I think I may have touched on this briefly in the FAQ and in a comment before, but it’s worth explaining in its own post.

Sometimes when my numbers differ slightly from other sources, the reason (if I or they didn’t just make a typo, which happens!) is because I’m using, in rare instances, estimates for DVD sales numbers if I feel we can get reasonably close (within 100 or so, ideally within 50). But how do these estimates get made?

Below is a lengthy but hopefully clear example of the process (used by 2ch for a while now, though I only figured out exactly what they were doing not too long ago) for Nagi no Asu Kara Vol. 2 DVD.

**Step 1**

The first step is to check Oricon’s website to see how the disc ranked for the week. As long as a DVD was in the all-genre Top 300, it’ll have a ranking number in red under the product image. If you look at the product page you’ll see that NagiAsu ranked #93 for the week.

(I should point out here is that in this case, it’s not actually necessary to estimate the DVDs. Since it’s in the all-genre Top 100, we’ll get the numbers in the full list anyway. But, I can still use it to illustrate the process.)

**Step 2**

Next we see if we can place a reference point above it, but not too far away. In this case we have a perfect one – we already know from the preliminary listing that Gundam Build Fighters DVD v2 sold **597**. If we check its product page (or just wait for the full list) we see it ranked #92. We could in theory stop there. Nagiasu sold “very close to 597 DVDs”. Maybe call it 590 and walk away very confident that we have a very very close estimate – virtually guaranteed to be far closer to the truth than 0!

**Step 3**

The next step is worth doing to see if you can shore up the estimate though, and that’s to find a reference point *below*, but not too far below. This can be harder, because you need to find a disc that ranked, yet sold less than the disc you’re looking into. How do you do that? You look at the sub-genre specialty rankings. If you don’t have access to Youtaiju (which only provides a handful of sub-genre lists anyway) one potential source is the weekly ranking summary on Oricon’s public website, where they publish actual sales numbers for the top few sellers in other lists.

The list I relied on for NagiAsu is the live action drama DVD ranking. In that list we can see that The Walking Dead 3 DVD Box 2 sold 581 discs. Its product page confirms that it ranked #95 for the week. That’s almost as perfect as it gets, because we can say the following with certainty:

#92 – 597 – Gundam Build Fighters v2

#93 – ??? – Nagi no Asu Kara v2

#95 – 581 – Walking Dead 3 Box v2

It sold between 581 and 597 copies. That’s a range of only 16 discs. So pick 582 to be super conservative (it’s probably more like 590) and you have a completely legitimate number that is at the absolute most off by only 15 discs, which is a much smaller margin of error than choosing 0 and being off by upwards of 596!

This is *exactly* the same method we used to estimate Non Non Biyori‘s v1 DVD number so closely. In that case we knew it ranked 1 spot higher than a drama that sold 874, so it sold at least 875 and with that you’ve got a nearly perfect estimate.

So why don’t we use this all the time? Primarily because getting such a tight estimation range is *pure luck*. Sometimes the disc in question is dozens or even more than a hundred ranks away from another reference point. If your range to choose from is ~400 discs, you risk being quite far off and I prefer to not estimate in that case.

As mentioned, in NagiAsu’s case, we’ll get the exact number in the full list anyway. But if we didn’t, we’d have a damn close estimate available! One more than good enough to officially use in rankings and such.