Free ways to track AI search visibility
How to track AI visibility for free: a weekly manual audit with a copy-pasteable spreadsheet, what each free route gives you, and when to move to a tool.
- The manual audit costs nothing: 10 to 15 buyer prompts, run on ChatGPT and Perplexity once a week, logged in a spreadsheet.
- Log five things per answer: mention, position, tone, cited sources, and the date.
- Google Search Console added a generative-AI performance report in June 2026, which shows impressions for Google's own AI surfaces only.
- Free stops working when the prompt list outgrows the sheet, a second person needs the numbers, or someone asks for a report.
- advisible runs on your own provider keys: unlimited team members, token cost at your provider's own prices.
The manual audit: ten prompts, two engines, one routine
Write 10 to 15 questions your buyers actually ask, run them on ChatGPT and Perplexity once a week, and log mention, position and cited sources in a spreadsheet. That is the whole method, and it produces a baseline you can defend in two afternoons.
Cover category questions, comparison questions, and a couple of branded ones, then leave the list alone. Same prompts, same day each week. Answers wobble from run to run, so a single check measures nothing; the line you build over a month is the story. Keep the full answer text in the last column rather than just your notes, because the reasons behind a mention stop being obvious three weeks later.
| Prompt | Engine | Date | Mentioned | Position | Cited sources | Full answer |
|---|---|---|---|---|---|---|
| best help desk software for startups | ChatGPT | 2026-10-06 | Yes | 3 | g2.com, example.com | Paste the answer text here |
What each free route gives you, and what it costs
Four routes run at zero licence cost. They differ in whether they produce a series or a snapshot, and only a series survives contact with a busy week.
| Route | What you get | What it costs you |
|---|---|---|
| Browser checks by hand | Real answers today, exactly as a user sees them | Nothing but time; it dies after about five prompts |
| Spreadsheet with a weekly reminder | A series you can chart and count | Discipline; it breaks when a second person touches it or engines multiply |
| One-off free checkers | A snapshot of one moment, sometimes a short report | Usually an email address; no history, so nothing to compare against next month |
| A workflow tool wired to your own keys | The manual audit on autopilot, reading prompts from a sheet and writing results back | A few dollars a month in tokens plus an afternoon of setup |
The fourth route is the one worth knowing about. A walkthrough of exactly this pattern, prompts in a Google Sheet, calls to ChatGPT, Gemini and Perplexity with your own API keys, results written back on a schedule, put the running cost at low digits per month for a 25-prompt list (TechRadar's guide to tracking AI visibility). Free-ish rather than free: the tokens are billed at your provider's prices, which is the same economics described in bring your own API key vs bundled pricing.
What Google Search Console shows you now
Search Console gained a generative-AI performance report announced in June 2026: impressions for your pages inside Google's AI surfaces, broken down by page. If it is in your account, turn it on today and set a baseline, because it is the only free report here that comes with history.
Google's announcement is the Search Generative AI performance reports post. Reporting on the rollout adds the caveats that matter: impressions only for now, with no click-through or query data yet, Google's AI surfaces blended together rather than split, and a gradual rollout that reached a subset of sites first. Check the report before promising anyone a trend from it.
The boundary to hold in your head: this covers Google's own AI surfaces and says nothing about ChatGPT, Perplexity, Gemini, Grok or Claude. Those answers only exist if you ask them, which is what the manual audit does.
When free stops being enough
Free breaks at a predictable point: the prompt list outgrows what one person will faithfully paste, a second person starts asking for the numbers, or someone upstream wants a report that cannot be built by hand without rounding away the sample sizes.
Prompt count is usually the first wall. Past roughly fifteen prompts on two engines, the weekly hour quietly gets skipped, and a skipped week is worse than no data because the series develops a hole nobody notices until the deck. The second wall is people: spreadsheets do not survive a colleague sorting the wrong column, and two versions of the truth end the usefulness of the audit. The third is reporting, where a monthly obligation turns a careful sheet into a chore performed at 11pm.
A warning that applies whichever route you take: never present a small sample as a big number. Count what you measured, say how many answers it came from, and let a thin week look thin.
The obvious next step
When the spreadsheet starts groaning, advisible runs on your own provider keys: no seat limit, and daily runs on ChatGPT, Perplexity, Gemini, Grok and Claude that cost only what your own provider charges for the tokens.
- Bring your own key: runs call the provider with your workspace's key, so token cost stays at the provider's published prices with no markup on top.
- Unlimited team members, so the second-person wall from the spreadsheet never arrives.
- Every metric declares its sample size, and the answer archive keeps the original text behind each number.
- The public demo runs on Sample data: generated by advisible for the demo, not a reply from a real AI engine. The demo shows the shape of the product; your own key shows your own category.
The prompts you wrote for the manual audit carry straight over, which is the real payoff of doing it by hand first. How to pick them well is covered in which prompts to track, and the buying criteria when you get there in how to choose an AI visibility tool.
Common questions
What is the fastest free way to check AI visibility?
Write 10 to 15 questions your buyers actually ask, run them on ChatGPT and Perplexity this week and next, and log mention, position and cited sources in a sheet. Two afternoons gives you a baseline you can defend.
How often should a free audit run?
Once a week on the same day, with the same prompts. Answers vary run to run, so one check is a snapshot and the trend across weeks is the measurement.
Does Google Search Console show ChatGPT or Perplexity mentions?
No. Its generative-AI report covers Google's own AI surfaces and reports impressions for your pages there. Chatbot answers have to be checked separately, by asking them.
Are the free AI visibility checkers worth using?
As a snapshot, yes; as a series, no. Most run once behind an email form, so treat the output as one moment in time and check which prompts and engines it actually ran before quoting the number.
When should I move off the spreadsheet?
When prompt count passes what you will faithfully paste, when a second person needs the numbers, or when someone asks for a report you cannot build by hand without lying about sample sizes.
How to track brand mentions in ChatGPT
Three ways to track whether ChatGPT mentions your brand, from manual checks to daily API runs, and what each one costs you in time and accuracy.
Which prompts to track for AI visibility
How to build an AI visibility prompt set from real buyer questions, how many prompts to start with, and why branded prompts belong on their own line.
How to choose an AI visibility tool
A checklist for buying AI visibility software: engine coverage, collection method, sample sizes, prompt limits, pricing model, and the questions to ask in a demo.
What is AI search visibility?
AI search visibility is how often your brand shows up when buyers ask ChatGPT, Perplexity or Gemini about your category. What it means and how to measure it.
Track this for your own brand
advisible runs your prompts on ChatGPT, Perplexity, Gemini, Grok and Claude every day, and every number shows how many answers it is based on.