What a GEO audit checks, and how the score is worked out
Guide · By Aashir Abdullah, founder of Taelvix · 20 November 2026
A GEO audit (generative engine optimization audit, also called an AI visibility audit) tests whether AI tools such as ChatGPT, Google Gemini and Google AI Overviews name a business when people ask the questions its customers ask. It then finds out why the business is or isn't named, and ranks the fixes. Below is the exact method Taelvix uses for local businesses.
Step 1: Collect the inputs
We need the business's name, website, city and state, two or three competitors, its top five services and, where they matter, the insurance plans or price points it offers. That is enough to write questions a real customer in that town would ask.
Step 2: Ask the questions customers ask
Our library has 24 questions in five groups. Every question includes the city and state, because AI answers change by location. The examples below are for a dental practice; the same groups work for any local business.
| Group | What it tests | Example |
|---|---|---|
| Discovery | Who should I go to? | What is the best family dentist in Austin, Texas? |
| Service | Who is best for a specific treatment? | Where is the best place to get dental implants in Austin, Texas? |
| Constraints | Insurance, price and opening times | Which dentists in Austin, Texas accept Delta Dental? |
| Comparison | This practice or that one? | What are good alternatives to [Competitor] in Austin? |
| Brand | Are the facts about the practice right? | What are the opening hours and address of [Practice] in Austin? |
Each question is asked on ChatGPT, Gemini and Google AI Overviews in a fresh session, with no follow-up questions. For every answer we record whether the business was named, its position, the tone, whether the facts were right, which competitors were named instead and which sources were cited. Key questions are asked a second time to check consistency.
Step 3: Score live AI visibility
The results become a weighted score from 0 to 100, so a client can see the number move month to month.
| Factor | Weight | Question it answers |
|---|---|---|
| Presence | 30% | Is the business named at all? |
| Position | 20% | Is it named first, or after competitors? |
| Consistency | 15% | Does it appear across ChatGPT, Gemini and AI Overviews, and across repeat runs? |
| Accuracy | 15% | Are its services, location and hours described correctly? |
| Sentiment | 10% | Is it described positively? |
| Completeness | 10% | Does the answer give a buyer what they need to act? |
Step 4: Check why
A low score is a symptom. The second half of the audit looks for the causes, across the website and the wider web, and produces a website readiness score from 0 to 100.
| Area | Weight | What we check |
|---|---|---|
| AI citability | 25% | Can AI tools quote clear, consistent facts from the website? |
| Brand authority | 20% | Is the business mentioned on sites AI tools trust and cite? |
| Content trust | 20% | Experience, expertise and trust signals, such as a named author with credentials |
| Technical | 15% | Can AI crawlers reach and read the pages? |
| Structured data | 10% | Is there correct, consistent schema markup for the business? |
| Platform presence | 10% | Is the business listed, with the same details, on the directories AI tools use? |
In practice this means crawling the site, comparing the opening hours, insurance details and review counts on every page, reading the robots.txt file and structured data, and checking the directories and ranking sites that the AI answers cited.
Step 5: Rank the fixes
The audit ends with an action plan ordered by impact and effort. In our sample audit, the top fix (correcting opening hours in two website templates) is estimated at about an hour of work, while building listings on the sites AI cites is ongoing work.
What a GEO audit can't tell you
- It is a snapshot. AI answers vary between runs and change over time, so the date of every test is recorded and monthly re-tests matter.
- It can't see inside the AI tools. The causes it lists are the most likely ones based on evidence, not a confirmed ranking formula.
- It can't guarantee a result. No one can promise that ChatGPT will name a business. A good audit tells you what to fix first.
Agencies can order this audit white-labeled through our GEO services, or start with a free pilot audit on one client.
Questions agencies ask
What is the difference between a GEO audit and an SEO audit?
An SEO audit asks how well a site ranks in Google's list of links. A GEO audit asks whether AI tools name the business in their answers, and why or why not. The two overlap on technical health and content quality, but a GEO audit also tests live AI answers and checks the third-party sources AI tools cite.
How many questions does a GEO audit test?
Our full audit tests 20 to 30 questions from a library of 24 prompts in five groups, adapted to the client's city and services. The free pilot uses a shorter set.
Which AI platforms does the audit cover?
ChatGPT, Google Gemini and Google AI Overviews. Each question is asked on each platform in a fresh session with the city included.
How long does a GEO audit take?
We deliver a full audit as a PDF with the agency's logo in 3 business days.