Should Multi-Location Brands Opt Out of Google AI Overviews?

Google made its Search generative AI control available worldwide on August 31, 2026. Site owners can choose whether their links and content appear in AI Overviews, AI Mode, and generative AI in Discover. For a publisher, that can become a debate about pageviews and content reuse. For a 60-location HVAC group, plumbing roll-up, med spa platform, or franchise service brand, the decision is different.

Last verified: September 10, 2026.

These businesses publish to get selected. Their location pages, service explanations, warranties, financing details, staff credentials, and local proof exist to help a buyer choose a branch. Removing that owned evidence from Google's generative answers can make the brand less inspectable at the moment a customer asks who to hire.

The default recommendation for most multi-location service brands is therefore to stay included, measure the new reporting, and document the property-level setting. Exclusion should be a narrow risk decision, not a reflexive AI policy.

Google's Search generative AI control documentation is explicit about the surfaces affected and the property inheritance model. The broader Google AI Search guide for local businesses explains the source foundation that inclusion makes available. To see which businesses and sources already appear for your markets, start with the Cheers AI Visibility Grader.

What Google's AI Search control changes

The setting appears in Search Console under Settings, then Search generative AI, with worldwide availability confirmed on August 31, 2026. The property is included by default. A verified owner can choose to exclude the website's links and content from AI Overviews, AI Mode, and generative AI in Discover.

Google says exclusion means the site will not receive traffic or impressions from those generative AI experiences while the setting applies. It also says the setting does not affect inclusion or ranking in other Google Search experiences. Those other Search experiences remain separate from this control.

That separation needs careful language. Google notes that excluded content may still inform how Search understands and generates language, even though the website's links and content cannot appear in the generative response. The setting also is not a model-training control. Google-Extended is the separate training control identified in that documentation. Search inclusion, model training, and indexing are three different decisions.

Google says changes generally take a few days to apply. Content removal can begin within one or two days after the control is live, but cached results can take longer to clear. The right operating state is therefore requested, then observed, then verified, not "we flipped the switch."

Why local service websites are not publisher websites

A publisher may evaluate an AI answer as a substitute for a pageview. A local service operator should evaluate whether the answer helps a qualified customer identify and contact the correct location.

Consider an electrician with 35 branches. A useful location page might explain generator installation, license coverage, response times, financing, the service area, and the phone number that reaches that market. If that page appears as a supporting link in an AI Overview, the customer can inspect the claim and continue to the owned conversion path. If the page is excluded, Google's answer may still rely on directories, review platforms, competitors, or other available sources.

That does not guarantee that staying included wins the recommendation. Google says its AI features use normal Search systems, and eligible pages still have to earn selection. But a service brand usually has more to lose from withdrawing its strongest owned evidence than it has to gain from withholding public service information.

The exception is content with a real exposure cost. A legal team may identify licensed material, contractual restrictions, member-only research, or a premium knowledge product that should not appear in generated answers. That is a content-governance reason. It is not a reason to remove every public location and service page from AI Search.

Property inheritance is the multi-location risk

Google applies this setting through Search Console properties. That matters because a multi-location portfolio rarely has one clean website boundary.

A domain property can cover protocols, subdomains, and paths across the whole domain. A URL-prefix property covers only URLs under the stated prefix. Google's property documentation explains those boundaries, while the generative AI control documentation says child properties inherit the parent setting unless the child overrides it.

For a roll-up or franchise, one decision can therefore affect more than the page in front of the operator. A corporate exclusion may reach location directories, acquired-brand subdomains, resource libraries, and booking paths. A child override may preserve inclusion for one section, but only if the property actually covers the canonical URLs and the right owner has access.

Before changing the setting, inventory the domain property, URL-prefix properties, canonical hosts, subdomains, acquired-brand domains, and site owners. Google says only verified owners and delegated owners can change the control. Its permissions guide is the authority for who can do what.

This setting controls the brand's website. It does not remove Google Business Profile facts, third-party reviews, directories, news coverage, or competitor pages from Google's source environment. A portfolio can withdraw its owned evidence without withdrawing the rest of the story.

Do not substitute robots or snippet controls

Several controls can look similar in a policy meeting while doing different jobs.

• Search generative AI setting: controls whether site links and content can appear in Google AI Overviews, AI Mode, and generative AI in Discover
• Google-Extended: controls model-training uses separately from Search generative AI visibility
• noindex: removes a page from Google Search when Google can crawl and process the directive
• nosnippet and max-snippet: limit how much page content Google may show in search-result snippets and can also limit use in AI features
• robots.txt: controls crawling, not guaranteed removal of an already known URL from Search

The AI crawler access guide separates search retrieval, training crawlers, and user-requested fetches across providers. Google's robots meta documentation explains noindex, nosnippet, and preview limits. Pick the control that matches the risk instead of stacking broad blocks and hoping they mean the same thing.

A decision rule for multi-location brands

Stay included when the website is public lead-generation infrastructure and the content helps a customer evaluate a real location, service, or proof point. This is the normal case for home services, franchise services, hospitality groups, and multi-location clinics.

Consider a limited exclusion when counsel or a content owner can name the protected material, the reason it should not appear in a generated answer, the smallest property boundary that contains it, and the acceptable visibility cost. A paid research library under its own URL prefix is a better candidate than the entire domain. Even then, verify that the child property and canonical URLs behave as expected.

Keep separately owned franchisee and acquired-brand domains separate unless the organization has authority to make the change. A corporate policy memo is not proof that the parent company controls every Search Console property. The subfolder versus subdomain guide explains why website architecture and operating ownership need to line up before a portfolio-level change.

The wrong decision rule is "AI is taking our content." The useful rule is "Which public evidence are we willing to remove from a local hiring answer, for what documented risk, and how will we verify the result?"

Run a reversible 14-day pilot

If the case for exclusion survives review, test the smallest useful scope. Do not begin with the domain property for a 200-location portfolio.

• Record the current control, parent and child properties, canonical URLs, verified owners, baseline AI Search impressions, classic Search performance, and conversion path.
• Change one eligible URL-prefix property with a clear rollback owner and timestamp. Capture the Search Console confirmation instead of relying on a meeting note.
• Allow several days for the setting and cached results to change. Inspect representative queries and URLs in AI Overviews and AI Mode without claiming one prompt is a complete measurement set.
• Compare the dedicated Search Generative AI report when available, overall Web performance, page-level classic Search data, AI referrals, calls, forms, and booked outcomes.
• Restore inclusion if the risk case is not validated, the property scope is broader than expected, or the visibility cost is not acceptable.

Google's Search Generative AI performance report shows impressions from AI Overviews and AI Mode by page, country, device, and date. Google says those impressions also remain inside overall Web Search data, so do not add the dedicated report to the Web total. The report is limited to 1,000 rows and aggregates pages by canonical URL.

The AI search traffic measurement guide shows how to keep visibility, citations, referral visits, and booked demand separate. An opt-out test needs all four. A drop in AI impressions is expected after exclusion. The business question is whether any protected-content benefit justifies the lost discovery and whether classic Search and conversion paths stayed intact.

What the operating record should contain

One row per tested property is enough. Record the property identifier, parent property, included or excluded state, inheritance source, owner, reason, approved scope, change time, expected application window, affected URL sample, AI Search baseline, classic Search baseline, conversion baseline, observation dates, rollback decision, and public readback.

For a med spa group, the row might cover a premium training library under one path while confirming that treatment and location pages remain included. For a restoration platform, it might document that acquired-brand domains stay unchanged because local operators own them. For a franchise system, it may reveal that the corporate team has no authority over franchisee sites and should publish guidance instead of promising a portfolio-wide switch.

That record turns an ideological argument into a reversible operating decision. It also stops a future site migration, acquisition, or permissions change from silently inheriting an old policy.

The practical recommendation

Most multi-location service brands should keep their public websites included in Google's generative AI Search experiences. Their economic goal is qualified local discovery, and their owned pages are the evidence they can make accurate, current, and conversion-ready.

Do not confuse that recommendation with unlimited permission. Use Google-Extended and provider-specific crawler controls for separate training decisions. Use noindex or snippet limits for page-level Search concerns. Use the new Search Console setting only when the organization has a documented reason to remove its links and content from Google's generative AI experiences.

Then verify the live result by property, URL, market, and customer route. The control is reversible. The visibility you give up while it is active may not be.

Sources

• Google Search Console Help: control content in generative AI experiences in Search. Supports the included-by-default state, affected surfaces, Search separation, property inheritance, owner permissions, timing, and model-training distinction.
• Google Search Console Help: Search Generative AI performance report. Supports report availability, metrics, dimensions, canonical aggregation, Web Search inclusion, row limits, and Search Labs exclusions.
• Google Search Central Blog: Search Generative AI performance reports. Supports Google's report rollout and the separation of AI Overview and AI Mode performance views.
• Google Search Central: AI features and your website. Supports Search eligibility, snippet eligibility, normal SEO foundations, and preview controls for AI features.
• Google Search Central: Google crawlers and Google-Extended. Supports the distinction between Google Search crawling and Google-Extended training and grounding controls.
• Google Search Console Help: add a website property. Supports domain and URL-prefix property boundaries.
• Google Search Console Help: owners, users, and permissions. Supports verified-owner and delegated-owner roles.
• Google Search Central: robots meta tag specifications. Supports noindex, nosnippet, and preview-control distinctions.

Dylan Allen-Arnegård is the CEO & Co-Founder of Cheers, the local search platform for multi-location service businesses.