7 Ways to Check If AI Knows Your Brand (Prompts + Tracking Sheet)

The 7 tests

1. 1. Brand test (ChatGPT and Claude)

   Ask: 'Search the web. What is [BRAND]? What do they sell, where are they based, who runs it and what does it cost? List your sources.' Mark each fact right, wrong or missing. Run it in Claude too, with web search on.

2. 2. Category test (Perplexity)

   Ask: 'What is the best [SERVICE] in [CITY]?' or 'Best [PRODUCT TYPE] for [CUSTOMER]?' Don't name yourself. Are you named? In what position? Perplexity numbers its sources, so copy them all.

3. 3. Competitor test (any tool)

   Ask: 'Why would someone choose [COMPETITOR] over [BRAND]?' This one stings. The reasons it gives are what the web says you lack, or what it can't find about you.

4. 4. Google test (AI Overviews and AI Mode)

   Search your top 5 buyer questions on Google. Note whether an AI Overview appears and which sites it links. Repeat in AI Mode where it's available. Then check Search Console's Generative AI performance report to see which of your pages get AI impressions.

5. 5. Gemini and Copilot test

   Run tests 1 and 2 in Gemini and Microsoft Copilot. Different tools lean on different search indexes, so answers often differ. Then open Bing Webmaster Tools > AI Performance to see which of your pages Copilot cites.

6. 6. Access test

   Check your robots.txt doesn't block the search crawlers: Googlebot, Bingbot, OAI-SearchBot (ChatGPT search), PerplexityBot and Claude-SearchBot. Also check Search Console > Settings > Search generative AI hasn't been switched off (sites are included by default). Make sure key facts are plain HTML text, not only images.

7. 7. Traffic test

   In Google Analytics 4 open Reports > Acquisition > Traffic acquisition, change the dimension to Session source, and look for chatgpt.com, perplexity.ai, gemini.google.com, claude.ai and copilot.microsoft.com. Some AI visits arrive with no referrer and show as Direct, so treat this number as a floor.

Free tools to run the tests

1. ChatGPTFree · web search on Free plan and signed-out (usage limits apply)

   Answers with web search and shows sources when it searches.

   Best for: Tests 1 and 3

   Ritik's note: Start the prompt with 'Search the web' so you see live sources, not just what the model remembers.

2. PerplexityFree plan available

   AI answer engine that searches the web and cites numbered sources in every answer.

   Best for: Test 2 and reading sources

3. ClaudeFree plan available

   Anthropic's assistant. With web search on, it fetches pages and links its sources.

   Best for: Test 1

4. Google GeminiFree with a Google account

   Google's AI assistant.

   Best for: Test 5

5. Microsoft CopilotFree

   Microsoft's assistant, grounded in Bing search.

   Best for: Test 5

6. Google Search Console: Generative AI performance reportFree

   Shows impressions of your links inside AI Overviews and AI Mode, by page, country, device and date. Launched June 2026; Google says it reached all sites by 31 August 2026.

   Best for: Test 4

   Ritik's note: It shows impressions only: no clicks and no queries. Use it to find which pages AI features pick, not to measure sales.

7. Bing Webmaster Tools: AI PerformanceFree · public preview

   Shows how often Copilot, Bing's AI summaries and some partner integrations cite your pages, plus sample 'grounding queries'.

   Best for: Test 5

8. theStacc AI Visibility CheckerFree · no signup

   Requests your page as seven AI crawlers and a normal browser, reads your robots.txt rules per bot, and checks text before JavaScript, noai/nosnippet tags, llms.txt and structured data.

   Best for: Test 6

   Ritik's note: It checks whether AI can read your site, not whether AI mentions you. Pair it with tests 1 to 5.

Copy-paste: a robots.txt that lets AI search in

# Search crawlers: these put you in AI answers
User-agent: OAI-SearchBot
Allow: /

User-agent: PerplexityBot
Allow: /

User-agent: Claude-SearchBot
Allow: /

# Training crawlers: your choice. Blocking these does NOT
# remove you from ChatGPT, Perplexity or Claude search.
# User-agent: GPTBot
# Disallow: /
# User-agent: ClaudeBot
# Disallow: /
# User-agent: Google-Extended
# Disallow: /

User-agent: *
Disallow: /wp-admin/

Sitemap: https://yourdomain.com/sitemap.xml

Tracking sheet and score

Paste into cell A1 of a Google Sheet, then Data > Split text to columns (comma). One row per prompt per tool.

Date,Tool,Test #,Prompt (exact),Named? (0/1),Position (1=first),Facts right (0-2),Our site cited? (0/1),Competitors named,Sources cited (URLs),Fix to do
2026-11-07,ChatGPT,1,What is [BRAND]? ...,1,,1,0,,example.com/a; example.com/b,update pricing page
2026-11-07,Perplexity,2,Best [SERVICE] in [CITY]?,0,,,0,Comp A; Comp B,listicle-site.com/best,ask to be added to listicle-site.com

Named in the answer: 1 point
Named first or second: +1
Facts right: 2 = all right, 1 = some wrong, 0 = mostly wrong or 'no information'
Your own site cited as a source: +1

Monthly score = total points / (number of answers x 5) x 100

Worked example (illustrative, not a real business)

A made-up physiotherapy clinic in Pune runs 4 prompts in 3 tools: 12 answers, 60 possible points.

Turn the gaps into a fix list

After the test: your first 30 days

1. Week 1: fix your own site

   Correct every wrong fact on your site first: prices, hours, services, location, who runs it. Put them in plain text on the About, contact and pricing pages.

2. Week 2: fix the sources AI cited

   Update your Google Business Profile and any directory or listing the tools cited with old facts. Most let you edit for free.

3. Week 3: get onto one cited page

   Pick the page cited most often in your category test. Ask its owner, honestly and with your facts ready, whether they'd consider including you.

4. Week 4: re-run the same prompts

   Same prompts, same tools, 3 runs each. Compare your score. Small or no change after one month is normal; the pages you fixed may not be recrawled yet.

What's confirmed vs what's guessed

A lot of AI visibility advice online is guesswork. Here is where each claim in this kit stands.

Common testing mistakes

• Testing while logged in with your own history

  Tools can remember you and lean the answer your way.

• Running a prompt once and trusting it

  Answers change between runs. Run category prompts 3 times.

• Naming your brand in the category prompt

  Real buyers ask 'best X in Y'. They don't name you.

• Rewording prompts every month

  Then you can't tell if the change came from the web or from your wording.

• Blocking every AI bot to 'protect content'

  Blocking the search bots removes you from those answers. Decide on training bots separately.

• Buying fake reviews or list placements

  It breaks Google's spam policies and review platform rules, and buyers can tell.