EROMANGA
HALL OF FAME

EROMANGA HALL OF FAME, Research Paper No. 2 ── first published 2026-10. Survey date 2026-10-27. The URL of this page will be maintained permanently; should the contents be revised, the edition marker will be advanced and the revision stated.

Put out one doujin book, and how many copies will it sell?
Every artist has wondered at some point, and readers sometimes imagine it too ── yet the question has no proper answer. What one hears is either the story of a jackpot that sold tens of thousands, or the lament of a book that sold nothing at all. How much the great majority in between actually sells, we have found no published count of.

So we counted. The sales of 155,154 doujin manga on sale at DLsite ── every last one.

A note on scope: these are Japanese doujin works ── adult manga self-published in Japan, in Japanese ── as sold on the country's two largest download stores, DLsite and FANZA. They are not figures for "hentai" sold worldwide, nor for print copies sold at events such as Comiket. DLsite also sells to readers outside Japan, and its counts include them; translated editions, sold under separate numbers, are excluded.

Cumulative copies sold of DLsite doujin works of type "manga", paid works with at least one sale (155,154 works). As of 2026-10-27.

The answer first: the middle work sells 170 copies. Line every work up by sales, and the one that falls exactly in the middle sold this many ── half sold fewer, half sold more. What follows looks inside that figure, chart by chart.

Fig. 1 ── Shrunk to a hundred works

The 155,154 DLsite doujin manga shrunk to 100 works, one square per work, shaded by sales band. Touch or hover over any bar, point or square to see its figure.

Of a hundred works, 39 have not reached 100 copies. The largest group is the 46 that sold 100–999; 15 pass 1,000. Works past 10,000 copies number 1 in a hundred ── 1,501 works in all (0.97%).

Fig. 2 ── The distribution of sales

The horizontal axis is the sales band (each step right is roughly threefold); the vertical axis is the share of works in that band.

The peak is the 100–299 band. The best-selling work has reached 410,901 copies; that the mean (750) is 4.4 times the median is the work of such a handful of hits pulling the average far upward. This is why "on average, X copies" sounds so far from the doujin artist's experience.

Show the figures (quantiles)

Position	Copies	Reading
Bottom 10%	13	a tenth of all works at or below
Bottom 25%	47	a quarter at or below
Median	170	exactly the middle
Top 25%	546	a quarter at or above
Top 10%	1,543	1 work in 10
Top 5%	2,909	1 in 20
Top 1%	9,798	1 in 100
Top 0.1%	34,081	1 in 1,000

Fig. 3 ── Where the sales gather

The total of all copies sold (116,399,881), divided by lining the works up from the best-selling down.

The top 1% of works take 26.8% of all copies sold; widen it to the top 10% and it is 67.2%. The bottom half, meanwhile, share a mere 4.0%. Doujin manga sales are heavily concentrated in a very few works.

Fig. 4 ── Ten years: what grew?

Three figures by year of release, on a common horizontal axis. Sales are cumulative to the survey date, so the latest two years (pale) are still selling and come out low.

The release year with the most works past 10,000 copies is 2024. We compare it with 2014, ten years earlier. New works rose from 3,974 to 16,616, about 4-fold. Works past 1,000 copies rose from 413 to 3,098, about 8-fold. Works past 10,000 went from 8 to 264, about 33-fold. Yet the middle work sold 185 then and 182 now ── hardly any change.

The market for doujin manga on DLsite has grown large over these ten years. But the growth gathered at the top; it has not reached the middle. And this comparison favours 2014, whose works have been on sale ten years longer. The gap opened even so. Works released from 2025 on are still early in their sales, and are not used for the comparison.

Note ── on the figures for early years

Works from the earliest year, 2005, show a high middle figure of 279 copies. Works no longer on sale do not enter the survey, so the older the year, the more its figure is that of the survivors still on sale today.

From the circle's side ── those who carry on, and those who stop

We took the 3,330 circles that released their first work in the same period (2016–2019), grouped them by how well that first work sold, and counted whether they went on to release five works or more.

Fig. 5 ── First-work sales, and the share of circles that went on to five or more works

Of circles whose first work passed 1,000 copies, 71% have released five or more. Of those whose first work fell short of 100, 28%. The first response and whether one goes on drawing are plainly linked. Put the other way round: even with a first work under 100 copies, about 3 circles in ten have gone on to five or more.

Fig. 6 ── Circles with ten or more works: copies sold by the 1st through the 10th work (median)

Then does carrying on make each work sell more? Line up the works of circles with ten or more (3,733 circles) in order of release, and from the first work's 302 copies to the tenth's 283, the line is nearly flat. What grows by carrying on is not the per-work figure but the number of works. Among circles that began in the same 2016–2019 span, total sales come to 58 copies for circles that stopped after one work, against 8,094 for those that released 10–29 (both medians) ── a 140-fold difference.

A first work's sales include copies bought later by readers arriving through subsequent works. The longer a circle continues, the more its first work's figure tends to grow afterward, and part of the gap in Fig. 5 is due to this. Works released in the latest two years are excluded from Fig. 6, their sales having had little time to accumulate.

Show the figures (circles by number of works)

Works	Circles	Works	Copies per work (median)
1	6,096	6,096	63
2–4	6,162	16,798	111
5–9	3,399	22,220	192
10–29	2,957	48,023	267
30–99	776	36,326	231
100+	100	25,691	39

The sellers with 100+ works include many labels that distribute serialized stories split episode by episode in large numbers; they are a different kind of thing from an ordinary doujin circle, and are excluded from Figs. 5 and 6.

Circles that began in 2016–2019	Circles	First work (median)	Total (median)
released 1 work	828	58	58
released 2–4 works	937	144	373
released 5–9 works	659	247	1,685
released 10–29 works	729	411	8,094
released 30–99 works	177	381	19,553

Fig. 7 ── Price and sales

Copies sold by the middle work in each price band.

The middle work sells best in the ¥551–880 band (216 copies); the relation is not a simple one in which the cheaper a work, the better it sells.

Prices are the selling price on the survey date; works on sale enter at the discounted price. Price is tied to page count and the character of a work, so this chart does not mean that changing the price would change how a work sells.

Show the figures (price bands)

Selling price	Works	Median	Entry to top 10%
up to ¥330	21,069	87	930
¥331–550	31,201	144	1,139
¥551–880	79,409	216	1,760
¥881–1,320	17,322	183	1,808
¥1,321+	6,153	136	2,207

Fig. 8 ── On FANZA, they sell more

Everything so far is DLsite alone. But doujin manga has one more large storefront: FANZA. We counted FANZA's doujin "comics" (152,670 listings) the same way, and set works at the same position side by side.

Copies sold; each gridline is tenfold. Each point is the sales of the work at that position when each store's works are lined up by sales.

At the middle, FANZA's work sells 243 copies, 1.4 times DLsite's. The gap widens toward the top: at the entry to the top 1%, FANZA stands at 93,903, about 10 times DLsite. In the way hits sell, FANZA is by far the larger market.

Show the figures (DLsite and FANZA)

Position	DLsite	FANZA
Works	155,154	138,268
Bottom 25%	47	63
Middle	170	243
Top 25%	546	960
Top 10%	1,543	4,046
Top 1%	9,798	93,903
Highest	410,901	465,278

Fig. 9 ── Of one work's combined sales, the share sold on FANZA

Across the 1,707 works we could match as the same work on both stores, FANZA's share of combined sales was 66.7% for the middle work. FANZA takes more than half for 69.7% of them, and more than nine-tenths for 21.3%. For such works, saying "this work sold so many" from the DLsite figure alone will often mean seeing less than half of the real total.

Both stores' figures are cumulative and include copies sold during discount campaigns. The matched works lean toward the well-selling works in and around this hall's ranks; the share need not be the same across all works. On FANZA, 8.3% of listings (12,659) do not display sales and are excluded from Fig. 8.

Fig. 10 ── How many buyers stand behind one star

A DLsite work page shows the number of star ratings next to the copies sold. Among works with 100 or more sales, about 1 buyer in 4 leaves a star rating at the middle work.

The ratio falls, though, the better a work sells. For works with 100–999 sales there are about 3.4 buyers per star; for works past 10,000, about 6.3. When estimating sales from the star count alone (for works that hide their sales, say), this spread has to be borne in mind.

Show the figures (share of buyers who rate)

Copies sold	Works	Share who rate (median)	Buyers per star
100–999	72,114	29.8%	3.4
1,000–9,999	21,971	23.4%	4.3
10,000+	1,501	15.8%	6.3

Summary

A "typical" doujin ero-manga sells 170 copies.

Stories of tens of thousands of copies are real. But works past 10,000 are fewer than 1 in 100 ── the very top. Sell 1,000 and you are in the top 15%; the middle work sells 170. Most works reach readers numbering in the tens to the hundreds.

And this "typical" has not moved as the market grew. In the ten years from 2014, new works multiplied about 4-fold, and works past 10,000 copies went from 8 to 264. Only the top grew; the middle stayed where it was, 185 then and 182 now. Whether one goes on drawing divides sharply on the first response, and the circles that do go on see a nearly flat figure per work, raising their totals by piling up works.

So for a work in the middle, one reader weighs a good deal. To a work of 170 copies, one purchase is one 170th. Leave a star, and it stands on the work page as the voice of 4 buyers. Rating a work you liked is a larger vote than you may think.

Method

DLsite: every work number in the sitemap of the doujin floor was queried in bulk, and works of type "manga" were counted (retrieved 2026-10-27, queries at 10-second intervals). Sales are the cumulative figure shown on each work page. Only paid works with at least one sale or rating were included (177,045 works). Of these, translated editions (21,890 works, released under separate numbers with a prefix such as 【◯◯語版】 in the title) are separate editions of the same work and were excluded, leaving 155,154 works. Circles were counted by seller ID (19,490). For how circles continue (Figs. 5–6), sellers with 100 or more works were excluded.
FANZA: the full list of doujin "comics" was retrieved, split by price band (2026-10-27, 10-second intervals). Sales are the cumulative figure shown in the list. Listings without a figure were counted as undisclosed, by number only.
Matching across stores: pairs from the same-work table this hall maintains for its own operation (matched by title, circle and cover), where both stores show sales.

Limits

(1) Sales are the stores' published figures; no estimate of ours is included. (2) Works with no sales and no rating, free works, and works no longer on sale do not enter the survey, so the lower figures come out higher than reality. (3) Being cumulative, newer works come out smaller. (4) Prices are the selling price on the survey date and are affected by discounts. (5) Compilations and bundles are each counted as one work. (6) FANZA's list carries no published release date, so the analyses by year and by circle use DLsite alone; a circle with a single work on DLsite may be continuing on FANZA or elsewhere. (7) Doujin manga is sold through other stores too (Melonbooks, Toranoana and others), so a work's full sales exceed the figures shown here.

How to cite this page

Survey: EROMANGA HALL OF FAME (this hall's database). Related: Research Paper No. 1 ── A Survey of Long Careers in Adult Manga · Doujin Works Hall