By Greg Nowak. Last updated 2026-11-01.
Should your business let Google use its content in AI Search? A service company may welcome another route to discovery. A publisher selling original research may have different priorities. The useful starting point is what the content needs to achieve, followed by evidence about whether participation supports that purpose.
Google’s control is now available worldwide. Before changing it, agree which content matters, what success would look like and who owns the decision. That makes the review manageable for business owners, operations leads and agency teams.
What Google’s AI Search toggle actually changes
Open Settings > Search generative AI in Search Console. Inclusion is the default and permits links and content in AI Overviews, AI Mode and generative AI features in Discover. Exclusion removes that eligibility, including grounding responses, and the associated impressions and traffic.
Google says the choice does not affect ranking or inclusion elsewhere in Search, override Ads or Merchant Center participation, or control AI training. Content may still support Search more broadly. See Google’s control documentation for the boundaries.
Keep related controls separate. Google-Extended manages specified training and grounding uses involving Gemini Apps and Vertex AI; it does not control Search inclusion or ranking. Google documents those uses in its crawler reference. A noindex directive has a broader purpose: removing a page from Google Search. Neither should be substituted casually for this Search Console setting.
Start with the pages that pay their way
A site-wide impression total mixes content with very different jobs. Group your important URLs by business purpose, then choose evidence that helps assess each group. You do not need to classify every archive page before starting.
| Content group | Business question | Evidence to review |
|---|---|---|
| Service and product pages | Does visibility accompany qualified demand? | Relevant landing-page visits, enquiries and sales quality |
| Educational articles | Does discovery bring potential buyers closer? | Engaged visits, onward journeys and assisted enquiries |
| Support documentation | Is surfaced information accurate and maintained? | Exposed URLs, outdated instructions and support feedback |
| Premium or licensed material | Does participation fit the commercial model? | Access policy, licensing commitments and subscription priorities |
For an agency, the client’s commercial or publishing owner should approve the policy. The agency can prepare the analysis and implement the setting. If licensing or access requirements already determine the answer, resolve those before proposing an exclusion experiment.
Read the reports without overstating the results
Google’s generative AI performance report for Search shows impressions from AI Overviews and AI Mode. It supports page, country, date and device views, plus exports. An impression records a displayed link; it does not establish a visit or enquiry.
The report excludes Search Labs experiments. The usual 1,000-row table limit applies, and page totals can differ from chart totals because of aggregation. Dates use Pacific Time. Crucially, this data is included in the ordinary Search performance report’s Web data: adding the two reports together would double-count exposure.
Discover has a separate generative AI report. Its page dimension identifies the canonical source page, which can differ from the landing page. Keep Discover analysis separate when matching reported exposure to website behaviour.
Review analytics and CRM outcomes alongside these reports. Unless your tracking reliably identifies the origin, describe a change in landing-page demand as supporting evidence. Do not calculate an AI conversion rate by dividing all organic leads by AI impressions.
Build a test someone else can review
- Write the decision question. For example: “Does participation support qualified enquiries for our service content sufficiently to justify continued inclusion?” Choose one primary business outcome and supporting measures. Agree what would justify continuing, reversing or extending the test.
- Save a usable baseline. Export exposure by page and date, ordinary Search performance, landing-page analytics and relevant CRM outcomes for the same period. Keep a fixed URL list. Check forms and conversion tracking before relying on their numbers.
- Confirm the boundary. Owners can override inherited settings for child properties, including suitable URL-prefix paths. Check parent settings and child overrides before choosing a section. Record the exact property and affected URLs.
- Log the change and recovery plan. Save the previous choice, timestamp, operator and reason. Exclusion generally takes a few days, with caching potentially delaying some content. Mark a transition window separately. Identify who can restore the previous choice if agreed commercial limits are crossed.
- Review a normal buying cycle. Schedule weekly checks and a decision date that fits your sales volume. Log campaigns, content edits, redesigns and tracking changes. If enquiries are rare, a short comparison may remain inconclusive.
A practical handover record can be as simple as property | previous setting | new setting | owner | change date | outcome measure | review date. Keep it with the exports so the next person can reconstruct the decision.
Decide what the evidence supports
A before-and-after comparison cannot isolate every cause. Seasonality, competitors and other site changes can move demand. A fall in AI impressions after exclusion mainly confirms that eligibility changed; it does not establish lost revenue. An unchanged enquiry count may also conceal effects on a longer buying journey.
Continue inclusion where the commercial case and content policy support it. Consider exclusion where access, licensing or governance priorities outweigh the opportunity. When evidence is thin, record that uncertainty and set another review date.
If your team needs help turning this into a workable review, Greg can help connect Search Console, content priorities and business outcomes, with a clear scope, decision owner and handover.
Related on GrN.dk
- Google’s AI Search Toggle Is a Publishing Decision, Not an SEO Setting
- Your AI Visibility Dashboard Needs a Methodology, Not More Charts
- AI Content on Google: Which Pages Are Worth Keeping?
Need help with this kind of work?
Plan your AI Search review with Greg Get in touch with Greg.
