remember a bunch of posters a while back talkin about how A. I.'s don't really take into account the amount of time (per day) people listen to the station...not sure how the "quarter hours" work, but the prob. is (it seems) that stations that rotate powers 90-100x a week (and give their powers huge A. I.'s if they're big stations) tend to have audiences that spend less time per day listening to that station (that's why they rotate their powers so much...so their listeners are sure to "catch" their fav. song when it comes on) - not sure how it actually works though