Skip to main content
Zhihe Growth 中文
Platform updates · Google AI Search

Google Tests Generative AI Search Controls and Reports: How Should Brands Decide Whether to Participate?

The announcement cited by the original article described a limited test of website-owner controls for generative search features, with page, country and impression insights. It offered more visibility into participation decisions, but opting out could also remove traffic and visibility from those AI features.

Author: Zhihe Growth Research Center Original publication: Date of event/observation: Sources checked: 2026-07-22

Key Takeaway

Do not change participation settings without a baseline and an authorized decision. Check availability for the property, account and region; record impressions, pages, countries, clicks and business outcomes; then involve growth, content, legal and management owners. Google's announcement states that the control is not a ranking signal outside the generative features, while opting out removes eligibility for the corresponding AI visibility and traffic.

What's the official change?

Google's cited announcement described a Search Console control for participation in AI Overviews, AI Mode and AI Overviews in Discover, including grounding answers. The initial test involved selected UK website owners, not every account.

The announcement also described insights into generative-search impressions, participating pages and countries. Google stated that opting out would not be a ranking signal in ordinary results outside those features, but would remove the site's corresponding AI-feature traffic and visibility.

Source verification 2026-07-22
Initial test scope Selected UK website owners
Reporting scope Impressions, participating pages and countries

Implications for International Brands

The announced control offers a more explicit participation decision, but robots, snippet controls and Google-Extended cover different products and uses. Map each setting to search display, training or other product behavior; no single control governs every AI use.

A page appearing in an AI answer is observable, but does not prove first-position brand placement, factual accuracy or conversions. Combine impressions, pages, countries, clicks, mentions, actual citations and business leads as separate measures.

Country-level analysis matters for international teams. One English page may appear in the UK but not another market because of product availability, relevance, feature scope or competition. Investigate market questions and evidence rather than averaging global observations into an unsupported citation rate.

Zhihe Growth's Assessment

Zhihe Growth treats GEO as an intentional decision about which public facts should be discoverable, not a requirement to expose all content. Public products, services, evidence and FAQs may be appropriate for discovery; paid research, customer data and internal assets need access controls.

Prepare decision processes and baseline records before availability expands. Preserve initial settings and a comparison period. Where the control's actual scope permits, use suitable comparison groups; do not assume a property-wide control can be applied to individual holdout pages. Annotate platform changes in measurement records.

Implementation Checklist

  1. 1. Check whether the relevant control and reports are available for the Search Console property, region and account.
  2. 2. Export baseline impressions, pages, countries, clicks and conversions before changing settings.
  3. 3. Classify public and restricted content and identify the authorization owner for each category.
  4. 4. Record generative-AI impressions separately from ordinary search, Bing citations and manual AI-platform tests.
  5. 5. For experiments, use comparable holdouts only where the setting's supported scope allows them; keep other conditions stable.
  6. 6. Record each setting, date, operator and product scope to reduce attribution errors.

Limitations

The original June announcement described a limited UK test. Google's linked article now includes an August 31, 2026 note stating that these features had rolled out worldwide. This page preserves the original test context, not a current UK-only restriction. Verify the current interface and reporting definitions; do not use unofficial methods to simulate settings.

Reports cover only their supported generative-AI surfaces, not every Google product or other platform. Validate impressions, citations and commercial outcomes with multiple evidence sources.

Official sources