Skip to content
← All practitioner guides

AI Search · · Updated · 13 min read

AI Search and GEO: What a Business Can Actually Control

A hand holding a phone above an open notebook at a conference table

How to check whether AI Overviews, ChatGPT and Perplexity mention your business, and what you can control: crawler access, consistent facts and a prompt log.

Leandro Álvarez
Founder & Digital Director, Niagara SEO

Do not assume every AI answer engine works the same way

Some products retrieve live web results, some use search partners, some combine retrieval with model knowledge, and some expose citations more clearly than others. Even within one product, behaviour can change by query, market, account state, and product version.

That means statements such as all AI engines use the same RAG pipeline or schema directly increases recommendation probability should be treated with caution unless the specific system documents that behaviour.

02

Make first-party business information clear and verifiable

Publish clear descriptions of what the business does, where it operates, who is responsible for the work, what important services include, and what evidence supports key claims. Case studies, project documentation, useful FAQs, pricing context, policies, and accurate About information can all become valuable source material.

If the website contradicts the Google Business Profile, directory listings, social profiles, or other trusted sources, machines and people have a harder time deciding which version is current.

03

Use structured data as description, not persuasion

Schema.org markup can express entities and page content in a machine-readable form. Use it to describe what is on the page, not to insert facts that the public site does not support.

Stable @id values can help keep Organization, Person, Service, Article, Product, and other nodes consistent across a site. The value comes from clarity and accuracy, not from stuffing every available schema property into the graph.

04

Third-party evidence matters because it is independent

Directories, reviews, trade associations, news coverage, client references, and other external sources can independently describe a business. The most useful strategy is not to manufacture mentions, but to make business information consistent and earn credible references through real work and relationships.

Never create fake reviews, fake citations, or false local presence in an attempt to influence AI answers.

05

llms.txt can be useful context, but it is not a ranking switch

An llms.txt file can provide a concise machine-readable overview of important site resources. It is useful as a maintained context document, particularly for tools that choose to read it.

Do not claim universal adoption or guarantee that publishing the file will cause an AI product to cite the business. The human-facing pages remain the primary source of truth.

# Example Business

> Short factual description.

## Core Resources
- Services: https://example.ca/services/
- About: https://example.ca/about/
- Contact: https://example.ca/contact/

06

Measure mentions as observations, not permanent rankings

Define a set of relevant questions, record whether the business appears, whether facts are accurate, and which sources are cited when citations are available. Repeat the observation over time under comparable conditions.

A mention in one prompt on one day is not equivalent to a stable search ranking. Treat AI visibility as a changing observation and combine it with measurable referral traffic and business outcomes where possible.

07

Google's AI features draw on the same foundations as Search

For Google specifically, AI Overviews and AI Mode are part of Google Search. Google's own guidance for site owners states that there are no additional technical requirements or special optimizations needed to appear in these features: a page needs to be indexed and eligible to show in Search with a snippet.

In practice, that means the unglamorous work still comes first. If key pages are not indexed, if the site blocks Googlebot, or if a page has nothing specific to say, there is no separate AI trick that compensates for it.

08

Check whether your site blocks the crawlers you want

Several AI companies publish the names of the crawlers they use, and many of them respect robots.txt. Some sites block these crawlers by accident through a security plugin, a hosting default or a copied robots.txt file. Others block them on purpose without realizing that some of those agents are used for search and citation features rather than only for model training.

OpenAI, for example, documents separate agents for training (GPTBot) and for ChatGPT search (OAI-SearchBot). Google-Extended is a control for certain Gemini uses, while Google Search and its AI features are crawled by Googlebot. Allowing or blocking each one is a business decision, but it should be a deliberate one. Check each company's current documentation before changing anything, because these agents and their purposes change.

# robots.txt: an explicit, deliberate policy (example only)
User-agent: OAI-SearchBot
Allow: /

User-agent: PerplexityBot
Allow: /

# Training crawlers can be allowed or blocked separately
User-agent: GPTBot
Disallow: /

09

Write pages that answer the questions people ask

People tend to ask AI assistants fuller questions than they type into a search box: who installs custom stairs in Niagara Falls, whether a painter in St. Catharines works on older homes, or what a small business website should cost in Ontario. A useful answer needs specific, verifiable facts to draw on.

For a local business, that usually means service pages that say exactly what is offered and where, project or case-study pages with specific details, plain-language FAQs, transparent pricing context where the business is willing to publish it, and an About page that identifies the people responsible for the work.

Vague claims such as best in the region or trusted by everyone give a system nothing it can confirm. A sentence like we build and install custom interior stairs for homes across the Niagara Region, and here are three recent projects, gives it something concrete.

10

Measure with a simple prompt log and referral data

Pick ten to twenty questions a customer might ask, run them in the AI products your customers are likely to use, and record the result in a spreadsheet: date, product, question, whether the business appeared, whether the facts were accurate, and which sources were cited. Repeat monthly under comparable conditions.

Combine that with data you can measure directly. Analytics tools can show referral visits from domains such as chatgpt.com or perplexity.ai, and Search Console reports performance for Google Search, which includes its AI features. Neither source is complete, but together they are far more reliable than a single screenshot of a good answer.

date        product     question                                    appeared  accurate  cited source
2026-10-01  ChatGPT     custom stair builder in Niagara Falls?      yes       partly    company site
2026-10-01  Perplexity  custom stair builder in Niagara Falls?      no        n/a       directory listing

11

A one-hour baseline you can do today

You do not need a tool to start. Write down ten questions a customer might ask before choosing a business like yours: some that name you directly, some that describe the problem and the area without naming anyone. Run each one in the assistants your customers are likely to use, and save the full answer with the date, the product and the place you searched from.

Then read the answers as a customer would. Is the business mentioned? Are the facts correct: services, location, prices, opening hours? Which sources are cited? Do not score yourself on a single answer, because assistants vary between sessions. The value of the first pass is a record of what exists today and a list of facts to correct.

  • Ten questions: some branded, some about the problem and place.
  • Save the whole answer, date, product and location.
  • Mark every factual error and the source it likely came from.
  • Repeat monthly under similar conditions.

12

Fixing wrong facts at the source

When an assistant gets a fact wrong, the cause is usually somewhere else on the web. Trace the error: check your own website first, then your Business Profile, then directories, social profiles and any article that mentions the business. An old address in a directory or a service you stopped offering can sit in several places at once.

Correct each source rather than only the page you control. Update directories, claim and edit profiles, and ask the owner of a third-party article to correct a clear factual error. Some AI products offer feedback tools for reporting incorrect answers, and it is reasonable to use them, but do not expect an immediate change. Recheck monthly, record what changed and avoid claiming credit for a correction unless the evidence supports it.

  • Check your website, profile, directories and press mentions in that order.
  • Correct the source, not just your own page.
  • Use a product's feedback tool where one exists.
  • Recheck monthly and keep the record.

13

What not to spend money on

Be cautious with anyone who guarantees citations or a particular answer, because nobody outside the AI provider controls them. Mass-produced question-and-answer pages add volume without adding information, and they can make the site worse for people. Paying for mentions on low-quality sites does not make a business a better source, and it can cause problems if the sites are poor.

An llms.txt file is optional and cheap, and it can be a useful summary for tools that read it, but it is not a ranking switch and no provider has promised to rely on it. The work that holds up is slower and plainer: accurate business facts, clear pages that answer real questions, consistent listings and a measurement habit. If a proposal does not fit that description, ask what evidence supports it.

  • Do not pay for guaranteed citations.
  • Do not publish pages of generic questions with thin answers.
  • Do not treat llms.txt as a requirement.
  • Do put the time into facts, clarity and consistency.

14

Why local questions behave differently

Questions about local businesses usually depend on where the person is and which sources an assistant can reach. Some products draw on map data, review sites and directories, and others rely mainly on web pages. The same question can return different businesses for different locations or sessions.

For that reason, test local questions from the place a customer would search, and note the location in your record. Keep the Business Profile, directory listings and website saying the same thing about services, hours and area. Reviews matter because they are an independent signal that real customers exist. Treat each result as one observation and judge change only over several months and comparable checks.

  • Test from the location a customer would use.
  • Keep profile, directories and website consistent.
  • Collect reviews honestly, without incentives.
  • Compare results across months, not single answers.

This guide explains a method. Service scope, access, responsibilities and fees are agreed for the actual project; rankings, AI citations and conversion gains are not guaranteed.

AI search optimization services

About this guide

Frequently asked questions.

Is AI search optimization different from SEO?
Mostly it is an extension of good SEO rather than a replacement. Indexable pages, accurate business information, useful content, structured data and credible third-party references help in traditional search and in AI-assisted answers. The additional work is mainly checking AI crawler access and measuring AI answers and referrals directly.
Should I block GPTBot and other AI crawlers?
It depends on what you want. Some agents are used for model training and others for search or citation features, and companies document them separately. Blocking a search-related agent can reduce the chance that the product reads your pages. Decide deliberately for each agent instead of relying on a plugin default.
How can I see traffic from ChatGPT or Perplexity?
In most analytics tools, visits that arrive from links in these products appear as referral traffic from domains such as chatgpt.com or perplexity.ai. Not every AI interaction produces a click or passes a referrer, so treat the numbers as a minimum rather than a complete picture.
Can a local business appear in ChatGPT, Perplexity, Gemini, or other AI answers?
Yes, local businesses can appear in AI-assisted answers, but the conditions differ by product and query. Clear first-party information and credible third-party references can improve the source material available about a business, but they do not guarantee a mention.
Does llms.txt guarantee that AI systems will understand or cite my website?
No. It is a useful optional context file for systems that choose to consume it. The website's accessible pages, factual consistency, and external evidence remain more fundamental.
Can Niagara SEO guarantee the top AI recommendation?
No. AI products are dynamic systems outside our control. We can improve information quality and monitor observed visibility, but we cannot guarantee a recommendation, citation, or fixed position.

Free Website Review

Already have a website? Start with the Free Website Review.

Start with the Free Website Review: a concise, human-written recommendation from Jesse and Leandro, delivered by email. No automated hundred-item PDF. No sales call. Or talk to us about a scoped SEO or content project.

Get a Free Website Review

Know what you need?

SEO services in the Niagara Region Discuss an SEO or content project

Custom work has its own written scope and fee. A free review is not a paid technical investigation.