The post that started it was unremarkable: a developer saying, on a Tuesday evening, that he had bought a domain. Four days later the site on it had taken $181,804 from 1,188 companies, and roughly 250 copies of it existed. By the time I finished writing this it was past $207,000.
It has been picked over by everyone at this point, and I couldn't leave it alone either. I wanted the actual numbers, so before the wave passed I captured the whole board: all 1,188 rows, on day four, at hour 89. Rank, price, category, and the clicks each listing reported.
One thing in there changed how I read the whole story.
The board is not a crowd
Ten listings paid $102,084 of the $181,804. That is 56% of the money from 0.8% of the board.
The median listing paid $5. Six hundred and thirty-nine of them, 54% of the board, paid five dollars or less, and between them they account for $2,506, which is 1.4% of the total. Run it from the other end and the shape is the same: by the halfway point of the board, 98.7% of the money is already in.
So "1,188 companies paid" is true and tells you almost nothing. This was a handful of large purchases with a very long tail of participation attached, and the tail is what made it look like a market.
Should you build one
Only if you already have the audience, and then the board is not the point.
254 people built the same thing inside four days and the median one earned nothing. Five days in there were 394 of them. What could not be copied was nineteen thousand followers who had spent two days watching him look for something to build, and a willingness to post the outage, the broken analytics and the hosting bill as they happened.
Should you buy a place on one
Only against your real cost per click, and only with a ceiling decided before the first payment.
The credible range is around $1.59 a click for untargeted traffic that bounces at 75% and stays 57 seconds. There is no search value: every listing link on the board carries rel="sponsored", which passes no ranking weight by design. I checked all fifty on the first page.
And the top of the board did not buy attention. The $15,000 spot was not the most clicked listing: four listings paying $5 to $10 reported more clicks than it did, and three of those four are rival leaderboards.
What is in the write-up
Twenty-two pages, and every figure in it opens its source.
- Minute by minute from an idea posted to an empty timeline to a live product, every timestamp decoded from the id of the post rather than read off a label
- Where the money came from, by slice and by category
- What a click actually cost, band by band, and why click counts below rank 250 cannot be believed
- What $15,000 bought, taken apart into search value, traffic, and what buyers say they got against what the board says they got
- The copies, and the two registries that cannot agree with each other
- A chapter listing every claim I could not verify, plus the eight figures I printed wrong before correcting them
The capture ships with the document. Part II is arithmetic on that file and nothing else, so you can recompute any figure in it, including the ones that disagree with the coverage. The zip carries a README describing how it was taken and what it is not.
This board gets read again
Day 7, then 14, 30 and 90. Same method, same file, because the interesting question about a wave is not how high it went.
Those readings go out from here, and I post them as they land at @dklymentiev.
