Your website can welcome new customers while keeping a firm hand on how search products use its content. The awkward part is choosing the right switch. Pull the wrong one and a sensible content decision becomes a discovery problem.
At UltiMedia, we recommend treating AI search controls as a business decision with a technical checklist. A rushed instruction to “block AI” leaves too much room for interpretation. Which product? Which pages? What outcome?
What do AI search controls actually do?
AI search controls let website owners manage particular uses of their content in search experiences. Different settings affect different activities, including crawling, indexing, previews and generative answers. Start with the intended outcome, identify the relevant product, and check the scope before changing a setting.
Google’s Search Console now includes a Search generative AI control. Its help page says worldwide rollout completed on 31 August 2026. The setting covers AI Overviews, AI Mode and generative features in Discover. Find it under Settings, then Search generative AI. Google explains the control and its scope.
That gives South African website owners a practical question for October: does the current setting match the business decision? We would ask that before buying another report about missing citations.
Start with the commercial question
Before touching AI search controls, write down the result you want in ordinary language. “Protect our content” sounds sensible, but it does not tell a developer what to change.
A Cape Town consultancy might want buyers to discover its public service pages. A publisher might want a deliberate policy for valuable editorial material. A wholesaler might have a customer portal that should never allow anonymous access. Those are three separate briefs.
Our suggested decision note starts with a sentence: “We want these named pages to remain available through these named channels, while restricting this specific use.” If nobody can finish that sentence, leave the switches alone for now.
- Name the pages.
- Identify the business owner who can approve the decision and the person who can reverse it.
- Write down what success looks like, including the customer journey you want to preserve.
- Keep a dated copy.
AI search controls work best when everyone understands the brief. The developer should not have to guess whether marketing means fewer automated requests, less reuse of copy, or removal from search altogether.
Separate inclusion from training
The Search Console choice and model training preferences answer different questions. Google says its Search generative AI control does not affect training. It points owners to Google-Extended for that purpose. Excluding content from the covered Search features also does not set ordinary Search ranking or inclusion signals. Read Google’s explanation of exclusion.
Google’s crawler documentation describes Google-Extended as a product token for specified Gemini training and grounding uses. It also says that token does not affect inclusion or ranking in Google Search. The document carries an update date of 15 October 2026. Check the Google-Extended product description.
Our recommendation: keep separate rows in your decision record for Search participation and training preferences. Do not bundle them under one vague “AI blocked” tick box. That label hides the very distinction your team needs to remember.
With AI search controls, precision beats a dramatic announcement. Ask the person implementing the change to name the exact setting and expected effect before they touch production.
Know what the other switches change
AI search controls sit beside older website settings. Those older settings still deserve respect. They can change much more than the feature that started the conversation.
A robots.txt file manages crawler access. Google warns that it does not reliably keep a URL out of search results: a blocked URL can still appear without a description. The same documentation recommends password protection for private material. Read Google’s robots.txt limitations.
A noindex rule tells supporting search engines not to index a page. Google must crawl the page to see the rule, so blocking the crawl can prevent it from detecting noindex. Google documents how noindex works.
Preview controls have their own scope. Google’s nosnippet rule suppresses text snippets and video previews and prevents direct use of the content in AI Overviews and AI Mode. Its max-snippet rule limits preview text and the amount of content available as direct input, with documented exceptions. Check the robots metadata specifications.
We would therefore avoid applying a blanket noindex instruction to public service pages simply because someone dislikes generative summaries. First choose the intended result. Then select the narrowest suitable setting with your technical lead.
Map the property before changing anything
For AI search controls, the selected Search Console property deserves a deliberate check. Google supports inherited settings from parent properties and allows child properties to override that inheritance. Owners of a top-level domain receive a notification when a child property overrides its setting. Review Google’s inheritance examples.
In practice, we recommend drawing a small map of the domain, relevant subdomains and registered URL prefixes. Put the current choice beside each one. This is an operating worksheet, not a new software purchase.
Imagine a fictional training business with a public course catalogue and a separate editorial section. The owner wants one policy for course discovery and another for paid editorial work. Before implementation, the team should identify the relevant properties and document the intended scope.
That example describes a planning exercise. It does not imply a measured result or suggest that a Search Console preference protects paid material. Access protection needs its own technical design.
Good AI search controls documentation should let a second person answer three questions without a meeting: what changed, where it applies, and who approved it.
A practical change procedure for small teams
We suggest the following process for AI search controls. These steps are our operating recommendations, not Google’s promised recipe for better rankings.
1. Save the starting position
Record the selected property, current setting, inherited setting where relevant, date and responsible owner. Take a screenshot for the internal change record. Keep the exact property address visible so a colleague can identify the scope later.
Also save a short list of commercially important pages. For a South African service business, that might include the main service description, contact page and location information. Choose pages because customers need them, not because a template contains five empty rows.
2. Check access and ownership
Choose a named person to make the approved change and a second person to check the result. We recommend keeping business ownership of accounts clear, especially when agencies or contractors help maintain the site.
Ask for an explanation that the owner can repeat: “This setting changes this use of these pages.” If the explanation needs twelve acronyms, simplify the brief before moving forward.
3. Change one relevant setting
Keep AI search controls changes separate from a redesign, a new SEO plugin configuration or a large content rewrite where possible. This makes later investigation less messy. If several changes must happen together, list each one and its intended effect.
Record the value before and after the change. Do not rely on memory or a message saying “sorted”. That word has caused enough trouble in website maintenance to deserve its own warning label.
4. Verify and schedule a review
Reopen the setting and confirm the selected property and saved choice. Ask the checker to compare them against the approved note. If they differ, investigate before making more changes.
Google says exclusions generally take a few days and notes that caching and propagation can extend the process. Avoid promising an instant result. See Google’s timing guidance.
For AI search controls, a scheduled review should examine implementation and business impact separately. First establish whether the intended setting remains in place. Then examine the available evidence about discovery and enquiries.
Measure the decision without inventing a victory
AI search controls do not turn a before-and-after chart into proof. A change in enquiries can coincide with a promotion, seasonal demand, new competitors or a broken form. Write down those possible explanations.
Google describes Search Console as a tool for understanding how a site performs on Search and for identifying indexing and technical issues. Use the relevant reports alongside your own business records. Google’s Search Console introduction explains the available monitoring tasks.
We recommend a simple review sheet with the observation period, affected pages, available search measures, qualified enquiries and known changes elsewhere. Label missing data as missing. A blank cell tells the truth more clearly than an invented estimate.
For example, a fictional installer could record whether enquiries concern its actual service area and whether the customer requests work it offers. Those questions help the owner evaluate lead usefulness without pretending that every form submission has equal value.
Keep AI search controls decisions reversible in your operating process. Specify who can restore the previous choice and what evidence would justify doing so. Reversal is a decision too; give it a reason.
For each review, keep the original decision beside the evidence. That simple habit stops AI search controls discussions drifting into personal preferences. If the evidence remains weak, say so, keep the uncertainty visible, and choose a sensible date to examine the decision again carefully.
Common mistakes worth avoiding
The biggest mistake with AI search controls is using a narrow technical setting to solve a poorly defined business problem. A switch cannot repair an unclear offer, inaccurate opening hours or a contact form that refuses submissions.
- Do not confuse discovery with confidentiality.
- Do not assume a Google setting governs every other search or AI provider. Check each provider’s own documentation before extending the policy.
- Avoid changing unrelated crawler rules during the same task.
- Keep agency handovers specific: property, setting, reason, date, owner and review plan.
Our suggested maintenance habit is modest: revisit the record after a domain change, agency handover or major site restructure. You want the current owner to understand the choice without digging through somebody else’s old inbox.
Frequently asked questions
Should every business exclude its website?
No universal choice fits every business. We recommend deciding which content uses support your commercial goals, identifying the affected pages, and documenting the tradeoff. Review the implementation and available business evidence later. A strong opinion about AI does not replace a clear website policy.
Can AI search controls protect a private customer portal?
Use authentication and appropriate access restrictions for private customer information. Search preferences govern particular discovery or content uses; they do not replace a security boundary. Ask your technical team to test anonymous access directly and document which people should have permission to view each resource.
How should a small team manage AI search controls?
Give one person responsibility for the decision record and another responsibility for checking changes. Save the property, previous value, new value, reason and review date. Keep related technical changes separate where practical, and record the business evidence you will use when reviewing the choice later.
Make the policy understandable before making it live
At UltiMedia, our view is simple: AI search controls should express an informed business choice. They should never become a guessing game between the owner, the marketer and the developer.
UltiMedia is a Cape Town-based AI-first digital agency offering Generative Engine Optimisation and web development alongside AI automation. If your team needs help connecting content policy to discoverability, ask us to review your AI search inclusion setup. Bring the property list, your current settings and the customer journey you want to preserve.