Free AI visibility audit: how to check if ChatGPT, Google AI Overviews and other AI engines recommend your brand

Shoppers now ask ChatGPT, Google AI Overviews, Perplexity, Gemini, Claude and Copilot what to buy, and those engines answer by naming a few brands. If yours is not one of them, you will not see it in your analytics. You will just see fewer people arriving who were already sure what they wanted.

An AI visibility audit tells you where you stand. This guide shows you how to run one yourself for free on every major engine, what to record so the results mean something, and what you get if you would rather have the free Eldor AI audit done for you.

The short version

• An AI visibility audit answers three questions. Do AI engines name your brand when shoppers ask for your category? Who do they name instead? And is what they say about your products true?

• You can check it yourself. You need a list of 20 to 30 real shopper prompts, a clean browser session for each engine and a spreadsheet. Plan on half a day for two engines.

• The hard part is making results comparable. Memory, location, plan and the wording of the prompt all change the answer, so one quick question in your own ChatGPT account tells you very little.

• The free Eldor AI audit does it for you. It runs 25 shopper prompts on ChatGPT and Google AI Overviews, checks 100 of your products against your live store and gives you a fix plan ranked by revenue at stake. No card and no install. Get your free AI audit.

What an AI visibility audit measures

A useful audit records four things for every prompt and every engine:

1. Presence. Is your brand named, linked or recommended? Being linked in a source list is not the same as being the product the answer recommends, so record which.

2. Share of voice. Which competitors are named in the same answer, and how often across all your prompts?

3. Sources. Which pages did the engine cite? Your own product pages, a marketplace, a review site or a listicle you have never heard of?

4. Accuracy. When the engine describes your products, are the price, sizes, materials, stock and policies right?

For a D2C brand the fourth one matters more than most audits admit. AI engines often answer from cached pages and third-party data, so an answer can recommend you and still quote last season's price or a returns policy you do not offer. That shopper arrives expecting something you cannot give them.

Before you start: write the prompts

The prompts decide whether the audit is worth anything. Do not ask "what is [your brand]?". Shoppers rarely ask that, and every engine will find your homepage. Ask what a shopper who has never heard of you would ask:

• Category and budget. "Best linen shirts for men under $80."

• Problem first. "Moisturiser for oily skin that does not cause breakouts."

• Comparison. "[Your brand] vs [competitor] for everyday wear."

• Occasion. "Gold earrings to gift my mother for her 60th birthday."

• Trust and policy. "Which jewellery brands offer free returns and lifetime exchange?"

Aim for 20 to 30 prompts across those types, written in the words your customers use in reviews, support chats and search terms. Fix the list before you run it. If you change prompts between runs, you cannot compare the results.

How to check each engine yourself

The method is the same everywhere: a clean session, the same prompts, the full answer recorded. What changes is where each engine gets its information, which tells you what to fix when you are missing.

ChatGPT

How to check. Use a logged-out session or a Temporary Chat in a private browser window, so your own Memory and custom instructions do not shape the answer. Ask each prompt and record the products named, the order, any product carousel and the sources listed. If your shoppers are in more than one country, run the prompts from each market, because ChatGPT's product results and feed options differ by region.

What it reads. ChatGPT blends OpenAI's own index, built by its OAI-SearchBot crawler, with results from outside search providers and merchant product data. OpenAI says sites that block OAI-SearchBot will not appear in ChatGPT search answers. Sponsored products are labelled and shown only to Free and Go users where ads are live, so do not count them as organic visibility. Our deep dive on how ChatGPT picks products in the US and Europe covers the sources in detail.

Google AI Overviews

How to check. Search each prompt on Google in a private window, signed out, and note whether an AI Overview appears at all. Google only shows one when it judges it adds something beyond classic results. When it appears, record the brands named and the links shown in the overview.

What it reads. Google's own guidance says a page only needs to be indexed and eligible to show with a snippet to be a supporting link, and that no special schema or extra optimisation is required. In practice that means AI Overviews reward the same things as Google Search: crawlable product pages, clear answers and up-to-date Merchant Center data. Clicks from AI Overviews are counted in Search Console's Performance report, under the Web search type, but not split out.

Perplexity

How to check. Ask each prompt signed out and record the products named and the numbered citations. Perplexity shows its sources more openly than most engines, which makes it the easiest place to see which third-party pages are shaping answers about your category.

What it reads. Perplexity builds its index with PerplexityBot and fetches pages live with Perplexity-User when a question needs them. Perplexity recommends allowing PerplexityBot in robots.txt. If you block it, you are unlikely to be cited.

Gemini

How to check. Use the Gemini app with a fresh chat and record the products, any shopping listings and the links offered. Google has been rolling out shopping in the Gemini app market by market through 2026, so the answer may look different depending on where you run it.

What it reads. Gemini's shopping answers draw on Google's Shopping Graph, the same product database behind Google Shopping. A complete, accurate Google Merchant Center feed and indexable product pages do most of the work here.

Claude

How to check. Start a new chat in Claude with web search turned on, ask each prompt and record the brands and the sources it cites. Without web search, Claude answers only from its training data, which tells you about brand awareness rather than current visibility.

What it reads. Anthropic runs separate crawlers for search and for training. Claude-SearchBot indexes pages for search and Claude-User fetches pages when a user asks something that needs them. Anthropic says blocking either one reduces your visibility in Claude's answers, while ClaudeBot, the training crawler, is a separate decision.

Microsoft Copilot

How to check. Ask each prompt in Copilot signed out and record the brands, shopping results and citations. Then open Bing Webmaster Tools: its AI Performance report, in preview since February 2026, shows how often your pages are cited in Copilot and Bing's AI answers and which queries triggered them. It is the only first-party citation report any engine offers today.

What it reads. Copilot draws on Bing's index and, for products, on Microsoft Merchant Center feeds. Microsoft Merchant Center can import an existing Google Merchant Center feed, so it is a quick win if you have not set it up.

What to record in your spreadsheet

One row per prompt per engine, with these columns:

• Prompt, prompt type, engine, market and date.

• Your brand: not named, named, linked or recommended.

• Position in the answer, if you are named.

• Competitors named, in order.

• Sources cited, with the domain of each.

• Anything said about your products that is wrong, with the correct fact from your store.

From that you can work out the numbers that matter: the share of prompts where you are named, your share of voice against each competitor, the domains engines trust in your category and how many answers about you are wrong. Our answer-share scorecard shows how to turn those into a weekly report.

Why a quick DIY check can mislead you

Running the audit yourself is worth doing once. It is also easy to get the wrong answer:

• Your own account is biased. If you have asked ChatGPT about your brand before, its Memory may bring you up more often for you than for a stranger.

• Answers move. The same prompt can name different brands an hour later. One run is a snapshot, not a ranking. Judge trends over several runs.

• Plans and markets differ. Free and paid ChatGPT users can see different products, and so can shoppers in the US, Europe and India.

• Accuracy is slow to check by hand. Comparing every claim with your live store across dozens of answers and products is the step most teams skip, and it is often the one that costs sales.

• The audit is only half the job. Knowing you are missing does not tell you which fix will matter most.

The free Eldor AI audit: what you get

If you would rather not spend the half day, or you want the product-level checks done properly, the Eldor AI audit is free and done for you. You send your store URL. We send back a report and fix plan you keep.

• 25 real shopper prompts, written for your category, price band and the way your shoppers ask. You can swap any of them before the run.

• Two engines: ChatGPT and Google AI Overviews, run in clean sessions.

• AI visibility. A visibility score out of 100, your share of voice against the competitors named, and the exact answers with the sources each engine cited.

• Prompt coverage. Which prompts you win, lose or are absent from, engine by engine.

• Product truth. 100 of your products checked against your live store, with every wrong price, size, material, stock level or policy logged with its source.

• Agent readiness. Product schema and feeds, missing attributes, size guides, quotable shipping, returns and COD policies, llms.txt, and a test checkout run by an AI agent.

• A fix plan ranked by revenue at stake, with each task marked as something Eldor can fix or something your team does by hand.

How to run it on Eldor

1. Go to eldor.ai/audit and enter your store URL, your name and your work email.

2. Tell us who you compete with, if you like. Name up to five competitors and we will track them in every answer. If you do not, we use the brands the engines name.

3. Review the prompts. We write 25 for your category and price band and you can swap any before we run them.

4. Get the report. It usually takes a few working days. You get the report, the fix plan and a walkthrough call.

We work from your public pages, so there is nothing to install. If you choose to connect your store, we can compare AI answers with your live prices, stock and policies rather than what your public pages show.

What about Perplexity, Gemini, Claude and Copilot?

The free audit covers ChatGPT and Google AI Overviews, the two engines most shoppers use today. To track more engines every week, Eldor's paid plans add them:

• Starter adds Perplexity, for three engines and 100 prompts tracked weekly.

• Growth adds Gemini, Claude and Copilot, for six engines and 300 prompts tracked weekly.

• Enterprise adds Google AI Mode, Grok and DeepSeek, for nine engines.

Until then, the steps above let you check any of them yourself. See pricing for the full plans.

Why do it now

AI answers name a handful of brands, not ten blue links. The brands named today are the ones engines have the clearest, most consistent facts about, and that advantage builds on itself as other sites cite them. A free audit is the cheapest way to find out whether you are one of them, which competitor is taking your place and which fix to start with.

Get your free AI audit. 25 prompts, two engines, 100 products, a ranked fix plan. Free, no card.

FAQ

Is there a free ChatGPT visibility checker?

Yes. You can check ChatGPT yourself by asking real shopper prompts in a logged-out or Temporary Chat session and recording which brands it names. For a done-for-you version, the free Eldor AI audit runs 25 shopper prompts on ChatGPT and Google AI Overviews and checks 100 of your products against your live store.

How do I check if ChatGPT recommends my brand?

Write 20 to 30 prompts a new shopper would ask in your category, run each one in a clean session with no Memory, and record whether your brand is named, linked or recommended, which competitors appear and which sources are cited. Repeat the run before drawing conclusions, because answers change.

What is included in the free Eldor AI audit?

25 shopper prompts on ChatGPT and Google AI Overviews, a visibility score, share of voice, the exact answers with cited sources, prompt coverage, 100 products checked against your live store, an agent readiness check and a fix plan ranked by revenue at stake, with a walkthrough call.

Does the free audit cover Perplexity, Gemini, Claude or Copilot?

No. The free audit covers ChatGPT and Google AI Overviews. Perplexity is tracked weekly on the Starter plan, and Gemini, Claude and Copilot on the Growth plan. You can check any of them yourself with the steps in this guide.

How is this different from free AI visibility checkers?

Most free checkers score a domain on a few generic prompts. The Eldor audit uses prompts written for your category and price band, checks your products against your live store for wrong prices, stock and policies, and ends with a fix plan ranked by revenue at stake. It is built for D2C stores rather than for any website.

How long does the free audit take?

Usually a few working days from when you send your store URL. You get a report and fix plan you keep, and a walkthrough call.

Do I need to install anything?

No. Eldor works from your public pages. You can choose to connect your store so answers are compared with your live prices, stock and policies.

Sources: OpenAI, "Overview of OpenAI crawlers" and Help Center, "Shopping with ChatGPT Search" and "Ads in ChatGPT"; Google Search Central, "AI features and your website"; Perplexity documentation, "Perplexity crawlers"; Anthropic support, "Does Anthropic crawl data from the web?"; Microsoft Bing Webmaster Tools, AI Performance report (public preview, February 2026); Microsoft Advertising, agentic commerce and Microsoft Merchant Center help; Google, shopping in the Gemini app announcements (2026).