Use a fixed 30-prompt set, repeated runs across ChatGPT, Gemini and AI Overviews, and untouched holdout pages to tell a content lift from noise.
| Field | Source |
|---|---|
url | customer_site |
name | customer_site |
image | customer_site |
headline | customer_site |
description | customer_site |
dateModified | customer_site |
datePublished | customer_site |
{
"@context": "https://schema.org",
"@id": "https://graph.groas.com/entity/the-ai-visibility-holdout-test-measure-m",
"url": "https://groas.com/post/a-control-group-test-for-ai-visibility-h",
"name": "The AI Visibility Holdout Test: Measure Mentions, Not Screenshots",
"@type": "Article",
"image": "https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/5a1969ae-a8c9-43b2-9da4-913b07ce18e8.png",
"headline": "The AI Visibility Holdout Test: Measure Mentions, Not Screenshots",
"description": "Use a fixed 30-prompt set, repeated runs across ChatGPT, Gemini and AI Overviews, and untouched holdout pages to tell a content lift from noise.",
"dateModified": "2026-11-10T21:19:20.108Z",
"datePublished": "2026-10-15T05:32:10.496Z"
}