Before Turning On AdSense Auto Ads: Preview, Exclusions, and Experiments

There is no universal rule that Auto ads must always be on or off. Articles, lists, hubs, and error pages have different roles and content. Use AdSense preview, exclusions, and experiments to decide from your site.

Select representative pages

Review at least:

• long and short articles;
• articles with tables, code, and images;
• Japanese and English articles;
• home, list, tag, and author pages;
• 404, empty search, empty tag, and other excluded routes.

For excluded routes, the site should omit the production ad script itself, not only hide an ad slot.

Preview checks

The Auto ads preview shows candidate positions, counts, existing units, and mobile and desktop views. Google notes that actual size and position can differ.

Page:
Does not obstruct the article opening:
Does not over-fragment headings and paragraphs:
Cannot be mistaken for navigation:
Does not overlap tables, code, or images:
No horizontal scrolling:
No material layout-shift regression:

Change one setting at a time

Auto ads settings include overlay and in-page formats, ad load, minimum spacing, excluded areas, and excluded pages. Change one factor so that the result has an interpretable cause.

Verify exclusions

Page exclusions override site Auto ads settings. Exclude 404, error, empty result, admin, preview, and content-free pages. Check whether exact-page or section-prefix scope matches the intended URLs.

Keep a mapping between application eligibility and AdSense exclusions so that mismatches can be detected.

Compare on and off

Google's Auto ads experiments can compare Auto ads settings or on versus off. Metrics describe site-wide impact including existing ad units, so record the configuration and period.

Review not only earnings but representative-page rendering, layout, speed, content access, exits, and user reports. Do not declare a winner without sufficient evidence.

Check the scope of page exclusions

In exclusion previews, verify whether an entry excludes only the exact home URL or everything beneath a path. Excluding the root and its descendants may also exclude intended articles. Excluding only a Japanese historical URL may leave its English counterpart uncovered.

Keep a table of URL, matching scope, language, and check time. Visit one core article, one archived article, search, and a 404 page. Record page exclusions separately from excluded areas within a page. Do not re-enable loading on a code-excluded page merely because an account setting looks correct.

Conclusion

Decide in the order of representative pages, preview, ad load, exclusions, and experiment. Do not infer that on guarantees more revenue or off guarantees better experience. Isolate each change and preserve site-wide evidence.