How AI decides which businesses to recommend is not based on one published formula. ChatGPT, Gemini, Claude and Perplexity all use different systems, but the same question keeps coming up: why does one business appear in an AI-generated shortlist while another does not?
There is no public master formula that tells us exactly how each platform weighs every signal.
OpenAI says ChatGPT Search uses multiple factors to help users find relevant, reliable information, but it does not publish the weighting of those factors and does not guarantee placement.
Google takes a similarly cautious position. Its guidance for AI Overviews and AI Mode says there are no special optimisation requirements beyond the fundamentals of Search, and meeting those requirements still does not guarantee that a page will be shown.
So this is not a post claiming:
Do these six things and ChatGPT will recommend you.
It is a map of the six areas we repeatedly investigate when a business ranks reasonably well in traditional search but disappears when the buying journey moves into AI.
The six are:
- Consistent business details everywhere
- Content structured so it can be lifted
- Third-party mentions and earned coverage
- Review signals
- Crawlability and rendering
- Entity clarity
Some of these are supported directly by platform documentation.
Others are observed signals and practical diagnostic areas, not confirmed AI ranking factors.
That distinction matters.
First, what do we actually know?
Before getting into the six areas, it is worth separating evidence from assumption.
OpenAI confirms that ChatGPT Search considers multiple factors intended to return relevant and reliable information. It also confirms that a website needs to allow OAI-SearchBot if it wants its content to be eligible for inclusion.
Read OpenAI’s official ChatGPT Search guidance
Google says its AI search features build on existing Search systems. It specifically advises website owners to make important content available in textual form, keep structured data aligned with visible content, ensure crawling is allowed and keep Business Profile information current.
Read Google’s guidance for AI features in Search
Neither source publishes a weighting such as:
Reviews = 20%.
Schema = 15%.
Third-party mentions = 30%.
So we should not invent one.
The useful approach is to treat these six areas as levers to investigate and strengthen, then measure whether visibility changes.
That is the same principle behind the wider Ranked Not Recommended campaign.
1. Consistent business details everywhere
What it is
An AI system needs to work out whether different references across the web are talking about the same organisation.
If your website says:
Lobo Example Ltd
but another profile says:
Lobo Example Marketing
and a directory lists a different phone number, address or service category, the web contains conflicting evidence.
The practical goal is simple:
Make the core facts about the business agree wherever they appear.
That includes things such as:
- business name
- website
- phone number
- address or service area
- core services
- categories
- opening information
- leadership details where relevant
Google explicitly advises businesses to keep their information complete, accurate and current, and says complete business information helps it understand and match businesses to relevant searches.
Google also gathers business information from multiple places, including official websites, third-party data and user contributions.
That does not prove that every AI assistant uses business-detail consistency as a direct ranking factor.
It does demonstrate why conflicting information can create an entity-understanding problem.
Why it may move the answer
If one source says you serve Manchester and another says London, which version should the system trust?
If your website has one company name and authoritative profiles use another, the system has more reconciliation work to do.
Consistency reduces ambiguity.
One thing to do this week
Search your company name and compare the first ten meaningful results.
Check the same basic facts across all of them.
Deeper campaign article: Business Details Must Match Checklist
Use the dedicated article once live rather than expanding this section into a full NAP-consistency guide.
2. Content structured so it can be lifted
What it is
A page can contain the answer and still make that answer difficult to retrieve.
For example, imagine a buyer asks:
Does this company provide commercial office cleaning in central London?
One page answers immediately:
Yes. We provide daily and periodic commercial office cleaning across central London.
Another begins:
Since our founding, we have believed that every building deserves to reach its full potential…
The second version may eventually answer the question.
The first one is easier to extract.
This is what we mean by content structured so it can be lifted.
Clear question.
Clear answer.
Supporting detail underneath.
Why it may move the answer
Generative systems often need compact factual passages to construct answers.
Google’s guidance for AI Search does not tell businesses to write for an AI template. It does, however, recommend making important content available in textual form and maintaining normal helpful-content fundamentals.
Structured data can also help machines understand what a page contains. Google says structured data helps it understand page content, while Organization markup can help disambiguate administrative and identity details.
Again, that is not evidence that adding schema gives you a fixed AI recommendation boost.
It is evidence that clarity and machine-readable structure help systems understand content.
One thing to do this week
Take your five most commercially important service pages.
For each page, ask:
Can someone understand what this service is, who it is for and where it is available within the first few paragraphs?
If not, rewrite the opening.
Deeper campaign article: Schema Types That Matter for AI Search
3. Third-party mentions and earned coverage
What it is
Your website tells systems what you say about yourself.
The rest of the web tells them what other people say about you.
That external footprint can include:
- industry publications
- trade associations
- directories
- partner websites
- interviews
- conference pages
- local press
- independent comparisons
- genuine editorial mentions
This is where digital PR and GEO begin to overlap.
Why it may move the answer
Search-enabled AI assistants can use external web sources when constructing an answer.
ChatGPT Search, for example, can query external search providers and return responses containing cited web sources.
But OpenAI does not publish a rule saying:
Three press mentions increase your recommendation score by X.
So third-party coverage should be treated as a practical authority and corroboration signal, not a confirmed mathematical ranking factor.
The commercial logic is still strong.
If ten independent sources associate your company with a particular service and location, the web contains more corroborating evidence than if the only page making that claim is your own homepage.
One thing to do this week
Search:
“[your brand] + [your main service]”
Then remove your own domain mentally.
What remains?
If the answer is almost nothing, your external evidence is thin.
4. Review signals
What it is
Reviews create independent evidence about:
- whether customers have actually used the business
- what services they associate with it
- the language customers use to describe the experience
- location relevance
- perceived quality
- repeated strengths and weaknesses
For local and service businesses in particular, that makes reviews difficult to ignore.
Why it may move the answer
We need to be careful here.
There is no public OpenAI document saying that Google review count is a direct ChatGPT recommendation factor.
That claim would go too far.
What is confirmed is that reviews are important signals within Google’s own local ecosystem.
Google says local prominence is informed by information such as links and reviews, and that more reviews and positive ratings can help local ranking.
Google also gathers business information from web sources and third parties, including reviews.
For GEO methodology, reviews are therefore something worth examining because they add independent descriptive evidence about the business.
That is an observed and commercially useful signal, not proof of a universal AI ranking rule.
One thing to do this week
Read your last 20 genuine reviews.
Ignore the star rating for a moment.
Ask:
Do customers actually mention the services and outcomes we want to be known for?
If all your reviews simply say “great company”, the evidence is much less descriptive than reviews that naturally mention the work performed.
Do not script reviews or incentivise specific wording.
5. Crawlability and rendering
What it is
This is the least speculative lever on the list.
Before a search crawler can use your content, it needs to be able to access it.
OpenAI says websites should allow OAI-SearchBot and permit traffic from OpenAI’s published crawler IP ranges if they want to be eligible for ChatGPT Search.
Google similarly states that pages need to be crawlable and eligible for Search before they can appear as supporting links in AI Overviews and AI Mode.
Why it may move the answer
If the crawler cannot retrieve the content, the rest of your optimisation does not matter much.
Problems can include:
- robots.txt rules
- firewall blocks
- CDN configuration
- 403 responses
- rate limits
- JavaScript-dependent content
- inaccessible HTML
- login requirements
This is why crawlability belongs in every AI visibility investigation.
Not because it guarantees a recommendation.
Because it determines whether the system has a reasonable opportunity to access the source.
One thing to do this week
Take one important service page.
Check:
- robots.txt
- page source
- JavaScript dependency
- server response
- firewall or CDN rules
Then repeat for your homepage and highest-value location page.
6. Entity clarity
What it is
Entity clarity means making it obvious that:
this company, this website, this service, this location and these external profiles all belong to the same real organisation.
It overlaps with consistent business information, but they are not identical.
Consistency is:
Do the facts match?
Entity clarity is:
Can the system confidently understand what this organisation actually is?
That includes relationships between:
- the legal or trading name
- website
- founder or leadership
- services
- sector
- location
- social and review profiles
- associated organisations
- external mentions
Why it may move the answer
Google’s Organization structured-data documentation explicitly says the markup can help Google understand administrative details and disambiguate an organisation in Search. It also supports sameAs properties linking to profiles elsewhere on the web.
That is confirmed for Google Search.
We should not automatically translate it into:
Organization schema makes ChatGPT recommend you.
There is no evidence for that.
But entity disambiguation is still a sensible methodology area because recommendation systems cannot confidently name a business they do not clearly understand.
One thing to do this week
Look at your homepage, About page and organisation schema together.
Could an outside system clearly answer:
- What is this company called?
- What does it do?
- Where does it operate?
- Who is behind it?
- Which external profiles belong to it?
If not, start there.
For the broader methodology, see Lobo Digital’s GEO and AEO services.
The six levers at a glance
| Lever | What we are trying to improve |
|---|---|
| Consistent business details everywhere | Reduce conflicting information |
| Content structured so it can be lifted | Make useful answers easier to extract |
| Third-party mentions and earned coverage | Increase independent corroboration |
| Review signals | Strengthen real customer evidence |
| Crawlability and rendering | Make important content accessible |
| Entity clarity | Make the organisation easier to identify and understand |
The key point is that these do not operate independently.
A business can have technically perfect schema but no external authority.
It can have hundreds of reviews but a website AI crawlers cannot access properly.
It can have excellent content but inconsistent business details across the web.
The methodology is designed to find which gap is holding the business back, rather than telling every client to do the same thing.
Which lever should you work on first?
That depends on what the evidence says.
If OAI-SearchBot receives a 403, fixing reviews first makes little sense.
If the site is technically perfect but the business has almost no external footprint, another schema plugin probably will not solve the real problem.
If an assistant repeatedly describes the company incorrectly, entity clarity may deserve priority.
If competitors are consistently named and have materially stronger independent reviews and mentions, authority is likely the more useful area to investigate.
This is why Lobo Digital starts with measurement rather than selling the same optimisation package to everyone.
The free AI Visibility Checker provides the quick first view.
The deeper AI Visibility Audit is designed to show which questions you disappear from, who appears instead and which of these areas deserves investigation.
Frequently Asked Questions About AI Business Recommendations
Common questions about AI recommendation factors, GEO timelines, reviews, schema, paid advertising and what may influence whether an AI assistant names your business.
Which of the six factors matters most?
There is no published evidence supporting a universal weighting.
The most important factor is usually the one currently creating the biggest weakness.
If crawlers cannot access the site, technical access comes first.
If the business is accessible but poorly understood, entity or content clarity may matter more.
If the website is strong but competitors dominate external sources, third-party authority may deserve priority.
How long before changes show up in AI answers?
There is no reliable universal timeframe.
Crawlers need time to revisit pages, search systems need time to process changes and external authority signals can take longer to develop.
Google notes that recrawling can range from days to months depending on how frequently its systems revisit a page.
OpenAI does not publish a fixed timetable for when broader business-information changes will affect ChatGPT recommendations.
Treat GEO as an ongoing visibility discipline rather than a one-week switch.
Can I do this myself?
Yes.
A business can audit listings, improve important service pages, maintain genuine review activity, fix crawler access and build credible third-party mentions internally.
The challenge is usually not knowing that these activities exist.
It is diagnosing which one actually deserves priority and measuring whether the answers change.
Does paid advertising influence AI answers?
There is no good basis for assuming that buying advertising automatically improves an organic AI recommendation.
Google explicitly says businesses cannot request or pay for better local organic ranking.
OpenAI says ChatGPT Search ranking uses factors intended to surface relevant and reliable information and does not guarantee placement.
Paid distribution can still increase brand awareness and create downstream business signals, but that is different from saying an advertising budget directly buys an AI recommendation.
Keep paid placement and organic AI visibility conceptually separate unless a platform explicitly labels a result as sponsored.
Can schema make ChatGPT recommend my business?
There is no published evidence supporting that claim.
Schema can help search systems understand structured information. Google explicitly documents that structured data helps it understand page content and organisational details.
That makes schema useful technical infrastructure.
It is not a recommendation guarantee.
Do reviews make ChatGPT recommend a company?
No platform has published a universal rule saying a particular number or rating guarantees inclusion.
Reviews are still worth examining because they provide independent evidence about a business, and Google confirms that reviews are relevant to prominence in its own local-ranking systems.
For GEO, treat reviews as one part of a wider authority and entity picture rather than a standalone trick.
AI visibility is usually a systems problem
Businesses often look for one missing setting.
A schema plugin.
An AI file.
A perfect prompt.
A technical toggle.
That is rarely how the problem presents itself.
AI visibility sits across several layers:
Can the system reach you?
Can it understand you?
Can it extract a useful answer from you?
Does the wider web support what you claim?
Do customers provide genuine evidence that the business does what it says?
Do all of those signals clearly point to the same organisation?
Those are the six levers.
They do not guarantee that an assistant will name you.
But they give you a practical framework for investigating why it currently does not.
And that is more useful than pretending anyone outside the platforms has access to a secret AI ranking formula.
If you want to see where your business currently stands before changing anything, start with Lobo Digital’s free checker.
Run the free AI visibility check
