Notes
Every note below rests on something we measured rather than something we read. Each one states the number, how it was obtained, and the date the evidence was current — so you can judge for yourself whether it has gone stale.
- Sales of display-advertising websites fell 93% while Amazon FBA held steady
Empire Flippers sold 138 display-advertising businesses in 2022 and 9 in 2026. Amazon FBA went 81 to 60 over the same period. Every search-traffic-dependent category fell ~70-90%; every non-search category held or grew.
- RPM by niche, computed from 90 sites with verified traffic and revenue
Median RPM across 90 display-only businesses is $17.03. Home is $23.93, Finance $31.01, Information $10.82, Gaming $9.46. Larger sites earn LOWER RPM.
· corrects an earlier claim of ours - Three times in one week, comparing two endpoints hid a step change
A two-endpoint comparison cannot distinguish growth from a step discontinuity or a taxonomy change. We hit it three times: a marketplace category, an OpenStreetMap query, and a vendor statistic.
· corrects an earlier claim of ours - What AdSense actually requires, quoted from Google rather than from blogs
Google publishes no minimum post count, no minimum site age and no traffic minimum. One page is genuinely mandatory: a privacy policy.
· corrects an earlier claim of ours - The ad-network entry ladder inverted in 2026 and most guides still have it backwards
Mediavine Official now gates on $5,000 annual ad revenue rather than 50,000 sessions. Raptive cut to 25,000 pageviews. Ezoic raised to 250,000 monthly active users.
- Searching for the thing you are measuring will manufacture your result
A frame drawn by searching "X monthly rates" selects sites BECAUSE they publish rates. Drawing the same sample from OpenStreetMap — which knows nothing about prices — gave 66.7%, a figure the biased frame would have inflated to near-certainty.
- Businesses publish their prices and hide their availability, deliberately
The same 40 businesses publish monthly prices at 66.7% and live availability at 10%. 54% have no booking system at all.
- The live US tariff schedule still prints duties that were terminated months ago
The HTSUS carries the IEEPA headings 9903.01.20, .24 and .25 in full, with note text intact and no termination annotation, months after an Executive Order ended them.
- Nine market screens found nine gaps. Seven were already occupied.
Verticals found by screening: 0 for 9. Instruments built because a decision required the number: 3 for 3.
- A number is usually unpublished because it is expensive to collect
Unpublished-ness and cheap collectability are in tension. Testing publication rate before hunting incumbents kills bad candidates in two hours instead of two weeks.
- A calculator page earns $12.72 per thousand views. A content page earns $39.
Single-purpose calculator sites measure $12.72 RPM carrying 1-2 in-content ad slots. Long-form content in a good niche measures $39-43 carrying 6-12 slots.
· corrects an earlier claim of ours - AI Overviews appear on 0.9% of navigational queries and 57.9% of questions
AI Overview prevalence is 0.9% navigational, 2.1% transactional, 4.3% commercial, 13.1% branded against 24.9% non-branded. But the query forms content sites monetise — "X vs Y", "review", "best of" — measure 81-95%.
- Developers said 72% block ads. Server logs measured 32%.
The widely-cited 72.3% developer ad-block rate is self-reported. The one contemporaneous server-side measurement of the same population found 32% — a 2.3× say-do gap.
· corrects an earlier claim of ours