TL;DR
Some buyers now use ChatGPT, Google AI features, Perplexity, and other assistants during research. That does not mean every buyer starts there, or that an AI answer replaces the rest of the decision. It means your public evidence needs to be crawlable, clear, consistent, and useful wherever research begins. Nobody can guarantee a citation, recommendation, or top placement.
The behavior is changing. The promise should not outrun the evidence.
A buyer may hear your name from a peer, encounter a LinkedIn post, search your company, ask an AI assistant for options, or do all four in one afternoon.
That mixed journey matters more than any claim that “search is dead” or that every next client will ask an AI first. AI-assisted discovery is another research surface. It is not a predictable vending machine where the right metadata produces a recommendation.
OpenAI says any public website can appear in ChatGPT Search and advises publishers not to block OAI-SearchBot if they want their content included in summaries and snippets. It also says there is no way to guarantee top placement. OpenAI’s publisher guidance and ChatGPT Search documentation are refreshingly direct about that boundary.
Google draws a similar line. Its guidance says the usual Search fundamentals remain relevant for AI Overviews and AI Mode, with no additional technical requirements, special AI files, or special schema needed. Eligibility is not a promise that a page will be indexed or shown. Google’s AI features guidance is the useful baseline here.
That changes the practical question from “How do we make an AI recommend us?” to something more honest:
If a buyer or retrieval system encounters this business, is there enough clear, credible evidence to understand what it does and verify why it belongs in the conversation?
Citation, mention, and recommendation are different outcomes
A source can be cited because one passage answered a question well. A company can be mentioned because its name appears consistently across the web. A business can be recommended because the available evidence suggests it fits a buyer’s situation.
Those are not the same achievement.
Your own website can improve how clearly your business is represented. It cannot manufacture independent consensus. Recommendations are also influenced by sources you do not control: client experiences, credible third-party references, public reviews, interviews, communities, and the exact context of the buyer’s question.
This is why “AI visibility” is too small—and often too slippery—as the strategy. The stronger goal is a public body of evidence that helps a person make a decision and gives search or AI-assisted systems something accurate to retrieve.
What a business can control
1. A clear public identity
The site should make the basics unambiguous:
- who the company serves;
- what problems it is equipped to solve;
- what its people actually know;
- what it has built or changed;
- how someone can verify the claims; and
- what the next step is.
This is ordinary positioning work. It becomes more important when a system may extract one passage without the full sales conversation around it.
2. Crawlable, indexable pages
Important content should be available as text, reachable through internal links, and eligible for indexing. Standard titles, descriptions, canonical URLs, sitemaps, and robots controls matter because they help retrieval systems access the same useful pages people do.
For ChatGPT Search, allowing OAI-SearchBot is the relevant crawler decision. GPTBot is a separate control related to potential model training; the two should not be confused.
3. First-hand evidence
Generic explanations are easy to reproduce. First-hand work is not.
A live readiness assessment, a package builder, a members portal, an implementation note, a measured observation, or a case study with approved numbers gives a buyer more to evaluate than another article summarizing the same public advice.
This is where small firms can be more distinctive than larger publishers. They may not have the largest domain, but they can publish evidence no other site possesses.
4. Claims that can survive extraction
Each important claim should still make sense when read on its own. That means naming the subject, stating the boundary, citing the source when one exists, and avoiding certainty the evidence does not support.
Structured data can help describe visible content, but it must match the page. It is context, not a ranking override. Invented ai-* meta tags and keyword lists do not create authority.
5. Consistency across the places buyers check
A website, founder profile, LinkedIn presence, directory listing, and third-party mention should not tell five different stories. Consistency helps people recognize the same business across a fragmented research journey.
That does not require repeating identical copy everywhere. It requires stable facts: name, expertise, audience, offer, proof, and point of view.
What we would not sell as a promise
We would not promise that:
- a particular AI system will cite or recommend the business;
- schema markup will cause an AI Overview;
- an
llms.txtfile is a ranking signal; - one page can establish authority by itself;
- more AI-written content automatically creates visibility; or
- appearing in an answer automatically produces qualified demand.
Those claims go beyond what the platforms themselves say and beyond what any operator can control.
The operating approach
For an expertise-led business, the useful work is broader than “AI SEO”:
- Clarify the positioning and the audiences that matter.
- Build an owned, technically sound source of truth.
- Publish first-hand thinking and proof that a buyer can evaluate.
- Use distribution channels without making any one platform the foundation.
- Keep the facts, examples, and point of view current.
- Measure search, referral, and lead behavior, then adjust without pretending attribution is perfect.
At SocialTide, that is part of the same operated presence as the site, content, analytics, and custom software. AI-assisted discovery is one surface the system should support. It is not the product and it is not a guaranteed outcome.
Start with what is verifiable
The strongest next move is rarely “publish more about AI search.” It is usually to make the business itself easier to understand and substantiate.
Show the work. Name the judgment behind it. Correct stale claims. Give buyers useful tools and useful explanations. Keep the technical path open for legitimate search and retrieval crawlers. Then measure what actually reaches the business.
That foundation serves a person who arrives from a referral, a search result, a LinkedIn post, or an AI-assisted answer. It is worth building even when no platform provides a citation.