Primer: The “Overall” estimate

I’ve written about this before, but as I can see it will be relevant in the estimates for 2008, here’s a quick overview of how the Overall Comics, Trade Paperbacks, and Magazine Dollar Sales statistic used here is calculated.

Unlike the data work most commonly done with the Diamond Top Sellers charts, this measure does not involve order index numbers and individual titles, but rather the Market Share chart that Diamond releases every month. Diamond reports the Final Unit and Dollar shares of its Top 20 Publishers of comics, trade paperbacks, and magazines. This is commonly seen here in the left sidebar column of any given month; last month’s example is here. Scroll to the middle, where it says “Market Shares.”

These figures presumably cover everything Diamond sold in the three categories — and they do definitely include magazines, since Wizard and Eaglemoss are ranked. One would suspect that sales of Mad would be attributed to DC‘s share — and so on. These shares are different from the shares you’d get from just looking at the Top 300 Comics Dollars — or even the Top 300 Comics plus the Top 300 Trades dollars — because (a) magazines are included, and (b) so are all the comics and trades beneath 300th place.

You can see the difference in the shares the samples here, again, from November. Let’s look at just IDW‘s and Fantagraphics‘ dollar market shares:

TOP 300 COMICS DOLLAR SHARES:
IDW: 3.31%
Fantagraphics: 0.05%

TOP 300 COMICS + TOP 300 TRADE DOLLAR SHARES:
IDW: 3.04%
Fantagraphics: 0.18%

FINAL DOLLAR SHARES FOR COMICS, TRADES, AND MAGAZINES:
IDW: 2.96%
Fantagraphics: 0.49%

The Top 300 Comics Dollar Share finds IDW way ahead, because IDW ranked 20 comics to Fanta’s 1. When the Top 300 Comics are added, IDW’s share slips relative to other publishers, and Fantagraphics’ share more than triples — because Fantagraphics has more trades. But when you get to the widest category, which includes everything on the backlist, Fantagraphics’ share nearly triples again — such is the length of its backlist!

OK, so the shares are different, and the Comics, Trades, and Magazines total represents the widest circle of items that Diamond reports. Now, at the same time that Diamond made its shift to reporting Final Orders only in its Top 300 list, in 2003, I began a new calculation using the final monthly order file that Diamond provides for publishers. This source material includes every single item shipped in the month, right down to single copies of backlist trades. It also accounts for a tiny number items subtracted from the system, in the case of refunded orders.

If you’ve got that data for a publisher, and the publisher made the Top 20, you can derive the size of the whole pie. Imagine a publisher with a dollar share of 20%, and with known orders of $8 million:

Publisher Actual Dollar Share: 20%
Publisher Actual Orders: $8 million
Overall Estimated Dollars: $40 million (or $8,000,000/20%, or $8,000,000 x 5)

That’s all there is to it — but before first using the calculation years ago, I did some checking. For example, the projected Actual Dollars for every publisher in the Top 20 Market Share list must be higher than the figure observed for the publisher by adding the dollar values of its items in the Top Comics and Top Trades lists. And it is. The major publishers, most of whose revenue is captured in the Top 300s, are close to the average number, but always below it — and then you get the Fantagraphics, where the “Overall” projection for the publisher is much higher, because of all the backlist items not captured in the Top 300s.

So the measure holds up there, but is it measuring what we think it’s measuring? Over the years I’ve been calculating it, there have been a number of data points from Diamond and other publishers, some corroborating the estimates, others placing the estimate a bit high.

In a press release about Free Comic Book Day a couple of years ago, Diamond announced that “comic shops sold over 100 million comic books and graphic novels in 2006 worth approximately $385 million at retail — an increase of approximately 18% over 2005.” That 2006 figure tracks well with the $396 million I had estimated for Diamond’s overall comic book, trade paperback, and magazine sales for 2006, seen here — and the magazine portion would close the remaining gap. So that one’s pretty close indeed.

On the other hand, following the “18% increase” number suggests 2005’s comics and trades were at $326 million; the “overall” estimate was $352 million. That’s more double the gap. Did magazines sell more dollars in 2005 than 2006? Probably. Enough to cover the gap? Probably not. It’s looking like 2008, where the Overall estimate looks for the year to be up by maybe 1%, is running a few points ahead of where recent statements from Diamond would put it. Still just a few points, but if true, a few points high.

There are a number of mathematical possibilities on both sides: the Diamond statements might not be reflecting the exact category of Final Comics, Trades, and Magazines the Overall estimate looks at. They might just be talking comics — or they might be talking about everything Diamond sells. In December 1998, back when Diamond reported its category market shares, 23.5% of its sales came from non-comics products. That’s certainly changed (and, wow! magazines were outselling trade paperbacks back then), but it is possible for multiple things to be true: comics down, trades up, comics and trades flat, all products down.

But it’s also possible that the yearly Overall figure — which is actually an aggregate of 12 monthly estimates — is not measuring what we think it’s measuring. I think we can be reasonably confident that the Market Share table is an accurate statement of the market shares in the category for a specific period; we can also be reasonably confident that the Final Order files I’m talking about also reflect sales for a specific period. Are the periods exactly the same? I think so — after all, the Order Index Numbers in the Top 300 square up with the Final Order files perfectly. But things get a little more complicated in dealing with the trades, as I found in November; those figures do not square up as perfectly. This suggests that the snapshots — the publisher’s data and the Market Share table — might be slightly off with relation to each other when it comes to the products referred to in each, introducing a margin of error. But again, an argument in favor of the estimates: Using the monthly market shares and the publisher data as I have actually predicts, almost exactly, the annual market shares Diamond published. What that means is that if the size of the projected overall pie is off, it must be off consistently, by the same factor, every single month.

I realize this has gotten pretty detailed — more so than intended for a primer — but to those of you who’ve stuck it out, this statistic is something I intend to continue looking at to see if refinement is possible. As stated, the December 2008 numbers, nearly done, look to raise the issue, so this seemed a good time to get under the hood and have a look around.