The original VisibilityOS study found that brand presence changed in 82 of 817 matched comparisons (10.0%). Using the same public counts, the pooled rate ranges from 8.3% to 11.1% when we leave out one engine view at a time.
Each entry shows the remaining changes, completed comparisons and pooled rate.
Grok: 61 of 736 (8.3%)
Gemini web: 67 of 711 (9.4%)
Claude: 68 of 710 (9.6%)
Gemini: 73 of 710 (10.3%)
Google AI Mode: 75 of 722 (10.4%)
Perplexity: 75 of 710 (10.6%)
ChatGPT: 76 of 710 (10.7%)
ChatGPT web: 79 of 710 (11.1%)
Use the eight rows in the published study. For each row, subtract its presence changes from 82 and its comparison count from 817, then divide the remaining changes by the remaining comparisons. For example, leaving out Grok gives (82 minus 21) divided by (817 minus 81), or 61 of 736.
The pooled result gives each comparison equal weight. To give each engine view equal weight, divide the changes by comparisons within each row first, then average the eight rates. That gives 10.5% for this sample.
The range shows how sensitive this sample’s pooled rate is to the views included. Each calculation uses the same observed comparisons. The study tracked PodPitch’s own sites between October 19 and November 17, 2026 UTC. Report the views, dates and comparison counts alongside any combined rate.
Open the public study notebook · Calculate coverage for your audit