An AI brand visibility competitive analysis measures how often ChatGPT, Perplexity, and Google AI Overviews name your brand compared with competitors. Run 15 to 20 buyer prompts five times on each platform, log every brand and cited URL, then compare mention rates and fix the sources that cite competitors instead of you.
Most clients come to me convinced that AI either loves them or ignores them. Both beliefs usually come from one ChatGPT search on their own phone.
When I started working with a sewer line and CIPP pipe relining company in Dallas-Fort Worth, the site ranked on Google for its own brand name and very little else. Before changing anything, I measured where it stood against competitors, in Google and in AI answers, so every later change had a baseline to beat.
This is the process I use. It needs a spreadsheet, a few hours, and no paid tool.
What is an AI brand visibility competitive analysis?
An AI brand visibility competitive analysis is a structured test of how often AI assistants name your brand, and your competitors, when buyers ask questions about your category. The output is not a ranking position. It is a rate: out of five runs of the same question, how many named you.
That rate is the difference from rank tracking. Google shows roughly the same results page to people searching the same thing from the same place. ChatGPT, Perplexity, and Google AI Overviews write a new answer every time.
If you want the wider picture first, our guide to what AI visibility is covers how these engines decide which brands to mention.
Why does one check tell you nothing?
Because AI answers change on almost every run. In SparkToro's research on AI recommendation consistency, 2,961 prompts were run across ChatGPT, Claude, and Google's AI Overviews, and the same list of brands came back less than 1% of the time.
So when a client tells me "ChatGPT recommends us," my first question is how many times they asked. The answer is almost always once. One run is an anecdote; five runs per prompt, per platform, is the smallest sample I trust for a direction.
How to perform an AI brand visibility competitive analysis, step by step
Work through five steps in order: find the competitors AI actually names, write buyer prompts, run each prompt five times on three platforms, log the same fields every time, and turn the log into rates. Skipping the first step is the most common reason the rest of the analysis measures the wrong thing.
Find the competitors AI actually names
Start with a discovery run before you pick competitors. Ask each platform five broad questions a buyer would ask, and write down every business named.
The list rarely matches the one the client gives me. On the Dallas-Fort Worth sewer project, the answers kept naming the same three businesses: Mother (Call Mother), Roto-Rooter, and Elite Rooter. Those three became the benchmark for every later run.
Pick the three or four names that appear most often. Those are your AI competitors, whether or not you think of them that way. If your own brand never appears in the discovery run, read why a business does not show up in AI answers before going further, because the cause may be technical.
Write 15 to 20 buyer prompts across three stages
Write prompts the way customers speak, not the way your service pages are titled. Split them across three stages of the purchase:
The gap between marketing language and customer language is where most prompt lists go wrong:
I draft the first list with AI, then rewrite every prompt by hand. Use the questions the client hears on the phone and the queries already showing in Search Console. AI-drafted prompts sound like marketing, and customers do not talk like marketing.
Run each prompt five times on three platforms
Use ChatGPT, Perplexity, and Google AI Overviews (or AI Mode when no Overview appears). Run each prompt five times per platform in a fresh private or logged-out session, and record the date and the location you ran it from.
20 prompts × 5 runs × 3 platforms = 300 answers
That is enough to see who owns a question, and small enough to do by hand.
For local businesses, location matters more than people expect. A prompt run from Dallas and the same prompt run from New York return different businesses, so either run local prompts from inside the client's market or put the city in the prompt.
Log five things for every answer
Keep one row per prompt, per platform, per run, with the same five columns every time:
The last column catches what most trackers miss: the AI naming you with the wrong service area, an old phone number, or a service you stopped offering. Wrong facts are an entity problem, and our guide to entity SEO explains how to correct them at the source.
Turn the log into mention rate and AI share of voice
Mention rate is the share of answers that name a brand. AI share of voice is one brand's mentions divided by all brand mentions across the same answers.
Here is how the numbers look for 20 prompts run five times on ChatGPT, which gives 100 answers. These are illustrative figures, not client data:
Calculate each platform separately. Averaging ChatGPT and Perplexity together hides the case where you win one and are invisible on the other.
Where are competitors getting cited from?
Open every URL the AI cited for a competitor on the prompts you lost. Each one falls into one of four types, and the type tells you where the fix happens.
The first gap was on the client's own site. Every sewer service sat on one general page, so neither Google nor an AI engine had a specific page to cite for hydro jetting or CIPP relining.
I rebuilt the structure with a Sewer Line parent section and a child page for each service: sewer cleaning, pipe repair, hydro jetting, and CIPP relining. Changing URLs on a live site carries risk, and I handled it the way our guide to redesigning without losing SEO describes.
Business listings came next, because review sites and directories are the other place these engines pull local businesses from. Our guide on why ChatGPT does not show your reviews covers which review content AI can actually read.
Within 15 days of the restructure and the first new posts, Search Console showed impressions for service keywords, not just the brand name, and clicks started rising.
For the page-level side of the fix, our guide to improving brand visibility in AI search engines covers what makes a page quotable.
How do you read the results without fooling yourself?
Act on large, repeated gaps and ignore small ones. A competitor named in four of five runs while you appear in none is a pattern. One run against two is noise.
Keep the prompt set fixed between checks. Change the prompts and every trend line becomes meaningless, because you are no longer measuring the same thing.
The manual method does not scale. Past about 20 prompts, logging becomes the whole job, and a paid tracker earns its fee; our comparison of AI visibility tools and what they cost covers the options. For a single local business, 20 prompts is plenty.
How do you report it to a client?
Report three numbers, three gaps, and three actions. Clients do not need 300 rows; they need to know whether they are gaining ground and what happens next.
My clients get a weekly report covering improvements, stats, leads, and the work done that week. The AI numbers belong in that report, but I would not rerun the full analysis weekly. Monthly checks of the priority prompts and a full rerun each quarter show real movement.
Then there is the conversation every SEO client has in the second month:
"Traffic is up, so where are the calls?"
Visibility, in Google or in AI, gets a business considered. Turning that attention into calls is a separate job.
I grow visibility first, then work on conversion for the pages that already get the most traffic. Setting that expectation at the start saves a difficult call later, and our take on whether GEO is worth it for a small business covers the same point for owners weighing the cost.
Key takeaways
An AI brand visibility competitive analysis replaces one-off checks with a rate. Fifteen to 20 buyer prompts, five runs each, on ChatGPT, Perplexity, and Google AI Overviews, gives you enough answers to see who owns each question.
The value is in the cited sources. Each one tells you whether the fix is on your own site, on a review platform, in a community, or in outreach, and that turns a visibility score into a work plan.
Start with the discovery run this week: five broad buyer questions, five runs each, on one platform. It is the smallest version of the analysis, and it shows you who your AI competitors really are.
Frequently asked questions
How often should you run an AI brand visibility competitive analysis?
Run a full analysis every quarter and check your priority prompts monthly. AI answers shift constantly, so weekly reruns mostly measure noise, while a quarterly rerun of the same prompt set shows real movement. Rerun sooner after a big change, such as a site restructure, a new service page, or a wave of new reviews.
Can you do an AI competitive analysis without a paid tool?
Yes. A spreadsheet, 15 to 20 buyer prompts, and five runs per prompt on ChatGPT, Perplexity, and Google AI Overviews give you a reliable picture for a small or local business. Paid trackers become worth it when you need hundreds of prompts, several markets, or daily tracking across many clients.
Why does ChatGPT name my competitor but not me?
Usually because the sources ChatGPT draws on describe your competitor more clearly. Check the cited URLs. If they are the competitor's own pages, those pages answer the question more directly. If they are review sites, directories, or Reddit threads, your business is missing or poorly described there. Fix the source, not the prompt.
How many prompts do you need?
Fifteen to 20 buyer prompts is enough for most businesses, split across problem, comparison, and decision questions, with each one run five times per platform. Fewer prompts miss whole buying stages. More than about 20 turns manual logging into a full-time job and adds little new information for a single business.
What is AI share of voice?
AI share of voice is the percentage of all brand mentions, across a fixed set of AI answers, that belong to your brand. If 100 answers contain 120 brand mentions and 12 of them are yours, your share of voice is 10%. Track it per platform and compare it with your top competitor over time.
Do you need special markup to appear in Google AI Overviews?
No. Google's documentation says there are no extra requirements and no special schema needed to appear in AI Overviews or AI Mode. Standard SEO decides it: indexed pages that answer the question clearly and are backed up by other sites. Structured data still helps search engines understand what your business does.
Want this analysis run for your clients?
Running the analysis once is manageable. Repeating it every quarter for every client, and acting on what it finds, is where most agencies run out of hours.
Survyc is an AI-first digital agency, and we run the full process for agencies and their clients: the prompt set, the runs, the source audit, and the fixes on the pages and listings that matter. The WordPress side of those fixes is our white label AI visibility implementation. We can deliver it under your agency's brand.
Tell us which client needs an AI brand visibility competitive analysis first, or email info@survyc.com.
Get in touchLast reviewed: September 2026. AI platforms change quickly, so this guide is reviewed every quarter.
