Ejc passed along something 2ch had discovered on the public Oricon website recently: http://www.someanithing.com/2608/comment-page-1#comment-42550

The big question was how to get a clean list, rather than blindly searching through over 1200 pages one by one on a site that is a pain to navigate at the best of times. I’ve managed to do that with some page scraping and a lot of text manipulation. The result is a list of 12,721 DVDs going back as far as Oricon tracks them, ranked by total sales, but without the sales number. You’ll see some listed with really old dates like “1988” which seems weird as there were no DVDs in 1988! But that’s just Oricon saying the original release came out in 1988, not that the *DVD* version came out in 1988.

The Data
DVD: Current as of 2014-12-24
BD: Current as of 2014-12-25update, added BDs

It’s rank/title/date/highest/#times separated by tabs, I recommend pasting it into Excel or similar to get proper columns. It’s their “animation” category so it’s a bunch more than just anime, but the more the merrier in this case as it gives us more to compare to.

Now the fun begins of how to use this. I mean one could go down the list one by one filling in what we know! But that’s tricky, and anything that ranked in the top 300 more weeks than it ranked in the all top 100/animation 30 is a release we can only fill in a partial number for, which is an issue. It’s definitely useful for the large list of discs that ranked only once, which is primarily what we care about. Most DVDs that sold enough to make top 300 more than once are likely to have

And of course the download and cleanup process will need to be completed every time you want new data, plus merging in any populated values with the new list… I have some ideas about how to do that but the initial site scrape itself takes about four hours, since I don’t want to hammer their site with parallel requests and try to keep at least a second between downloading and starting the new request. Don’t want to get banned (or have them wise up and remove it, if this isn’t something they intended for people to do!), you see.

I haven’t started doing it for BD either, but DVD seems more important right now.

48 Responses to “All Oricon-tracked DVD/BDs by rank (rank only, no sales number)”

  1. Avatar Tiago says:

    Hopefully some nisekoi sales are in this!

    Btw I can’t open those links, I wonder why.

  2. I know you’re really far ahead on things at this point, but I figured I’d post the results of what I had for the 2007-2010 crowd I’d been looking at. Shana II, Hayate (v11 was missing), and Star Driver all add some on some DVD volumes, Junjou Romantica I couldn’t find a single volume I could see a boost for.

    Hayate no Gotoku season 1: http://pastebin.com/wHt6zHpB
    Junjou Romantica: http://pastebin.com/k4STuTV9
    Shakugan no Shana season 2: http://pastebin.com/SKmS83AR
    Star Driver (DVD only): http://pastebin.com/be3VEGPQ

    Doing a little helped me appreciate how intensive the DVD forest can be. Good luck with it!

  3. something something says:

    Progress update.

    BDs DONE! All 3,457 BDs in the ranking as of when I pulled it on 12/24 are complete. The Oricon ranking should be updating again today/tomorrow but at least going forward I just need to identify and record updates.

    DVDs, though… well yeah that’s gonna be a long while yet. I’ve got about 10,000 left to go through. On the one hand, a larger percentage will be things I don’t track than in the more anime-dense BD list. But on the other hand, that sometimes makes it worse because I still need to look them up to try to find reference points for the stuff I do want to track.

    So, gonna be a while yet. But I do intend to get everything in eventually. I guess I took vacation at the right time, would such much worse if I were trying to get this done just a few hours here or there on weekends. Back in work on Monday though. ;_; No way I’ll be done by then.

  4. something something says:

    http://www.someanithing.com/2645/comment-page-1#comment-45160

    This is how I plan to handle comments (or lack thereof) for long tail and low sale additions obtained through the new method.

  5. something something says:

    As a reply to: http://www.someanithing.com/2733/comment-page-1#comment-46925

    1. Genre bias? I haven’t been looking but I’d guess no. The increases have more to do with release specifics like formats and timing and how old the show is than anything else.

    2. Bestsellers with longer tails? Yes, if only because they’ve just got more discs to work with. Their tails could for all I know be proportionally about the same as low sellers, but if the last 1% is 500 discs, that might make cutoffs that 50 or 5 won’t.

    Overall the large majority of new data is previously unranked wk1s or wk2s. Through 2002, we only got a top 50 and the extended rankings only went back to 100. And this when DVD had already taken over, so that was terribly insufficient.

    Thus a show like Seikai no Senki (2000-2001 releases) officially has a 9.6k average which you’d think would be a lock for extended rankings putting it over 10k. I’d wager it’s real average is upwards of 12k. Or hell, 15k. But v4-5 look like they’ll miss wk2s even in the extended rankings, despite together likely adding thousands. It *is* possible some of the other vols will get wk3s and push it over the top but I’ve got a few thousand discs to check between me and most of its volumes.

    Of course these super strict 1999-2002 thresholds provide many of the milestone markers used to boost estimates of the many many shows that came after them. So I respect your sacrifice, Lafiel. ::salutes::

    3. Bombs to not-bombs? Probably not, at least not anything drastic. If a series truly bombs, we only know because wk1s were very low or unranked. And while a significant chunk of sales can be hidden in the extended list, raising an average more than 2k is incredibly hard unless the show was selling a ton already.

    Some series have gone from never ranking to most volumes ranking (e.g. Gad Guard to 1145, Divergence Event to 966, El Cazador to 798, Cross Game to 579) but that’s still sub-“decent”.

    One series I’ve been noting however is Dennou Coil, a 2007 show that averaged 2,534 but with extended rankings is up to 3,831. This isn’t because of long tail though. Wk2s only account for ~298 of the gain. REs, none of which had ranked before, are the cause. 7 of the 9 made the extended rankings and they average 1.3k, which boosts the average by 1k.

    A note on REs and averages for those wondering: you can count two ways. One is to add an RE with its LE and use one number per volume for averaging. The other is to take separate LE and RE averages and then add those together. The latter results in higher numbers, Dennou Coil would be 4.1k for example.

    I favor the former approach though, not least because it makes the formula nightmare of my spreadsheet a little more predictable, but also because it addresses cases where only one or two volumes had an RE, or where only one or two of them ranked. They could have been especially expensive LE versions for those vols, driving unusually high RE sales.

    A few shows with decent RE sales and maybe a third of their REs missing (Dennou Coil, Scrapped Princess, Chrono Crusade) will miss out omewhat. But for most either all their REs rank (FMP Fumoffu/TSR, Marimite) or only 1-2 rank. And of course most have none rank. So I’d rather just be consistent. In some ways missing a few REs is no different than missing a few wk2s. It can’t be helped and you can drive yourself and Excel nuts trying to account for all the special cases.

    (This does mean Haruhi is likely to drop a tiny bit because I was using the other calculation before when REs were super rare, now there are enough I need consistency. Ah well.)

  6. Avatar mangamuscle says:

    One idea/petition. Would be to hard to extract the exact disc sales for Manabi Straight? I am curious if the “Manabi Line” was really set at 2899 (I think it was a bit higher, maybe in the 3200+ range)

    • something something says:

      Thresholds were high back in 2007 so only vols 1 and 7 got wk2s, and they were small. So v1 increases from 2899 to 3394 and the average from 2319 to 2435.

      That said, I’ve got to go into my Manabi Line rant now, sorry! It was never of any real utility, and it tells us nothing about whether a show really broke even or not.

      The 17,048 discs we’re aware of Manabi selling comes to ¥97,230,000 (MSRP/post-tax), If production committees get 55% (I believe it was) of MSRP on average, that’s ¥53,476,500 or about half a million USD, subject to wild swings in exchange rate of course.

      The price to animate a single episode of anime is variable, but ¥11,000,000 is a number I’ve seen a few times before. Even if we assume Manabi was the most average of average series, its disc sales wouldn’t even cover producing five episodes. That’s not counting any other costs involved in making anime not accounted for in the ¥11m claim. And if Manabi were cheap enough to break even on those sales, it would by definition be a poor yardstick, because it wouldn’t be an “average” show. (In general I don’t think talking about “average” budgets or costs is helpful anyway.)

      I always advise against measuring break even using just video disc sales, because video discs are never the only source of income. Manabi also had a manga, and CDs, and a PS2 game, and your typical merchandising effort.

      Another problem with the Manabi Line is that people usually cite the Vol. 1 numbers, but obviously average sales tell you much more about a show’s performance than Vol. 1 numbers do. Just look at the massive gap between Vol. 1 and average for Love Live! s2 and Utapri s2 for extreme examples.

      Most importantly, Oricon numbers only reflect what was reported as sold at the retailers Oricon tracked at the time. But the actual money made would be based on the number of units shipped, and we don’t know what that number is. They could have shipped way more than sold and “broken even” that way. Or maybe most did sell, but just didn’t make Oricon’s crazy DVD thresholds or weren’t at stores they tracked. And since their tracking and reporting changes over time, 5k in 2000 is not 5k in 2007 is not 5k in 2015.

      In other words, an average of “2,435” is not an absolute number, it’s better thought of as an abstraction of one measure of success among many. A way to roughly compare one show to another, but only if you keep many caveats in mind (thresholds, data availability, age of release) and talk in probabilities, not absolutes.

      The Manabi Line is arbitrary, just like “10,000” – a number we only use because as a species we’ve decided base-10 is a neat number system. Manabi Line is more problematic though, because while nobody claims 10,000 to mean anything in particular, the Manabi Line is cited as having real world implications for production budgets.

      tl;dr there’s no single break-even point, and even if there were – even if there were *and* it were equivalent to Manabi’s sales – we don’t actually even know all that accurately what the Manabi Line would be!

      • Avatar mangamuscle says:

        First and foremost, thanks for your work -_-

        I know the manabi line is not (and never was) an absolute. But since we will never know each series budget, animation costs, publicity costs, TV broadcast costs, etc. it gives us a vague idea of (i.e.) where to start losing hope for a second season. Of course, I cannot understand my husband got a second season and Shingeki no Bahamut getting one has little to do with disc sales and all to do with increased card sales.

        • something something says:

          I know it was a long reply, but it wasn’t aimed directly at you – just at the idea of the Manabi Line in general! It comes up a loooooot. I even used to use it, but it’s such a problematic idea because it is so overly reductive. What I try to stress above all is that no two shows’ sales numbers are ever perfectly comparable, and disc sales themselves are just one piece of a larger pie, and the importance of that piece varies enormously.

          Better to judge every show independently, taking into consideration who is on the committee, what else is being sold, and so on. It’s more complex of course, but I prefer to embrace the uncertainty over thinking we have a number that means something it doesn’t.

  7. something something says:

    Progress report… Still at it: http://i.imgur.com/hjz49VB.png

    BDs: 3,552 done, 0 remaining. Fully checked through this week’s additions, completely caught up.

    DVDs: 9,895 done, 2,999 remaining. Technically 3047 but some of that is the same release with the exact same name but on two different dates, which makes sorting funky. But they’re all mainstream movies I don’t care about. Latest data is merged in, so fully caught up for everything from 3000 and up.

    Will keep merging in the newest data each time Oricon updates so I always stay on top of that and use the latest data.

    I’ll be done, some day. Soooooome day.

  8. something something says:

    About 2,400 DVDs left. Two more notable things that have come out of it:

    1. Maria-sama ga Miteru season one joins the 10k club! Some strong wk3s for the RE versions put it over the top.

    2. Ayakashi, as in 怪 ~ayakashi~ from Winter 2006 shows probably the most extreme case of a long tail in any TV series I’ve tracked, on a relative basis. It’s fairly well known that of the three story arcs included in that show, the last arc (Bakeneko) was the most popular. It went on to spawn the Mononoke anime, which averaged at least 12,879.

    Before looking at the extended ranking data, all we knew was that the Bakeneko arc sold 2,140 while the first two did not rank. If Rakuten is correct, all three volumes came out on the same day, and the threshold that week was 1,055. All three had LE and RE versions, but the only disc to rank was the Bakeneko LE. The RE missed the cutoff at #108, implying it was around 1,000. It never did make the countable rankings.

    But I just came across the RE in the data, and damn, it ranked 11 times and sold 6,661. Add that to the 2,763 (with wk2 data) sales of the LE and compare it to the extended ranking numbers for the first two discs and Ayakashi’s sales look like:
    Arc 1: 510
    Arc 2: 375
    Arc 3: 9,424

    I have a column in my spreadsheet that calculates sales increase or decrease vs previous volume. For Ayakashi v2→v3, it’s up 2413.07%. I had the widen the column to display the number because nothing has ever hit four digits before. So for people who have sent me various questions on ask and elsewhere about whether discs ever have very low wk1s but go on to sell a lot over time, you have a new best example! Just don’t expect to see something quite like it for a long long time, possibly ever.

  9. something something says:

    Another day another new 10k!

    Full Metal Panic? Fumoffu gets a significant jump in the full data from 8,917 to at least 10,093, adding a bunch of missing wk2 data for both LEs and REs.

  10. Avatar Anon says:

    Dumb question. Is there a similar mass ranking chart for manga?

    • something something says:

      Good question, not a dumb one!

      Funnily enough, the way the “exploit” works for DVD/BD is that Oricon has a section of their website that includes “profiles” of various actors, musicians, writers, directors, etc. that each link to a information, including the sales ranking of the works they’ve created, starred in, or have some connection to.

      Conveniently for us, one of the profiles is for “アニメーション” (animation), which is a broad category they by default append to every single anime (and foreign cartoon and CGI as well) disc they track.

      So essentially we’re getting these lists by browsing the profile of someone named “Animation”, who is so incredibly prolific they’ve starred in nearly 12,800 DVDs and over 3,600 BDs! Heh.

      Unfortunately, I don’t see a profile for “コミック” (comic, manga) or “ライトノベル” (light novel) so the same trick can’t be used there.

  11. something something says:

    About 1.7k DVDs left. Newest 1-k is, unsurprisingly, My-HiME, which was already 9.994 before the extended rankings.

  12. something something says:

    Less than 1000 to go, finally. As far as new 10ks, like torisunanohokori already pointed out, Aa Megami-sama TV s1 and Otome wa Boku ni Koishiteru officially join the 10k club.

Leave a Reply to something