Sunday, March 22, 2009

Averages

For years I have been saying that the average Arbitron diary keeper writes 2.6 radio stations in their book during the survey week and the average BBM ballot contains an average of 2.2 station mentions, which convinced me over time after many diary and ballot reviews that pretty much every diary respondent in both the U.S, and Canada wrote down, more or less, ‘two or three’ radio stations.

I'm betting that a lot of us have programmed our radio stations based on those numbers.

Somewhere deep down I have always known that if two people are in a room, one of them weighs 300 pounds and the other’s weight is 100, there is no one in the room even close to the ‘average’ of the two, but it has taken PPM data emerging over the last few years from Montreal, Philadelphia and Houston to prove how far my (not so) "educated" guess was from the truth, and how different each market can be when you start to see data as granular as PPM permits.

BBM reports that the Montreal electronic data shows - in a city with English speakers, French speakers and bilinguals and radio stations for each of those groups - that people actually, on average, tune into four or five radio stations.

Meanwhile, in the U.S. PPM markets out so far the average meter measures listening to six or seven stations.

In Canada, panelists are permitted to stay in the sample for up to five years, while in America, ARB removes them after two.

ARB VP PPM Sales John Snyder and Michelle Barker of Arbitron’s Research staff crunched some PPM data from Philadelphia in November 2008 (each dot is one meter in the sample) and found that two people actually listened to 25 different radio stations.

During the same month in New York, five panelists cumed between 26 and 30 stations (click on the graphic to enlarge it).

Conclusion: those averages I have been living with for my entire radio career were wrong; which becomes more obvious now that we can see real usage without the bias of top of mind recall.

I’ve always believed it, but now I know for sure: when we see PPM data from the rest of Canadian major markets and more American large markets later this year, we’re going to find that radio listening in Quebec is not the same as it is in Calgary, which is different from Toronto, which is different from Vancouver, which is different from San Francisco, which is different from Atlanta.

Yes, we've all been saying it for years: every market is different.

Now, we are getting the data which proves it.

3 comments:

Chris Kennedy said...

Good blog post, Jaye. While PPM expands the number of stations listeners "hear" from the diary's "2 or 3", yes...PPM picks up that actually another 3 or 4 are listened to on at least a semi-regular cume basis. What's really interesting in a market like Montreal - split by Anglophone and Francophone sampling, there is actually a lot of cross-language tuning (English listeners tuning in French stations and vice versa).

More importantly, daily cume occasions are far more frequent with PPMs than as previously believed with the diary system, substantially increasing total daily cume reach (pummeling daily newspaper cume reach and nearly matching TV cume usage). This makes radio a far stronger advertising option for the $$$ than the diary system had the ad market believe.

Steve Casey said...

VERY Interesting blog!

But I’m not convinced that we should be too quick to throw out all your previous conclusions.

If I am reading the charts correctly, the PPM figures are over a month, not a week. Do you know if that is true?

Even if these are weekly values, we don’t have any clarity about this extra listening or its value.

1. How much of this is nearly meaningless? A study by Ball State showed that the diary captures 85% of listening volume. How many of those extra radio stations fall into the last 15%? Given the nature of the reach and frequency distribution curve, I would submit that we can dismiss that listening as misleading at best. I fear that it is more likely very dangerous because of the effect it has on the average values of just about every calculation. We should only interested in the number of stations receiving MEANINGFUL listening – not every time a meter catches the audio encoding.
2. Is this incidental listening – audio exposure – but nothing was actually “heard”? Anything not head is useless to programmers and advertisers? Diaries capture listening that people are aware of. Those are the stations that they will create relationships with.
3. Is some of this mere curiosity? To understand that, we would need to carefully study the number of times that a station might be sampled, but then not returned to for a week or more. While such behavior might have some small relevance to the sales effort, it would be of low priority to programming.

Whether we are talking about creating a real relationship with listeners, or even simply generating enough TSL so that an advertiser has a fighting chance of getting a commercial heard, our goal is average persons per quarter-hour.

Our focus has to be within our domain of influence: those who return to us regularly, because they like our programming.

The skewing of cume and TSL values that the extra stations the meter picks up audio signatures from over the course of a month are dangerous distractions from what we need to focus on. I think we must, now more than ever, make our programming decisions based on the people who put us into their top tier – the stations that they are involved enough with to write down in their diaries.

Steve said...

Great points, Jaye.

With the diary, there was never any way to measure how many stations a diary keeper listened to over a month - because, of course, the diary was kept for only a week.

It's interesting that there are a fair number of high station counts even in what appear to be the middle quartiles of the chart (midway between heavy and light listeners).