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.
- How it was measured
- Traced the 72.3% figure to the raw Stack Overflow 2018 results.csv (it is not on the published report page). Compared against Read the Docs' own server-side measurement from the same year, then against four independent vertical measurements.
- Evidence current as of
- This corrects us
- We treated 72.3% as usable for months, and then replaced it with "plausibly above 33%". Both were wrong.
If you have read anything about monetising a technical audience, you have met the figure: 72.3% of developers run ad blockers. It is used to argue that developer-facing sites cannot be funded by advertising.
It comes from the Stack Overflow 2018 developer survey. Two things about it are worth knowing.
First, it is not on the published report. Neither 72.3% nor its companion statistic appears on the 2018 results page. Both exist only in the raw dataset. Anyone citing "the Stack Overflow survey says" from the report is citing something that is not there.
Second, it is self-reported. Read the Docs measured the same population the same year, server-side, and published the result: "32% of Read the Docs visitors run an ad blocker."
That is a 2.3× gap between what developers say and what they do, and it explains why the survey figure never replicated anywhere.
Every measurement since lands in the lower band
Mediavine's Technology & Computing category: 10–30%. Publisher Collective: 32%. NitroPay: 25%. 9GAG: 33%.
Measured technical-audience ad-blocking clusters at 25–33% — at or below eyeo's 2026 figure of 33% for the US general population.
The figure cannot be refreshed
Stack Overflow asked once, in 2018, and never again — we checked the 2019, 2024 and 2025 surveys individually. Blockthrough's public series ended in 2023. eyeo's 2026 report has audience personas but no occupational cut, and its own methodology says content-category data was obtained and simply not published.
So there is no current developer-specific number anywhere. That is not "we could not find it" — it does not exist.
Our own correction
Having retired 72.3%, we replaced it with a direction: developer audiences skew US, desktop and highly educated, so a rate "materially above 33% is plausible and unmeasured."
That was also wrong. Where it has been measured, it is not above. We were still reaching for the shape of the original claim after discarding its evidence.
One caution on eyeo's personas, which are increasingly quoted: "Tech suspicious 2.04" is a GWI index value, meaning 2.04× likelihood of falling in that segment. It is not a percentage, and converting it to one is fabrication.
Tracked in our ledger as ADBLOCK-MEASURED-32, ADBLOCK-UNKNOWN. If a number here is wrong, tell us and we will correct it in place and say what changed.