After finishing up an individual analysis of manga and light novel adaptation markets, I had originally planned to toss the remainder of shows not covered in those analyses in one bin and call them “other”. It took about 5 minutes into assembling that sample to realize how incomplete that analysis would have been. There are at least 3 additional distinct categories that anime adaptations fall into: Game Adaptations, Spinoff/Franchise/Merchandise Series based around a larger product line, and True Originals.
Why games should be treated beyond simple disk sales is pretty obvious, but here’s one example. Persona 4: The Golden, released several months after the end of the anime and before the true final episode, sold a reported 248,242 copies in 2012 at an MSRP of over 7000 yen. If its anime was spending 10,000,000 yen per episode, on the low end of what’s been reported to be typical, then the total anime budget was on the order of (10000000*26)/(248242*7000)=.15, or fifteen percent of gross profits from those sales.* So Persona 4 only really needed to bump the game’s sales total up by about 10-20 percent to be worth it before even counting the 30k+ average it posted. Now, the Persona 4 anime hardly needed that money, but this does underscore that for anime series like, say, Starry Sky or Mashiroiro Symphony, being coupled to a PSP re-release of their source title is a pretty potentially big deal. I’ll be using vgchartz or something over the next several weeks to determine just how much, but it’s definitely something that needs to be looked at along with disk sales in determining how successful titles at all tiers of sales were.
The reason why spinoffs and original anime are not lumped together is a bit more nuanced. Though the distinction between the two is a tad fuzzy, the notion that truly original anime have stronger marketing pushes behind them that may prompt better disk sales is worth strong consideration. Not to mention that there’s at least some element of merchandise (however unquantifiable) being marketed beyond the disks. All of the 10k+ series in the non-Game/Manga/LN heap are true originals, so there may be more to that idea than a pipe dream. I can tell you right now that the list of originals makes for a fairly stacked chart; including things whose main goals were TV ratings (noitaminA, Phi Brain) and excluding Madoka Magica, the average original TV anime in this period sold over 8000 disks per volume.
Particular items of note regarding the category classifications that I will be using in the coming weeks:
-I classified the Marvel/Madhouse anime, Enma-kun Meramera, Appleseed XIII, Precure Series, Saint Seiya Omega, Aquarion EVOL, Blood-C, and the Mine Fujiko Lupin III series as spinoffs that, while carrying the name of the original series, are different enough to count as more than a straight continuation.
-Likewise, I classify series based on pachinko and merchandise as tangential because essentially all the makers had were a set of character designs, though they were based off things that presumably had some popularity. Ditto for series like AKB0048 and Natsuiro Kiseki based around real-life personalities.
-Black Rock Shooter is classified as a game adaptation. While the original source material is a vocaloid song, I classified it as such due to the abundance of side-characters who exist in the game continuity. Feel free to hit me up on how wrong this is.
-The Twin Angel and Tamayura TV series are lumped together as originals because they are the first direct continuations of OVA series.
-Level E, Manyuu Hikenchou and No 6, excluded by my own mistake from the manga and light novels lists, will be included in them when get to I breaking down the 5 source categories. For now, I want to focus on nailing down games, but I’ll get to them.
If there are questions you have about these classifications, I’d be happy to hear you out. Goodness knows I’m going to be spending some time figuring out the Japanese game market before I get to doing anything meaningful with any of them.
*Corrected an error made by not converting dollars to yen.
Animetics 
Regarding Persona 4 (and game sales in general):
You can’t count the game sales as pure profit, though. Even if you leave out the fixed costs of developing the game itself, there’s still the % taken by the retailers, and the cost of packaging to take account of. But yes, the extra game sales can definitely be a significant factor, just like extra manga/LN sales can be a significant factor for adaptations of them.
Regarding Fate/Zero – yes, the game (Fate/Stay Night) was the original, but Fate/Zero itself is adapted from a light novel prequel to this game.
Also, just a point in general – would it not be preferable to use a logarithmic scale for displaying sales on a graph?
The reason I tend to do order of magnitude rather than per-yen calculations is because it’s often easy to misrepresent how much we know. While we often have series sales averages to plus/minus 100 disks, the budgets of series are mostly within a black box constrained by the 10,000,000-30,000,000 yen range of episode budgets, as are the exact prices paid by each consumer for the product because of xx% off sales/etc., and crunching the 100% true profit out of that is impossible without that key bit of information. My philosophy is that it’s better to be doing an OoM/gross (I’m using that word to mean total money spent by consumers, which I hope is right) calculation that acknowledges the possibility of a spread in what actually happened than to claim to know the exact offsets of game sales to an unrealistic degree of accuracy. Game charts are better than manga charts in getting something close to total sales out with their >300 unit thresholds, so the uncertainty in this case comes most of all from the fact that I’m still trying to figure out the revenue splits and wanted to ballpark it. I would definitely factor in revenue distribution splits if I was trying to be an order more precise. For the same reason, while the costs of packaging aren’t zero, I generally take them as a negligible factor for disk-based media on the order I’m trying to calculate things.
This is more of a semantic point than anything, but I would argue that in case where the game was going to get made anyway (not always a trivial assumption, but probably valid in P4’s case), one typically doesn’t add development costs by spending more on advertising. I tend to think of anime that aren’t pure originals as paid advertisements that occasionally make back what is essentially an ad budget thorough disk sales. And since the fixed costs are fixed costs, they do matter proportionally less for each unit sold.
I’ll reclassify Fate/zero as a light novel. It makes more sense as part of the Another/Shinsekai/Kotenbu subcategory, given that. Hurts that it came out before mal news started storing charts, though. I don’t have data on how it actually sold, though based on other LNs from this period, I’m guessing the answer was “pretty well”.
Logarithmic charts might debatably be better if I was trying to display the data in a more natural form. What’s more important to me is my auxiliary goal; trying to write for an audience that might not be that well versed in statistical conventions. Maybe this is a misconception of mine, but I find that people without background don’t understand what half-log and log-log plots are, and explaining them takes away from both pithy column space and intuitivity of the articles for that audience. I want most of these posts to be self-contained things that people can grasp by looking purely at them and what I link to.
Regarding the Persona 4 stuff – I understand that your figures are meant to be rough, but if you take the high end of the range quoted in that same article, and assume the (realistic) estimate that 40% of the retail price is taken by retailers/discounts/sales tax (I forgot this last one before), then suddenly your 15% of game sales becomes 75%, as both of the sources of error pull strongly in the same direction. Of course, there’s no way to know how much each episode costed to produce, and without that figures can’t be given that are in any way reliable, and there are other sources of error as well – .for example, unless they come direct from the companies, sales figures are usually lower than actual sales. If they do come direct from the companies, they’re usually shipping figures and thus overestimates.
It’s strange, really. Messing around with statistics in this industry is completely futile as you’ll never get anything close to representative of anything of significance, yet at the same time it’s the best and most effective tool we have, and thus one we must rely on. And then there are things like merchandise which we often get no data of sales for whatsoever.
I’ve spent quite a bit of time trying to track down sales data for light novels before 2009 (or, more precisely, before November 2008, which is the start of the time covered by the 2009 yearly rankings). Indeed, this is what I was searching for when I came upon this blog. There is plenty of ranking data out there, although it is usually either based entirely on a single retailer or cover far too broad a period of time to actually give us information – and obviously don’t give us any actual figures either. There are also retailer shipment figures for individual series, though these are usually combining every volume of a series, and in cases where both sets of data are available the shipment figures are usually more than double the reported sales by oricon, so they’re useless for comparisons except between themselves. Other than that there’s just a couple of news posts around with data that looks extremely dodgy like this one by ANN: http://waa.ai/Zhz (sorry, a volume of Shakugan no Shana failing to sell 50k in the year in which the 2nd TV series aired? I just plain don’t believe that – and I’d be willing to wager every penny in my bank that at least one of the Saiunkoku Monogatari volumes released that year outsold everything on that list).
There is a way we can make statistics valid; by testing them against results (either against the future or against a sample/quantity independent from the one we originally used to draw the conclusion). Also, by acknowledging potential sources of error, we can at least start by not being too wrong and try to refine things from there. Shot charts in basketball tell you a lot, but it took some real time for the people studying them to get them developed. That is why I prefer large-scale analysis over simply comparing numbers, though.
For one example of how this can work, unit sales and licenses (for series in the 2000-4000 average range) can be said to be significant factors in anime’s financial success because we see those series get second seasons at phenomenally better clips (as opposed to series with poor unit sales and no license). Ditto for better-selling light novels generally producing better-selling anime.
That 2008 list is obviously bunk. Saiunkoku was getting to 100k in first weeks on a regular basis in mid-2009, well after what presumably was anime bump. Believing that that transition occurred well after the anime began airing is beyond me. If I had to guess, I’d say it may have come from them looking at the real list and only logging in the things they knew to be a light novel. This is my pet theory as to why volume 5 of the Kotenbu series is excluded from the myanimelist data in its release week. It’s always important to keep in mind that people repasting data from the charts are human beings.
It doesn’t help, of course, that there is no fixed definition of what a light novel is. MAL’s rankings are technically for light novels and anime/manga related books, which is why we’ve seen that manga de wakaru nanatsu no shuukan appearing regularly on the rankings recently. Plus their policy on what to include/what to leave out has changed several times over the years.
It becomes even more problematic when several of the borderline series are huge sellers. Sanbiki no Ossan has been raking in the sales consistantly for months now (in a similar manner to how toshokan sensou did at the start of last year), but yet MAL hasn’t reported its sales once. An even more extreme case is Moshidora, which came out in late 2009, but at the time wasn’t counted due to MAL being in its phase of having the strictest criteria for inclusion that it’s ever had. It appeared in the rankings a couple of years later, by which time it had become the single best selling volume that we have data from the weekly rankings for, with over 2 million sales reported by Oricon. But yet we so nearly had no data for it at all…
I guess a genre defined by the label publishers choose to put on it has some inherent issues that way. The main reason I’ve been keeping my analysis of manga and anime to 2011-2012 is that too far before that the data isn’t incredibly trustworthy (for manga, too). My kingdom for a Japanese weekly archive with the whole damn novel chart from every week.
I been researching the subject of anime adaption source recently as well. It’s a complex issue, especially since anime are often developed as a part of a mixed-media project for sales synergy. Here’s my spreadsheet for anime between 2000-2013 for reference:
https://docs.google.com/spreadsheet/pub?key=0Ajj_IfkmgvF0dDQzZVJKYmVjc0s3dTNHdVdlLVo2dnc&output=html
I referenced both wikipedia and ANN to develop this listing. I was mostly concerned with the concept of IP creation, so my categorizing scheme differs from Wikipedia’s [[Category:Anime with original screenplay]], instead uses the working definition of “anime created as one of the initial texts of an IP.” Thus:
* AKB0048 isn’t an Original Anime since it’s based on an pre-existing franchise (AKB48 idol group)
* Red Garden is an Original Anime despite the manga version precede TV premiere.
* Shigofumi: Letters from the Departed is an Original Anime, even though the LN came out two years earlier because of creator intent.
Thanks for the extra info. I’ll give that a look!
Regarding AKB/franchise series: On one hand, it’s probably fair to say that the AKB brand was big enough not to need advertising, but whether the series that are somehow spinoffs were made for disk sales or not, I find their performance as a whole interesting precisely because they seem to do worse than even print-sales boosted manga adaptations. Either they’re built primarily on alternate revenue streams/advertisement benefits even more so than any other source type, or they’re a money pit. Each case would pretty noteworthy, assuming that first-look observation holds up.
Generally speaking, I classify an anime as original if the print source predates it by less than a year (the rule of thumb I found while investigating the ton of poorly-selling sequels released t<1 year after the original). The Shigofumi case is an interesting one, and a pretty good case of why rules of thumb are usually worth a double-check.
That rule of thumb works fine based on my experience. Having gone through ~two thousand wikipedia pages in the recent weeks, Shigofumi was the only exception to the rules that I have found.
For an alternative list of anime sales by adaption source, check out:
http://www.someanithing.com/?page_id=452
Not sure where’s a good place to submit data, but here’re 2ch’s master list of anime TV ratings:
http://www.geocities.jp/animesityouritu/index.html
http://desktop2ch.tv/iga/1321776520
That second link it really useful, and seems more complete than the geocities link (thought I really wish I knew what the all-numbers posts were about). I’d been looking for more recent data figures on the noitaminA slot post-2012. The recent ratings a pretty dismal; Katanagatari broke 2 points, but everything since has been pretty typical-or-worse, with Fall 2013’s combo undercutting the Fractale and BRS ~1.7 lows with only 1.16.
Edit: Which is odd, because a later post gives a much higher figure of 1.74 for the same slot. Not sure what to make of that.
I had the first link (used it to compile the post on the most awesome timeslot nobody ever talks about): https://animetics.net/2013/11/20/noitmina-has-nothing-on-the-nihon-tv-wednesday-2450-slot/
The Puella Wiki explains it a bit:
http://wiki.puella-magi.net/TV_Ratings