How we measure paywall success rates
Unlike other paywall tools, we publish how often removal actually works — per outlet, updated continuously. Here's exactly how those numbers are produced, so you can trust (and cite) them.
Right now: 5,447 articles recovered across 257 outlets with enough data, a blended 63% success rate.
Measured from real usage, not lab tests
Every figure comes from anonymous, real attempts by real people using the tool — logged server-side, so ad-blockers don't undercount them. We never contact a publisher to build these numbers. When you paste a link, the outcome of that attempt (which method delivered the article, and whether it succeeded) is recorded against the outlet's domain. Nothing that identifies you is stored with it.
What counts as a success
A success means the tool returned a readable article — either a clean reader-view extraction of the article text, or a working full-page/archived copy — rather than a block, a challenge page, or an empty shell. A result that comes back paywalled, bot-blocked, or too thin to read is counted as a failure. We don't grade ourselves on partial credit.
Two signals, blended
Each outlet's rate blends two independent signals: the passive outcome our engine records for every attempt, and the explicit 👍/👎 feedback readers leave on whether the result actually worked for them. Combining both corrects for edge cases the engine can't self-assess.
Minimum sample size
An outlet's success rate is only published once it has at least 5 recorded attempts (12 for the annual report), so a single lucky or unlucky try never sets the number. Low-sample outlets are shown without a percentage until they cross the floor.
Continuously updated
The rates refresh automatically as new attempts come in (cached briefly for speed). A publisher that tightens its wall will see its rate fall in our data within days — and one that loosens will see it rise. The numbers describe the live web, not a snapshot frozen at launch.
Honest about the limits
Hard subscription walls — where the article text is never sent to a logged-out reader — can only be recovered when a public archive of the piece already exists. That's why outlets like the FT, WSJ and Bloomberg show lower rates: the number reflects reality, not marketing. We'd rather publish an unflattering figure than a false one.
See the data
The datasets are free to reuse with attribution (CC-BY 4.0).