Plain Reckoner

Everybody quotes the same numbers. Nobody measured them.

What a thousand pageviews actually earns. What an ad network wants before it will take you. What a site like yours sells for. Trace any of those numbers back and you find a blog citing a blog citing a post from 2019, with no original source at any point. So we measured them ourselves, from businesses whose traffic and revenue a broker verified before they sold. Every figure says how it was worked out and the date it was read, so you can check it instead of trusting us. All of it free, with no account.

pageviews measured
1.17bn
revenue behind them, broker-verified
$6.84m
of real sales read, price by price
$126m

Two measurements, not one survey: ad-funded sites followed month by month, and every completed sale of one in a broker's public record · every figure dated, sourced, and repeatable from the raw data

What a small site actually earns, per thousand views · measured across 1.17 billion pageviews from 123 verified sales

Latest finding

16,085sites in the frame; we checked 462 of them against their network's own customer list

1 in 9 Mediavine sites cannot prove, from their own ads.txt, that Mediavine sells their ads

Ad networks publish lists of the sites they sell for. Publishers publish a file called ads.txt naming who may sell their ads. Both are public, so the two can be checked against each other — and nobody does. Across 462 publisher-to-network pairings, 36 could not be confirmed from the publisher's own file: 20 serve no ads.txt at all, 6 name no ad network, 5 name a different one, and 4 return a web page where a text file should be. The rate is not the same everywhere. Mediavine sites failed to confirm 11.1% of the time against Raptive's 3.1% — a 3.6-fold difference on the same test.

Measured by us··Read the method →

The newest 8 of 24 findings
gr figure finding
Grade: A. Primary source, independently checked — 1 independent verification 16,085 — sites in the frame; we checked 462 of them against their network's own customer list 1 in 9 Mediavine sites cannot prove, from their own ads.txt, that Mediavine sells their ads sites in the frame; we checked 462 of them against their network's own customer list
Grade: A. Primary source, independently checked — 1 independent verification 89 of 1,020 — publishers blocking a file that does not do the job they blocked it for 89 publishers blocked Google-Extended to stay out of AI answers. Google says it does not do that publishers blocking a file that does not do the job they blocked it for
Grade: A. Primary source, independently checked — 1 independent verification 68% — of sites that blocked an AI bot used a name the company behind it does not list Most sites that blocked an AI bot blocked the wrong name of sites that blocked an AI bot used a name the company behind it does not list
Grade: A. Primary source, independently checked — 1 independent verification 0.036 — rank correlation between what a subject earns per visit and what it sells for The subject that earns the most per visit is not the subject that sells for the most rank correlation between what a subject earns per visit and what it sells for
Grade: C. Single source, or an observation rather than a measurement — 2 independent verifications 27,394 — pageviews a month to clear Mediavine's $5,000 bar, at the measured median RPM Mediavine Journey takes you at 1,000 visits. The real tier needs about 27,000 pageviews a month to clear Mediavine's $5,000 bar, at the measured median RPM
Grade: A. Primary source, independently checked — 1 independent verification 11.9% vs 0.4% — how often ChatGPT-User is blocked against OAI-SearchBot, across 1,020 publishers Publishers block the AI visitor that would have sent them a reader at least ten times more often than the one that only files them away how often ChatGPT-User is blocked against OAI-SearchBot, across 1,020 publishers
Grade: A. Primary source, independently checked — 2 independent verifications 134 → 5 — Amazon-affiliate businesses sold, 2020 → 2026 Amazon-affiliate businesses were 45% of verified sales in 2020. In 2026 they are 4% Amazon-affiliate businesses sold, 2020 → 2026
Grade: C. Single source, or an observation rather than a measurement — 0 independent verifications 2 — cases — an observation, not a measurement Tool sites keep getting rejected for "low value content" — an observation, not a finding cases — an observation, not a measurement

Why you cannot look this up

Nobody publishes it

Ad networks do not publish what the sites they serve actually earn. The last breakdown by subject from a major network is from 2020. Every table of earnings per thousand views you can find is a blog quoting another blog — and an AI will hand you the same recycled figure, because that is what it read.

So we measured it

123 businesses whose traffic and revenue were both checked by a broker before they sold, matched month by month across 1,334 business-months. Not a survey. Not an estimate. Figures somebody was legally obliged to get right.

And we show our workings

Every figure carries how it was obtained, the date it was last checked, and a grade saying how far we would push it. So you can check any of them yourself — there are 24 findings here, and each one names the source it came from and the day we last read it.

What it changes

Before you believe any of this

You have no reason to trust us. Go and check.

We are not a name you know, and this is a page full of numbers about money. The correct response to that is scepticism — not just about us, about anything you read on this subject. So here is how to settle it without taking our word for anything.

  1. Ask any AI assistant about us. Whichever one you already use. Paste this address and ask whether the site looks legitimate, and whether these figures — what a thousand views earns by subject, taken from sales a broker checked; what ad networks currently ask for; what buyers paid, year by year — are published anywhere else. The answer to the last one is no. That is the reason this site exists, and you should hear it from something that has no stake in it.
  2. Check that the underlying data is public. It is. The sale records come from a broker's own public listings, the entry requirements from each ad network's own page. None of it is secret — it has simply never been measured and written down. Ask the same assistant whether anybody publishes what a thousand views earns, broken down by subject, with the sample it rests on stated. Then look at what we publish.
  3. Then check us on one number. Pick any figure on this site. Every one carries the sample it came from, the method, and the date it was read. If a figure cannot show you those, it does not get published here — and if you find one that cannot, tell us and we will pull it.

We would rather you spent two minutes checking than took our word for it. A reader who has verified us once does not have to do it again.

Free, and answered by a person

Have us look at your site

Send the address and you get back what the numbers here mean for you specifically: what your traffic is worth at the measured rate for your niche, which ad networks would take you at your size today, and whether you pass the AdSense requirements as Google actually words them — including the one page almost everybody gets subtly wrong.

One email per finding

Get these as they land

One email per finding — the number, the method, and the date it was true. No schedule, because measurements do not arrive on one.

Data controller: Plain Reckoner. Contact contact@plainreckoner.com.