To improve your chances of being recommended by ChatGPT and AI search, make your business easy to find, understand and verify. Publish specific service information, keep public business details accurate and allow the relevant search crawlers. No change guarantees a recommendation; test realistic customer questions and check which sources support each answer.
A mention is only useful if it describes your business correctly to a suitable customer. A campsite recommended for large motorhomes when its entrance cannot accommodate them has gained the wrong attention. Aim for accurate recommendations that lead to qualified enquiries, rather than collecting screenshots of your name.
Start with the recommendation you deserve to receive
Choose one service and one customer need. “Recommend a tour operator” is too broad to guide useful work. “Find a small-group walking tour with a clear pace description and a stated cancellation policy” tells you which evidence a customer needs before booking.
Write five questions that real prospects ask before buying. Use enquiries, reception notes and sales conversations. Remove customer names and private details. Include one question that should rule you out, such as a request for a service you do not offer. Correct exclusion is a good result.
For an illustrative property maintenance firm, the questions might cover planned gutter clearing, photographs needed for a quote, access requirements, emergency availability and roof repairs. If the firm only provides planned clearing, a good answer should not recommend it for emergency roof repairs.
Do not begin by asking an assistant why it refuses to recommend you. Its explanation may be a plausible guess rather than evidence of how the earlier answer was assembled. Inspect the answer's linked sources and compare the actual pages instead.
Record the question, date, tool, whether search was used, the named businesses, the linked sources and any incorrect claims. Keep the wording unchanged for later checks. This small record will be more useful than a different optimistic prompt every morning.
A tour operator repairs two pages that answer booking questions
Consider an illustrative three-person tour operator with six main website pages. Its most important walking tour page says “unforgettable experiences for everyone”. Customers keep asking about pace, group size and wet weather. Those facts are known inside the business but missing from the page.
The owner sets aside four hours. They allow 45 minutes to collect approved facts, 75 minutes to rewrite the two priority pages, 45 minutes to check access and links with their website helper, 30 minutes to inspect other public listings and 45 minutes for review and baseline searches. These are planning allowances, not a promised delivery time.
At an assumed internal time value of $30 an hour, the exercise uses $120 of staff time. Any website supplier fee is extra and needs a quote. There is no need to buy an AI visibility subscription for this first pass; the immediate task is correcting information the business already knows.
| Customer question | Evidence added to the tour page | Who confirms it |
|---|---|---|
| How big is the group? | Maximum 8 guests on the standard departure | Operations owner |
| How demanding is the walk? | 4 km, about 2 hours, with uneven ground and steps | Lead guide |
| What happens in wet weather? | Published decision and notification process, matching booking terms | Bookings owner |
| Can a private group book? | Request process and what details to supply | Bookings owner |
| Is access step-free? | No; the route contains steps, with further details available before booking | Lead guide |
The owner also removes “suitable for everyone”. That phrase contradicted the route. The page now helps someone make a decision even if no AI service ever cites it. This is the strongest reason to do the work: better source information also improves direct enquiries and ordinary website visits.
For the baseline, the owner runs five fixed questions in two search services, giving ten observations. The business appears in two, one with an incorrect claim about access. Four weeks later, a repeat produces four mentions without that particular error. These illustrative numbers show how to record change, not how much improvement to expect.
Ten observations are far too few to establish a stable recommendation rate. The owner records “four mentions in this check” and continues monitoring. They do not report a doubling of market visibility or assume that the page edits caused every difference.
Check that search services can reach your public pages
A crawler is software that visits pages to discover their content. OpenAI distinguishes OAI-SearchBot, used for ChatGPT search, from GPTBot, used for model training. Its documentation says pages that opt out of OAI-SearchBot do not appear in search answers, though navigational links can still appear. Allowing training is a separate choice.
Ask your website supplier to check both crawler instructions and the hosting security rules. A page that works in your browser may still be blocked for a search crawler. Use the official OpenAI crawler documentation for the relevant names and access details.
Do not paste a generic “allow everything” file over your existing settings. Give the supplier a specific request: “Check whether our public service pages are accessible to OAI-SearchBot. Preserve protection for private customer pages and explain any proposed changes.” This keeps the task tied to the intended pages.
For Google's AI features, the normal search foundations still apply. Google says supporting pages need to be indexed and eligible for a search snippet. Its guidance does not require special AI markup or an AI text file. A site can meet the requirements and still not be selected.
If you do not know whether important pages are indexed, begin with the afternoon SEO audit. Fix a blocked service page before commissioning twenty more pages. “Published” and “available to a search engine” are different checks.
Give each important claim a visible answer and an owner
Build pages around decisions, not around repeated variations of “best”. A useful service page says what you do, who it suits, the limits, the process, how pricing is determined and what to do next. Add dates only when they help someone judge whether a fact is current.
Here is an illustrative guest house rewrite. Before: “Enjoy exceptional comfort, generous facilities and a perfect stay.” After: “We have six guest rooms. Two accept dogs by prior arrangement. Breakfast is served from 07:30 to 09:00. Tell us about arrival requirements before booking so we can confirm what is possible.”
The after version is not a formula that forces a citation. It simply gives a customer, a search service or a booking assistant something specific to assess. The owner must verify every detail. If breakfast times vary, the page needs the actual rule rather than an AI-generated compromise.
Use AI to spot missing answers in a draft. Do not ask it to fill the gaps from imagination. Paste approved public information and keep the task small enough that you can check its output.
Review this public service description for customer decisions.
Use only the supplied facts. Do not infer prices, credentials or availability.
List:
1. Questions the text answers clearly.
2. Questions a customer would still need to ask.
3. Claims that sound broader than the evidence.
4. A revised paragraph using only confirmed facts.
Facts: planned gutter clearing; single-storey buildings only;
quote after photographs; appointments agreed after assessment;
no emergency repairs.
Draft: Complete roof care whenever you need it.
An illustrative output might flag “roof care” and “whenever” as unsupported, then suggest: “We provide planned gutter clearing for single-storey buildings. Send photographs for an assessment and quote. Appointments are agreed after assessment; emergency repairs are not offered.” That preserves the limits.
What still needs fixing is the action: where should someone send the photographs? Add a real enquiry route after checking it. AI can identify an incomplete instruction, but the owner must supply the working process.
Make independent descriptions agree with the business
Inspect the public places customers already use to assess you: your business profile, relevant booking listings, partner pages and professional directories. Check the trading name, service description, contact route and current offering. Focus on sources that are useful in their own right.
For an illustrative letting agency, the website says it provides full management and tenant-finding services, while an old directory says “property sales only”. Correct the directory through its normal process and record the request. Do not create dozens of new listings while leaving the prominent wrong one untouched.
Where a partner describes your work, provide a factual update they can review. A tour operator might send an activity partner the new group-size limit and booking page. The partner decides what to publish. Do not ask for a fabricated endorsement or present a paid placement as an independent recommendation.
This is practical source maintenance, not a claim about a secret ranking factor. You cannot control which sources an AI search service will use. You can reduce contradictions in the material a customer might encounter.
Also resist artificial customer reviews. Ask actual customers for honest feedback through the normal process, without scripting their experience. A made-up review creates misleading evidence and does not become acceptable because an AI assistant wrote it.
Test fit, factual accuracy and the route to an enquiry
Repeat your small question set at a sensible interval, such as monthly. Keep a record of the service and search mode because answers can vary. Separate three checks: whether the business appears, whether the description is true and whether the suggested next action works.
- Fit: would you willingly serve this customer under the stated conditions?
- Accuracy: are the service, restrictions and contact details right?
- Evidence: does the cited page actually support the statement?
- Action: can the person enquire or book from the destination?
Suppose an illustrative campsite appears for “quiet camping with electric hook-ups”. The answer links to an old page saying all pitches have power. In reality, only 12 of 30 do. Correct the source page and any prominent conflicting listings. Save the original error so a later reviewer knows what was changed.
Do not keep prompting until the answer repeats your preferred wording and call that success. The test should resemble a customer's question. A prompt containing your business name, your selling points and an instruction to recommend you measures something else.
Commercial measurement needs a separate record. Ask new enquiries how they first found you, and keep that answer alongside any recorded website source. Use the AI search customer tracking tutorial to distinguish a recommendation, a visit, an enquiry and a completed booking.
Build a comparison answer without inventing superiority
Customers often need help choosing between two kinds of service. You can answer that question without claiming that your business is always the best option. Explain the trade-offs you know, identify the limits of your own offer and let the reader decide whether it fits.
An illustrative letting agency could answer “tenant-finding or full management?” using its approved service scope. Tenant-finding includes marketing, arranging viewings and the agreed setup work. Full management includes the specific ongoing tasks listed in the agency's contract. The page should name those tasks rather than describing full management as “everything taken care of”.
The owner asks AI to make a two-column comparison from the approved service list. An illustrative draft puts emergency maintenance cover in the full-management column. The service list only says maintenance coordination during office hours. Correct the row to that wording and ask the operations owner to confirm how out-of-hours requests are handled.
The useful comparison now has an honest boundary. It can help a customer decide, and it gives any service summarising the page a more precise source. It does not need a claim that competitors are slower, less qualified or more expensive.
Use the same discipline for “best for” wording. A campsite might reasonably explain that its quiet hours and pitch layout suit customers seeking a quieter stay. It should also disclose relevant limits, such as no group bookings, if that is the actual rule. Do not claim to be the quietest campsite without evidence.
Keep a small evidence register for statements that appear in several places. For an illustrative estate agency, one entry could read: “Accompanied viewings by appointment; confirmed by sales manager; appears on sales page and business directory; review when staffing changes.” Another could record the appraisal process and the page that explains it.
When the operating rule changes, the register tells you which public descriptions need attention. Without it, an owner may fix the website while an old partner page continues to carry the previous promise. AI can help compare the supplied descriptions, but someone must check the original sources and request the corrections.
A practical completion check is to ask a colleague unfamiliar with the service to read the revised page. Give them three customer questions and ask them to point to the answers. If they need to infer an important restriction, improve the wording. This tests the source you control without depending on whether an AI search service happens to cite it that day.
Spend on missing evidence before buying visibility promises
A sensible first month has a small scope: five customer questions, two improved pages, a crawler-access check and corrections to the most relevant public listings. Assign one owner to maintain the facts. Put volatile details, such as availability, into the process that already manages them.
If outside help is needed, request concrete deliverables: pages reviewed, changes proposed, source errors found and a method for measuring enquiries. Ask which claims are documented product behaviour and which are the supplier's working hypotheses.
Be cautious about promises to guarantee ChatGPT recommendations, mandatory special files or a fixed number of mentions. A supplier should be able to explain the work without suggesting they control the answer. The tutorial on choosing an SEO agency that understands AI search gives a fuller buying process.
Keep the work that makes your business easier to assess: clear facts, honest limits, accessible pages and a functioning enquiry route. Review actual customer outcomes before expanding the project. A small number of suitable enquiries is more useful than frequent recommendations for a service you cannot provide.
Further reads
- How to Write Your Website Copy With AI: Home, About, and Services — Turn your business facts into clearer service pages.
- How to Optimise Your Google Business Profile With AI — Keep your public business details accurate and consistent.
- What AI Can and Can't Do for Your Local SEO — Understand which search improvements AI can actually help with.
- How to Fact-Check AI Marketing Copy Before It Goes Live — Catch invented promises before they become public sources.
- How Independent Hotels Win More Direct Bookings With AI — Five places direct bookings leak to online travel agents, what AI can fix at each, and the commission sum that tells you what it's worth.
- How to Get Your Products Recommended in ChatGPT Shopping — How ChatGPT picks products in 2026, how your data reaches it from Shopify, Etsy or your own site, and the fixes that make a recommendation more likely.
- How Wedding Businesses Get Found When Couples Ask AI for Suppliers — How AI assistants build supplier shortlists for couples, a test to see what they say about you, and the five fixes that get a wedding business named.
- How to Write Website Articles With AI That Still Rank on Google — A step-by-step method for AI-assisted website articles that rank: evidence packs, outline and draft prompts, the edit pass, and what to check in Search Console.
- How to Do Keyword Research With AI for a Small Business Website — Brainstorm and group search phrases with AI, check them against Google's free data, and build a keyword map for each page, with a bookshop worked through.
- How Small Businesses Actually Use AI in Marketing: 15 Examples — Fifteen specific marketing jobs small businesses hand to AI, each with the steps, a sample, the time it takes and the catch to watch for.
- How AI Overviews Are Changing Local Search for Small Businesses — How Google's AI answers now sit around the map pack, what they read about your business, and a 12-search visibility check you can run in an hour.
- What Is Generative Engine Optimisation and Does Your Business Need It? — Where the GEO idea came from, how AI answer engines choose sources, a decision table by business type, and five checks that cover most of it.
- What to Do When ChatGPT Gets Facts About Your Business Wrong — How to find and fix the listings, pages and directories behind a wrong ChatGPT answer about your hours, prices or services, and check it stays fixed.
- AI Website Builder or Web Designer: What Each Really Costs — Builder prices for Wix, Framer, Durable and Hostinger, what designers charge according to Clutch, and a three-year cost that counts your own hours.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: OpenAI Overview of OpenAI Crawlers; Google Search Central, AI features and your website. Checked 28 September 2026.