Incogniton’s 90-day Facebook ban rate was 1.2%. Octo Browser’s was 1.6%. Multilogin’s was 1.9%. The ordering is consistent with the overall ranking, but the gap is smaller than the decimals suggest.
The sample is 400 profiles per platform
At 400 Facebook profiles, 1.2% is about 5 bans. A 95% Wilson interval for 5/400 runs roughly from 0.4% to 2.8%. For 6/400 (1.6%), the interval is roughly 0.6% to 3.4%. The intervals overlap heavily.
That does not make the ranking meaningless. It means the top-three ban-rate differences are directional. A team choosing between Incogniton, Octo, and Multilogin should not treat a 0.4-point difference as a guarantee.
What the ban rate is not
- It is not a causal estimate. We did not randomize accounts to browsers.
- It is not a prediction. Platform policy, proxy reputation, and behavior change.
- It is not the same as survival. Survival also counts permanent checkpoints and other losses.
- It is not a pooled count. The combined rate is the unweighted average of the Facebook and Instagram rates.
What would make the comparison stronger
A larger sample, more proxy countries, a longer window, and repeated cohorts. A vendor that wants to claim a ban-rate advantage should publish the sample size, the platform split, the proxy mix, and the status definitions. Without those, a percentage is marketing.
Practical read
Use the ban rate to screen, not to decide. Then run a pilot with your own proxies, warm-up process, and account mix. The methodology explains the status definitions, and the profile-level data lets you inspect the counts.