One bit of seemingly ubiquitous conventional wisdom is that makers of anime often face a choice between making works that sell and works with integrity. However, one thing I’ve learned over the years is that it’s usually worth taking the time to test conventional wisdom against actual numbers, because it can be wrong fairly often. So I took a look at the performance of Ecchi anime relative to the rest of the market over the past 8 years. Sure enough, the picture is a bit more complicated than “otaku only buy boobs”.
In this analysis, I adapted a fairly simple method of classifying Ecchi and Non-Ecchi. I took the shows from 2005-2012 that I have unit sales data for, and classified the shows tagged as “Ecchi” on myanimelist as the Ecchi sample. This sample can be found here (the Non-Ecchi sample is on the second sheet of the same google doc), or just read on for the analysis. For perspective, keep in mind that a Unit Sales total of 3000 per volume is typically considered break-even.
So, out of 761 total shows in the overall 2005-2012 sample, 102 are classified as Ecchi. These shows sell an average of 4374 units per volume, comfortably profitable on average. The Non-Ecchi sample contains 659 shows, which average a very similarly comfortable figure of 4533 units per volume.
However, 2005-2012 is largely an arbitrary period, selected because it matches the bounds of my original sample. There’s an important divide there that I haven’t accounted for yet. Namely, the fact that Blu-Rays came onto the scene in late 2006, and, after a short while, seriously altered the dynamics of the anime market by raising the price producers could charge without significantly altering production costs. In 2007, Blu-Ray sales made up less than 20% of the market. In 2008, they made up a little less than half. By 2009, they accounted for between 70 and 80 percent of units sold for top-selling shows. For the purposes of this article, I’m saying the transition to the Blu-Ray era is more-or-less complete in 2009, and the pre-Blu-Ray era is the years up to and including 2008. This happens to coincide fairly well with a huge increase in average unit sales; after Winter 2009, every season averaged more than 4000 per show, a feat that only 4 seasons from 2005-2008 managed.
The really interesting result comes in when you compare Ecchi sales over the two eras. In the pre-Blu-Ray era, 37 Ecchi shows averaged sales of 4076, while 327 Non-Ecchi shows averaged 3367 in sales, a much borderline figure. In the post-Blu-Ray era, average Ecchi sales are relatively unchanged, with 65 shows putting up 4543 in average unit sales. Non-Ecchi shows, though, take a huge jump; 302 Non-Ecchi shows log an impressive 5826 in average sales.
Now what does all this mean? It points to a particular line of reasoning; Ecchi anime was significantly outperforming Non-Ecchi anime in the near past, so they’re making more of it now. Meanwhile, Non-Ecchi shows raised their standards of production and got a big boost in average sales out of it. I don’t want to get big-headed and predict what the future means, but what the data says is that, while there is more Ecchi anime being made, its sales figures are growing a lot more slowly than the industry in general.
I often see people people point to individual cases, such as the lackluster sales of Shinsekai Yori, as indications that it’s near-impossible to make an experimental anime and have it be profitable. The claim itself is easily refutable, just dredge up the consistent megahit numbers of Natsume Yuujinchou (over 10,000 for the first 3 seasons), or the respectable successes of Mawaru Penguindrum (5243) and Tatami Galaxy (5665). It’s dangerous to point to any one particular example as evidence of a system broken beyond repair; individual shows fall through the cracks all the time for any number of reasons, especially when the field is as competitive as it is for seasonal anime. Saying that shallow Ecchi is somehow the only kind of anime that sells is to ignore the fact that a lot of those shows fall flat with Japanese audiences; for every Queen’s Blade that makes mad bank, there’s a few Chu-Bras that fail to recoup production costs.
Animetics 
Pingback: 3 Major Anime Industry Sea Changes Explained By Their Effect on TV Anime (Part 3: Blu-Ray Boosting) | Animetics
Pingback: Fun With Numbers: The Anime Industry Is Demonstrably Healthy (And Not Dying) | Animetics
Pingback: Lists Are Fun to Make: Questions I’d Like to Attack in 2014 | Animetics
Pingback: Fun With Numbers: Do Elements of Ecchi Content/Fanservice Provide a Boost for Blockbuster Shows? | Animetics
Did this sample took a long time to put together? This kind of analysis seems like a productive and straightforward way to answer all sorts of “X kind of anime always sells” type of question.
It’s hard to say exactly how much time it takes, but the mechanics of the process are something like:
-Open 2 windows, one with the genre query and one with the list of shows in the time period.
-Go through the list and mark series that show up in the query.
-Sort out the list by the marks, then get averages.
The second step is the only one that’s really grindy. For a 700 show sample like this one, the fastest you can do it is like 20-30 seconds per show that needs to be marked (so 102*.5~50 minutes). It takes longer the worse prepared you are, but it can be fairly streamlined if you’re just looking at one y/n variable.
The problems I encountered when trying to do more of these is that any database would need to be complete enough in a particular genre. I don’t have a database I trust for that sort of thing, generally. Ann is sometimes incomplete, which is more dangerous the smaller a theme category is. Anidb is likewise incomplete and also has a tag system with the maturity of a 14-year-old, and mal a) doesn’t track detailed stuff and b) stopped allowing general searches a short while ago. You can still search their database, but you have to specify text now, which sucks a whole lot – can’t use it to quick-check the shows in a given year or genre without a lot more finagling.
Convenience is a luxury that comes when queries are popular enough to get attention. It’d be really nice to be able to set up complex queries like that; “Find all shows where voice actresses also sing the show’s opening theme”, “Find all shows with 4 or more main characters made in the year 2012”. If I knew of such a site, I would use it all the time, but I don’t think there’s really enough demand for something like that to happen at this point.
Your example query makes my mouth water. I imagine with a reliable, query-able database even an anime version of “average win share” isn’t out of the realm of possibility.
In terms of discs, it definitely isn’t. You’d have to do work to quantify how important large/small casts are and thus how big a role voice actors actually play, account for larger animation budgets/staffs, and factor in source material advantages/disadvantages, but I’m fairly certain you could break down certain people’s contribution to disk averages to some degree with information that’s very much available.
I’ve done a lot of the work on source material already, and the voice actors stuff is something I’ve been toying around with without actually collecting data. The budget/animation part would be the trickiest bit of groundlaying, but there are people who study that stuff fairly religiously and could be asked about it (which is probably where I’d start). Of course, this is pretty much just a thought experiment than anything – I’m scared of signing on to large-scale projects like that because my last one took 2 months and yielded ambiguous results.
I think given the potential applications, it would be better to setup a database to automate the process as much as possible. If we put together a spreadsheet using something’s Series Data – Quick View as base, then add to it 3 columns, linking to each series’ MAL, ANN, Anidb entries. All that’s remain is figuring out a script that can datamine each link for a particular keyword (“Shinbo” “harem” “Kyoto Animation” “Aniplex” “moe” etc) and output rows with a match.
I don’t have a lot of scripting experience. Know anyone with some R and BeautifulSoup experience to make this happen?
Sounds ambitious, useful, and something I can’t really help with. I’ll keep on analyzing things one component at a time. It’s slow, but it has its perks; one of the the advantages of going through the figures one by one is that you occasionally find out new things examining the pages. If you manage to get any traction on that thing, though, let me know.
Are the databases convenient enough to easily generate graphs of fads over time (like peak maid, peak tsundere, etc?)