AI VISIBILITY AUDIT · CASE STUDY
Why ChatGPT Doesn’t Recommend Your Local Business: What an AI Visibility Audit Reveals
A real AI visibility audit across ChatGPT, Gemini, Perplexity and Google found zero appearances in 44 commercial answers. Here is what the measurement revealed and what changed next.
Oleksandr KoliesnikovAI Visibility, GEO & Local SEO consultantChatGPT may overlook a local business even when it has a working website, years of experience and active profiles. In most cases, the problem is not one missing keyword. The AI system may be unable to identify the business consistently, find enough independent confirmation or extract a clear and current answer about its expertise.
I discovered this through my own project, Koliesnikov.com.
On July 27, 2026, I tested 12 customer questions across ChatGPT, Perplexity, Gemini and Google AI. The measurement produced 48 answers. Forty-four were commercial, non-branded answers—and neither Koliesnikov.com nor my name appeared in any of them.
AI systems could find me when they were given my exact name. They simply did not connect me with the commercial questions I wanted potential clients to ask.
That distinction is the reason an AI Visibility Audit must measure more than whether ChatGPT “knows” a business.
MEASUREMENT SNAPSHOT
Measurement Snapshot
- Baseline date
- July 27, 2026
- Questions tested
- 12
- Systems tested
- ChatGPT, Perplexity, Gemini and Google AI
- Total answers collected
- 48
- Commercial non-branded answers
- 44
- Commercial mentions of Koliesnikov.com
- 0
- First spot-check appearances recorded
- August 12, 2026
My Baseline: Zero Mentions in 44 Commercial AI Answers
The first measurement included customer questions in English, Ukrainian and Russian. They covered several types of intent:
- Who offers AI Visibility Audits in Toronto?
- Who can help a Canadian business get recommended by ChatGPT?
- Who is a GEO specialist in Toronto?
- Who combines Local SEO and GEO for contractors in the Greater Toronto Area?
- Who are the best GEO consultants in Toronto?
- Should a local business invest in GEO or traditional SEO first?
- Who is Oleksandr Koliesnikov, and what does he specialize in?
Each question was tested in four systems: ChatGPT, Perplexity, Gemini and Google.
ChatGPT and Google identified the right person on the branded query. Perplexity initially selected another person with the same name. Gemini combined parts of my profile with information about a namesake.
The business existed online, but the entity was still ambiguous.
AI Knew My Name but Did Not Recommend My Services
This was the most important finding.
AI systems could extract a substantial amount of information about my background in digital marketing, Local SEO, paid advertising, lead generation, CRM and automation. I have worked in digital marketing since 2010, so much of this information was accurate.
However, my current focus—AI Visibility, GEO and Local SEO for local businesses and personal brands—was not yet strong enough to replace the older, broader description of my work.
The systems knew individual facts. They had not assembled those facts into the commercial conclusion I wanted:
Oleksandr Koliesnikov is a Toronto-based AI Visibility, GEO and Local SEO consultant.
This is the difference between being discoverable by name and being selected for a client’s question.
The Audit Revealed Three Separate Problems
I organized the findings into three practical layers.
1. Identity: was every profile describing the same person?
My name, website, older business identity and current services were not connected consistently enough.
I had previously operated LikeBooster in Ukraine and later used the LikeBooster name for a sole proprietorship in Canada. I then moved toward a personal consulting brand through Koliesnikov.com.
That transition was logical to a person who knew my history. It was less obvious to a machine collecting separate facts from different pages.
2. Independent confirmation: who supported my own claims?
My website could describe my services and experience, but it remained a source I controlled.
During the audit, AI systems specifically pointed to missing evidence such as independent reviews, external articles, interviews, professional profiles and client results that could corroborate the relationship between my name, website and expertise.
This was the largest gap in the first measurement.
3. Extractable expertise: could a system retrieve the right answer?
The systems found detailed information about my general marketing work. They did not consistently retrieve my newer GEO positioning.
This meant I needed more than a new headline. The site structure, author information, service pages, cases and external profiles had to support the same conclusion.
I Rebuilt the Website Twice After the Audit
After seeing zero appearances in 44 commercial answers, I continued studying how AI systems discover and assemble information. I then rebuilt the website twice.
The new structure had to perform several jobs:
- Support English as the primary market language
- Include Ukrainian as my native language and the site’s second full language
- Present AI Visibility, GEO and Local SEO as the main specialization
- Preserve my marketing experience since 2010 without allowing older services to dominate the current positioning
- Connect my personal name with Koliesnikov.com
- Bring cases, speaking experience, services and professional profiles into one coherent structure
I added a complete About section, case studies, speaker information and a page through which organizations can invite me to speak. I also connected my official profiles and implemented structured data to describe the relationship between the person, website and services.
Google’s documentation explains that structured data provides explicit clues about the people and companies described on a page. It helps a search system interpret those relationships, although it does not guarantee a recommendation by itself. Google Search Central: Structured Data
The objective was to make all relevant pages support one identity—not several disconnected versions of the same person.
My Main Mistake Was Trying to Communicate Everything at Once
My first instinct was to build a website that did everything immediately: fast, attractive, technically advanced and comprehensive enough to present every service I could provide.
That created a positioning problem.
I was trying to preserve the full range of my marketing experience while establishing a narrower commercial focus in GEO. The broader story was true, but it made the main conclusion harder to extract.
The correction was to organize the experience around a clearer specialization:
- GEO and AI Visibility for local businesses
- Local SEO as a connected service
- Personal GEO for entrepreneurs and experts
- Broader marketing capabilities as supporting experience
A business does not need to erase its history. It needs to make the current commercial priority unmistakable.
Early Spot Checks Show Progress, but Not a Complete Re-Measurement
On August 12, 2026, at approximately 11 p.m. Toronto time, I ran several new incognito checks.
For the Google query ai visibility specialist toronto, Google’s AI Overview included Oleksandr Koliesnikov as the second person listed during a spot check on August 12, 2026. It described me as a Toronto-based consultant specializing in AI Visibility and GEO alongside Local SEO. Koliesnikov.com also appeared among the supporting sources.
For the ChatGPT query Can you recommend a marketer in Toronto who speaks Ukrainian?, ChatGPT included me after ToroN2 Consulting and described me as a Toronto/Etobicoke marketing consultant. However, the answer still emphasized SEO, Google Ads, Meta Ads, lead generation and automation. It did not yet present GEO as my main specialization, and it suggested confirming whether I spoke Ukrainian.
The response also surfaced an older local profile separately. I renamed that profile immediately after the check to align it with my current positioning.
These were two spot checks, not a complete repeat of the original 44 commercial answers. They show early movement, but they do not yet establish a stable ranking or prove that every change caused the appearances.
Why GEO Confirmation Is Different From Traditional Link Building
For GEO, I do not evaluate a placement only by asking whether it contains a backlink.
I ask four different questions:
- Can the relevant AI or search system access the page?
- Does the page clearly refer to the same person or business?
- Does it confirm a fact that matters to a potential client?
- Can the system extract a useful answer from it?
Different AI products retrieve information in different ways.
OpenAI states that ChatGPT Search may use third-party search providers and partner content, while its answers can include links to web sources. OpenAI also provides a dedicated search crawler, OAI-SearchBot. A site that blocks this crawler cannot appear as a normal source in ChatGPT Search answers, although a navigational link may still surface. OpenAI: ChatGPT Search and OpenAI Crawlers
Google explains that its generative search features use retrieval-augmented generation and “query fan-out”: the system may run several related searches before composing its answer. A page must also be indexed and eligible to appear in Google Search before it can be shown in Google’s generative features. Google’s guide to generative AI search
This is why I treat a website, LinkedIn, YouTube, X, professional profiles, external publications and genuine reviews as different evidence surfaces. I do not assume that every model reads every source in the same way. I test the questions and inspect what each system actually retrieves.
Clear Sources and Concrete Evidence Matter
The original academic paper that introduced Generative Engine Optimization evaluated a benchmark of 10,000 queries. Its authors reported that adding relevant citations, quotations and statistics could improve source visibility by more than 40% in some settings. They also found that the effect varied by topic and that results on Perplexity reached improvements of up to 37%. These were visibility measurements—not guarantees of traffic, leads or recommendations. GEO: Generative Engine Optimization, KDD 2024
The practical lesson is straightforward: a business should publish evidence that an AI system can reuse.
That can include:
- Specific case details
- Dates and measurable results
- Clear service definitions
- Direct answers to customer questions
- A real, verifiable author
- Independent reviews and mentions
- Links to authoritative supporting sources
- Consistent relationships between the person, company and official profiles
General claims about being “innovative” or “the best” provide far less usable evidence.
How I Recommend Auditing a Local Business
An AI Visibility Audit should begin with real customer questions—not branded searches that already contain the company’s name.
My process is:
- Identify the questions potential clients ask before choosing a provider.
- Include commercial, comparative and best-of questions.
- Test several systems separately, using clean or incognito sessions where possible.
- Record whether the business is mentioned, linked, prominent or absent.
- Record which competitors appear instead.
- Ask each system why it selected those competitors.
- Ask what evidence is missing for the business being audited.
- Separate identity problems, independent-confirmation gaps and content gaps.
- Make changes and repeat the same measurement later.
The objective is not to obtain one flattering answer. It is to establish a baseline that can be compared with future measurements.
When This Approach Is the Right Fit
This work is best suited to a local business willing to clarify its positioning, correct outdated information and publish verifiable evidence.
It is especially relevant when:
- The business is found by name but absent from service recommendations
- AI systems confuse the owner with another person
- Old services dominate newer positioning
- Several profiles describe the same business differently
- Competitors appear repeatedly for commercially valuable questions
- The website contains expertise but AI systems do not reuse it
The process takes sustained work. Indexing, profile updates, external confirmation and repeated measurements happen over time.
Start With a Baseline, Not an Assumption
Before changing dozens of pages or creating more profiles, find out what AI systems currently say.
Test real customer questions. Record every answer. Identify who appears instead of you. Ask the systems why.
My own baseline was uncomfortable: zero appearances in 44 commercial answers. It was also useful because it showed exactly where the problem was.
Sixteen days later, I could see the first appearances in relevant answers. The new observations do not prove that the work is complete or represent a complete rerun. They show that the business is beginning to move from being recognized only by name toward being connected with the expertise I want to own.
That is what an AI Visibility Audit should provide: an honest starting point, a map of the missing evidence and a measurement that can be repeated.
AI VISIBILITY AUDIT
