Differences in extrapolating AQH Shares

We've had a few questions lately about XTrends AQH Share extrapolations vs. those done by hand or with spreadsheets.  This page attempts to explain why those differences exist.

The short answer to why we get calls about differences is that:

1.  XTrends extrapolates the AQH Persons and the Market Level Persons (or PUR, Persons Using Radio) to calculate a new share.

2. Most extraps done by hand or with a spreadsheet are done by extrapolating the shares alone.

If you are used to just typing the shares into a spreadsheet or calculator, then you are calculating extraps differently than XTrends.  

Here are some thoughts about these two approaches:

• When doing extraps by hand, taking the second approach is the only reasonable approach because it takes time!  We'll call this the easy method.  If you have to stop and enter in all the AQH Persons and PUR, extrap new AQH Persons and PUR, and then compute the new share you have to do 3 times as many calculations.

• BUT, an AQH share is the AQH Persons divided by the PUR (times 100).  That's it, that's how it's calculated, you know that.  When a new trend or book comes out, that's how it's calculated.  You can extrapolate the shares, but you run the risk of seeing a greater variance in your extraps because you have extrapolated a percentage (the shares).

• When your next book comes out and you compare your extraps to the actual months, XTrends will be closer because it's doing the extraps as precisely as possible.  XTrends produces XTraps that are going to be a better prediction of the month.

• While you are watching shares, make sure you are watching your AQH Persons and especially your cume and TSL!  Many times a station can retain the same AQH Persons but lose share because the monthly PUR has increased.  Did you lost audience?  No.  Did you lose share?  Yes.  While that's not great, it's not totally bad news.  Go deeper and look at your cume and your TSL.  Cume is really important because that's the people who are checking out your station.  Obviously it would be great to convert all your cume to P1s and have a huge TSL, but you've got to get them on your signal first.  Watch your cume!  XTrends xtraps that too (hey, that's another benefit!).

If this makes sense, then you can move on with your day, thanks for taking the time to learn about these differences and thanks for using XTrends.

If you want to see an actual example, then read on...

Average Share for a station is calculated by dividing the AQH Persons by the Market Level Persons (or Persons Using Radio).   

In easy terms, if a station has 10,000 AQH Persons and the PUR is 100,000 then the station's Share is 10.0 (or 10 percent).  This is the calculation that XTrends uses EVERY time you see an Average Share estimate, whether its a Trend, Actual, or XTrap.

Consider this recent real life example for a large market radio station in their target demo of Persons 18-34:

Phase I:

Let's start with Station X in April of 2001:

We'll begin with the first (easy) method.  The Phase 1 trend for Spring (April) arrives and the station has a 4.0 share.  The previous 2 months shares are ACTUAL shares from the Winter book.  March is a 4.0 and February is a 6.1 (going to lose a big month!).  Do the extrap, 

• Multiply the 4.0 times 3 = 12.0

• Subtract 4.0 and 6.1.  The new share extrap is a 1.9 (ouch).

Now lets look at the XTrends XTrap.  The Phase I trend for AQH Persons is 3,300.  March and February actuals are 3,400 and 4,600 respectively.  Do the extrap,

• Multiply the 3,300 times 3 = 9,900

• Subtract 3,400 and 4,600.  The new AQH Persons extrap is 1,900 (still ouch).  This is the number reported by XTrends.

Now you have to do the PUR.  The Phase I trend is 81,800.  March and February are 85,900 and 75,900 respectively.  By the way, that's a big change in market level listening for a month, 13.1 percent more people using the radio in a 30 day period?  OK, the target demo here is Persons 18-34, and the difficulty in sampling this demo, sample size, response rates, and all that jazz are a different discussion.  Do the XTrap,

• Multiply 81,800 times 3 = 245,400

• Subtract 85,900 and 75,900.  The new PUR Xtrap is 83,600.  This is the number reported by XTrends.

Now compute the Phase I share:

• Divide 1,900 by 83,600 and multiply times 100.  The Xtrap share is 2.27272, which rounds to a 2.3.  This is the number reported by XTrends.

Big difference between 1.9 and 2.3!  And guess what, when the Spring book came out, the actual monthly numbers for April were:

• AQH Persons: 1,700

• PUR: 82,300

• AQH Share: 2.1

Sure, the easy extrap is off by a .2 and so is the correct Xtrap, and if you didn't jump out the window after the 1.9 then you'd have been around to see you really had a 2.1, but NOW....see what happens in the second month when you create extraps off of an extrap.

Phase II:

First the easy method.  The Phase II (May) trend arrives and Station X has a 3.5.  Do the extrap,

• Multiply the 3.5 times 3 = 10.5

• Subtract the April extrap (1.9) and the March Actual (4.0).  The new share extrap is 4.6.

OK, glad you didn't quit radio, or God forbid get blown out?  Let's look at the XTrends approach.

The Phase II AQH Persons trend is 3,000.  Do the Xtrap,

• Multiply 3,000 times 3 = 9,000

• Subtract the April XTrap (1,900) and the March actual (3,400).  The new AQH Persons Xtrap is 3,700.  This is the number reported by XTrends.

Now do the PUR.  The May Trend is 86,400.  Do the XTrap,

• Multiply 86,400 times 3 = 259,200

• Subtract the April XTrap (83,600) and the March Actual (85,900).  The new PUR Xtrap is 89,700.  This is the number reported by XTrends.

Now you are ready to get the May share Xtrap:

• Divide 3,700 by 89,700 and multiply times 100.  The Xtrap share is 4.12486, which rounds to a 4.1.  This is the number reported by XTrends.

So how does this match up to the actuals?  Drum roll please....

• AQH Persons: 3,700 compared to our XTrap of 3,700.  Bingo.

• PUR: 89,400 compared to our XTrap of 89,700.  Off by 300 people, that's bingo too!

• AQH Share: 4.1 compared to our XTrap of 4.1.  Bingo again!

And, to wrap up our point about extrapolating shares, look at the May share from the easy method, it was a 4.6!  It was off by a half a point!!!  XTrends was right on the money.

So what happened to Station X when the Spring book came out?  It had a 3.8 share when it added an actual month where it had a 5.5 (which is calculated on an actual AQH of 4,300 and a PUR of 78,700).  Their cume was back down to 88,200, but the TSL was huge, up almost 2 hours to 6:09. In June everything went Station X's way.  Although Cume was down, their AQH went up because of the big TSL, and the PUR went down, so it had a big Share month.

The final footnote to Station X's Spring book was the Cume and TSL story.

Well, April was just one of those months.  TSL was way off and the Cume was down significantly.  But everything rebounded nicely and the 4.1 that came in May was predicted on the money by XTrends.

Now aren't you glad that XTrends does the work for you?  Not only is it a hell of a lot easier and faster and you get every demo and daypart for every station, but when it comes to extrapolating shares, it's also a hell of a lot more accurate than the old fashioned, easy method.

If you have any questions or comments about this, please don't hesitate to contact us at 888-882-4480 or email us at support@cstoneresearch.com

Link to the Cornerstone Research Home Page