I’ve done a few small bits of data analysis in response to questions received on ask.fm recently, and now that there are a few of them I’ve collected them here rather than just leaving comments in the top post. Click the header links for the full reply.

Questions from ask.fm
1. How often do sequels increase in sales?
Summary: “Nearly three quarters of series drop more than 20%, though only about a fifth drop more than 50%. Meanwhile only about one in ten increases by more than a quarter, while only a little over one in twenty doubles its sales.”

Note that some of this data has already changed since I posted that, for example Nanoha 1st went up a bit due to the extended ranking data. No change to the big picture though.

2. Do anime ever start very low, but build up a long tail over time?
Summary: “In that dataset, 6,054 individual discs ranked sub-2500 wk1. Of those, 1733 (28.6%) managed more than one week of ranking, either in the normal lists, or in what I have found so far in the extended rankings. (If you’re curious, fully 1372 of those 1733 are from the extended rankings – it’s added a ton of new data). This number will rise somewhat as I continue through the extended rankings. I have ~3.5k of ~12.8k DVDs left to check. Only 9 of those 1733 discs managed to go from sub-2500 wk1 to >5k total. Only two of those are BDs, and in one case (T&B) it was due to a severe LE stock deficiency pushing desperate fans to the RE version. NagiAsu is thus the only good example in the BD era.”

Since I did that post, I came across the pretty amazing case of the RE version of the Bakeneko arc of Ayakashi in the extended ranking data. See this comment for details. From maybe barely 1k and not ranking in the top 100, to over 6.6k total in 11 weeks.

3. Which sells better, original series or adaptations?
Summary: “Conclusion: No, sales of originals and adaptations are no different. In fact they’re virtually identical once you control for a tiny percentage of outliers.”

4. Does the Winter anime sell worse than other seasons?
Summary: “Winter and Summer have very similar averages. Summer and Fall have very similar averages. Winter and Summer shows do about 1.1k to 1.4k less in average sales than Spring and Fall shows.”

I found the last two fun because of how clearly they show the power of averages to smooth out outliers. You don’t always want to look at huge aggregate averages, because sometimes that hides interesting information. But when the question is broad like “Does [huge category of shows] sell better/worse than [other huge category of shows]?” having a big enough dataset makes clear that despite some Madoka Winters or Sunrise originals skewing some portion of the dataset, the general trends are very consistent and we shouldn’t make overly broad conclusions from outliers.

Questions from comments
1. 2013 total unit sales vs 2014 total unit sales?
To compare years we have to use disc release date, not airing date, so:

Airing Season Released in 2013 Released in 2014 % Change
Previous year series 1,699,724 1,663,828 -2.1%
Winter 601,882 500,989 -16.8%
Spring 1,284,937 1,175,158 -8.5%
Summer 624,415 516,994 -17.2%
Fall 111,928 79,664 -28.8%
Total 4,322,886 3,936,633 -8.9%

Because some shows may have been boxsets or gone a lower disc count/higher MSRP route, let’s see if measuring in yen changes anything:

Airing Season Released in 2013 Released in 2014 % Change
Previous year series ¥11,375,720,000 ¥12,081,360,000 +6.2%
Winter ¥3,932,150,000 ¥3,370,810,000 -14.3%
Spring ¥8,777,340,000 ¥7,740,490,000 -11.8%
Summer ¥4,081,890,000 ¥3,520,700,000 -13.7%
Fall ¥692,730,000 ¥550,980,000 -20.5%
Total ¥28,859,830,000 ¥27,264,340,000 -5.5%

Counting in yen gains ground for 2014 in every season (especially for Fall and previous years) except Spring, where a large percentage of the unit sales were the half-price Love Live! 2nd Season v1. This reduces the drop from 8.9% to 5.5%, a pretty significant change and a sign that measuring in unit sales can be problematic in some ways.

IMPORTANT CAVEATS
0. The above are all totals however, I did nothing to control for number of releases. And there were more releases in 2014 (771) than 2013 (706) by a fair margin. This means 2014 was a harder drop versus 2013 that the numbers show, because the units and yen per volume were lower.
1. This is release date, not airing date, hence why remnants from previous years are so much higher than the Fall of the actual year in question. Most Fall 2012 shows (and for that matter a good chunk of volumes for Summer and even some Spring and Winter) are in 2013’s number, and so the same for 2013 vs 2014. Fall 2013 and 2014 shows have 0 or 1 discs per show in their respective year’s number. This means 2013 gets most of Garupan’s sales while 201 gets most of Monogatari 2nd’s sales, for example.
2. Some discs released in December 2014 will still add additional discs. In particular, Love Live! 2nd Season v7 could have another 10k discs left in it, so the 2014 number will end up higher, just not enormously so.
3. This is only shows I track, which does not include children’s shows, or movies, or OVAs. We’ll want something like the full JVA report for a more broadly applicable “health of he industry” answer.
4. In other words, we can’t use this number to measure the year to year health of “all anime created in year {CONTENT}“, because it’s not intended to be that comprehensive. It only looks at one (albeit major) segment of the industry, primarily the late night TV anime market. We’d need to check out something like the JVA reports to get a wider view.

14 Responses to “Miscellaneous sales statistics from ask.fm questions”

  1. primadog says:

    Can this also serve as the “ask me” section of the website?

    Did TV anime see an unit sale decline in 2014?

    • something something says:

      I hadn’t really thought that far ahead, I suppose I was just going to do these every month few months if enough interesting questions came through ask.fm. I figure it wouldn’t last long because there are only so many ways I can (reliably, easily) slice the data and questions would eventually get repeated.

      For now, if questions come through the comments, and warrant a long enough reply to edit them into the post, I’ll do that. So I’ve added yours as #1 under “Questions from comments” above.

      • something something says:

        Updated it with an important qualifier, that is, the number of releases included in each of those years. 2014 had more releases, which means its drop vs 2013 is a little worse than the numbers imply, because there was less to go around per disc.

  2. It’s complicated anyway because of all the other markets and stuff, but the -10% per-season change seems somewhat significant. There’s a lot of ways to slice it, since so much is manga (and anything else) tie-ins and a lot of those did pretty well this year. Definitely a conversation starter, though.

  3. moonnya says:

    I have a question about boxes. Usually we will see 2 points in stalker , one use expensive No products correction (——————高額商品補正なし—————— ), has higher value, and one does not use that, which has lower value. So which one is more correct? Thanks

    • something something says:

      Stalker used to give high price items the same points per day per rank as normal priced items. But Amazon’s rankings factor in not just units sold, but MSRP. 100 sales of a ¥7,000 item will not rank as highly as 100 sales of a ¥20,000 item. Therefore, expensive items would get more points for any given rank than they should. So Stalker started applying a heavy penalty to high price items. (I believe high price is anything over about ¥13,000.)

      Of the two numbers, use the lower.smaller estimate. It’s more accurate.

  4. iku says:

    If two series that share the same type of audience get released in the same huge week, is it possible that they would each cannibalize off of the other’s larger potential sales?

    • something something says:

      I guess shows relying more on impulse buys or physical retailers could be more prone to this, but so much anime is preorder heavy that I don’t think it’d be a huge effect.

  5. primadog says:

    How wide are the estimate ranges in the high end?

    • something something says:

      As you’d expect, it ramps up exponentially.
      • In the bottom 95% of BDs, there’s no gap larger than 1,016 and the average gap is 30 and the median is 5.
      • In the bottom 95% of DVDs, there’s no gap larger than 268 and the average gap is 8 and the median is 2.
      So we’re talking suuuuuuper close numbers through the vast majority of the list and when you talk about the bottom half, it’s even more incredibly tight.

      In the top 5% there are some within a few hundred and others tens of thousands apart. In the top 25 gaps, DVDs are anywhere between 5,678 and 793,249. BDs are anywhere between 454 and 1,893,391.

Leave a Reply