Affiliate revenue per pageview runs from ¥0.1 to ¥417. Across 33 cases, the writer was not the one setting the rate

Two days ago we counted 225 AdSense blogs sold on a site marketplace and wrote that 1,000 pageviews is worth ¥386. That was the pay-per-click side. This time we count the other side, pay-per-result. Going through the cases on this site one by one and pulling out those whose main revenue is affiliate commissions (the advertiser pays when a reader signs up or buys) gave 20 Japanese cases and 13 overseas ones, 33 in all. The population is articles we wrote ourselves, and every figure traces back to the sources listed on each article.

The shape of the conclusion first. Under the same word “affiliate,” a pageview was worth anywhere from ¥0.1 to ¥417. That is a 4,000-fold spread. The 225 AdSense listings ran 13-fold from bottom to top. Three orders of magnitude apart. And the spread cannot be explained by how well the articles are written.

Median monthly revenue across the 33 is ¥1.4M, with 1,700x between top and bottom

Start with the revenue distribution. Nineteen cases are operating and publish an actual monthly revenue figure, 17 Japanese and 2 overseas, and the median is ¥1,400,000. That is Makurin’s number. Restricting to the 17 Japanese cases does not move it. Just below sit three cases at exactly ¥1,000,000 a month: Tsuzuki Blog, manablog and Unico Blog. As we wrote in three people who reached ¥1M a month, that is a figure people want to publish, not the distribution of figures people reach.

At the top are Kuronekoya’s ¥102M a year (¥8.5M a month on average), Kunitomi’s ¥4.84M a month, Hayato Ikeda’s ¥4.05M, Gadgemaga’s ¥3.78M and Santso’s ¥3.77M. At the bottom, Derako Blog’s ¥17,000 a month and the travel blog Single Traveler’s ¥5,000. Top to bottom is 1,700x.

A pageview is worth anywhere from ¥0.1 to ¥417

For the 18 cases where both monthly revenue and monthly pageviews are known, we divided one by the other. Overseas figures are converted at ¥150 to the dollar.

TravelMamas makes $15,000 on 200K to 350K pageviews, somewhere between ¥6 and ¥11, and is left out of the table. The median across the 18 is about ¥4, and ¥2.4 for the Japanese cases alone. The median for the 225 AdSense blogs was ¥0.39 per pageview, so a pay-per-result pageview sells for ten times a pay-per-click one. This site has a case of 3.5 million monthly pageviews at ¥0.5 each, the ceiling of ad-supported media. The top half of the table sits 10 to 800 times above it.

Read the table from the top and you notice that the genre is the sort order. At the top: high-priced wellness coaching, courses on making money online, careers, side jobs and investing, blogging tools. At the bottom: travel, fishing, manga analysis, general topics. Gadgets cluster in the middle at ¥2.3 to ¥6.4, three cases together. The Blog no Kamisama case settles it. Two sites written by the same person earn ¥0.15 and ¥8.2 per pageview, a 55-fold difference. And in terms of writing ability, the first site, 340 posts carried to 310K pageviews, is the stronger one, and its revenue stopped at ¥47,000 a month. The rate is set not by the writer’s skill but by the price of what the reader buys after reading.

The rate is what an advertiser can pay for one customer

Why does genre set the rate? Several per-conversion payouts survive in the articles.

• A crypto exchange account: up to ¥18,000 each
• A sports-betting customer referral: several hundred dollars each
• One WordPress theme: ¥48,000 a month (out of ¥53,000 in monthly revenue)
• One article comparing programming schools: about ¥500,000 a month
• A travel brochure request: ¥300 each
• Physical goods (Amazon, Rakuten): commission rates of a few percent

¥18,000 against ¥300 is 60-fold. The writer did not negotiate that gap into existence. An exchange can pay ¥18,000 because it will collect trading fees from that one account holder for a long time. A travel company pays ¥300 because the profit in one brochure request is thin. A conversion payout is the amount an advertiser is willing to pay the person who brought in one customer, and advertisers derive it from customer lifetime value on their own side. By the day the writer starts the first article, the number is already fixed.

This site has a standing position: what sets a price is not what the seller made but who the buyer is. It came out of tallying sale prices, and the affiliate payout is a small-scale version of it, where the buyer is the advertiser. Fin vs Fin could build $1M a year on 30K monthly visits because its readers were people who buy expensive coaching, and more than 300 advertisers were lined up putting a high price on each of them.

From here on we are guessing: this structure decides, almost by itself, the ceiling on what an individual can earn from writing. The effort of multiplying pageviews by ten and the choice of sitting in a genre whose rate is ten times higher do the same arithmetic. The first takes years. The second is settled before the first article is written.

Nine of the 33 record a drop of 50% or more in figures

Lining up only the numbers that went up would make this tally a lie. Among the 33, we counted the cases where the operator or the buyer wrote down a drop of 50% or more in actual figures. There are nine.

Beyond these, Tsuzuki saw his main article fall from 1st to 5th, and manablog closed two of three sites. Eleven of the 33 carry a record of decline.

The declines share a shape. When pageviews halve, revenue does not stop at half. Fin vs Fin lost 50% of traffic and 75% of profit. BitVirtual lost traffic and referral payouts at the same time and fell 98%. Monthly affiliate revenue is pageviews times conversion rate times payout, and of those three variables, ranking belongs to Google, payout to the advertiser and the affiliate network, and conversion to the product. What the writer holds is the article, and the article moves none of the three directly. In the Hitode panel, the remark that “when things are going well the network offers special rates, and when they are not it may cut them” is testimony that even the third term of the product is in someone else’s hands.

Our reading, for the record. These 11 are not 11 unlucky cases. People who kept publishing numbers for longer are the ones with a record of decline, so length of disclosure and presence of a decline measure nearly the same thing. The other 22 have no decline on record not because they did not fall, but because only the pre-fall numbers were ever published.

Sales ran from 21 to 60 months of revenue, and here too the buyer made the difference

Thirteen of the 33 ended in a sale, and for six of them a multiple can be computed from both the price and monthly revenue.

English-language individual sites landing at 21 to 40 months of revenue is the same band as the 19.4-month median of 225 sold AdSense blogs (on listed prices) and the 25-month median of 20 exits on this site. The two outliers sit above and below. CBWG got 60 months because a listed affiliate company was buying “search real estate in states that have not yet legalized,” ahead of time. The nursing-home directory got 81 months because the buyer was paying for a facility database, not ¥17,000 a month in profit. Multiples jump not when the business is better but when the buyer is looking at something other than monthly revenue.

The buyers are worth listing. Investor Junkie and CBWG went to XLMedia, Budgets Are Sexy to a Motley Fool network, and Tenshoku Antenna to the listed company Logly for ¥700M plus an earnout of up to ¥300M. In high-payout genres such as finance, careers and betting, listed companies whose core business is customer referral are always waiting as buyers, and only the cases that reached them landed in seven figures of dollars or hundreds of millions of yen. The buyers of Living Cozy, Root + Revel and WorldofTablet were individuals, and the prices stopped at six figures. What set the order of magnitude of the sale price was, again, not the articles but whether a company exists that buys that genre.

The ones who came back used affiliate income as raw material for something else

Among the 33, the only case with a record of climbing back above its previous level after a decline is Kuronekoya. The way back was not growing affiliate revenue. Four other cases had made the same move before their declines.

After falling to ¥50M a year, Kuronekoya sold more than 500 copies of a course priced above ¥100,000 and returned to ¥102M. The content of the course was the experience of reaching ¥9M a month through affiliate income and of being sunk by an update. Santso put an online community on top of seven sites and moved part of the revenue outside search. Hayato Ikeda’s ¥4.05M a month paired ¥2M in affiliate income with ¥1.07M in note sales. Pack Hacker tried a membership during its trough, and Gadgemaga added YouTube.

Calling this “diversifying revenue” is a little off, in our view. Diversifying would mean holding several revenue streams of the same kind. What these people did was use the experience and readers gained through affiliate work as material for products whose price they set themselves: courses, communities, paid articles. From a business whose rate is held by the advertiser to one whose rate they can write. Among the 33, Kuronekoya is the only one with a record of seeing that move all the way through, which is why there is only one record of climbing back above the previous level.

How we counted, and the limits

The 33 are cases whose text and metadata led us to judge that affiliate commissions are the main revenue. Many also run AdSense, and some, like Own The Yard, split roughly half and half with display ads. In the other direction, Half Marathon Guide was excluded as AdSense-only and BarBend as a 15-employee operation. We drew the line.

All figures are self-published, and only a few, such as Unico Blog, go as far as network statements or bank transfer images. Hitsuji’s ¥1.7M a month is “generated,” not “confirmed,” and he draws that distinction himself. The per-pageview table uses only cases where the revenue month and the pageview month correspond within the same article, but it includes cases like Laugh-raku where the month is not stated.

The population is people who wanted to publish their revenue, and the 11 who wrote about declines are people in a position to publish a decline. Publishing when revenue rises, and continuing to publish when it falls, is possible for someone who no longer lives on that number, or for someone whose decline becomes the next product. Kuronekoya, from the previous section, is the latter.

The only thing the writer could move was the first line

What stayed with us after counting is the sense that the order of the rate table was fixed from the start. Most of the 4,000-fold gap between ¥0.1 and ¥417 was decided by a single choice made before writing: what to write about. What can be moved afterward is pageviews, and pageviews belong to rankings, and rankings belong to Google. The payout belongs to the advertiser and the conversion rate to the product.

In the affiliate business, the writer holds two things: the right to choose where to sit at the start, and the right to turn the experience gained there into a different product. That is what these 33 cases left us thinking. Every other variable is on someone else’s desk.