Here is the uncomfortable part: you will never see it happen.
Someone opens ChatGPT, Claude, or Perplexity and types "what is a good tool for X." The model answers with two or three names and a line of praise for each. The person picks one and signs up. No search results page. No click on your site. No row in your analytics. If your name was not in that answer, you did not lose the deal, you were never in the room.
Google Analytics cannot show you this. Search Console cannot show you this. You can rank #1 on Google for your main keyword and still be completely absent from the answer an AI gives to the exact same question. Ranking and being recommended are now two different games, and most makers are only playing the first one.
So the real question is not "where do I rank." It is "when a machine is asked to recommend something in my category, does it say my name, a competitor's, or nothing at all." Here is a free way to find out. It takes about twenty minutes and needs no tools you do not already have.
The 20-minute audit
You are going to interrogate the AI models the way your future customer would, then log what comes back. That is the whole method.
1. Write your question list
Think like a buyer, not a founder. Buyers do not search your product name, they search their problem. Write 8 to 12 real questions someone would ask before they know you exist. For example:
"what is a good tool for [the job your product does]"
"how do I [problem your product solves] without [common pain]"
"best [category] for [specific audience]"
"alternatives to [the big competitor everyone knows]"
"is there an open source way to [thing you do]"
The last two matter most. "Alternatives to [big name]" is where challengers get discovered, and if you are not on that list, you are invisible at the exact moment someone is ready to switch.
2. Ask every model, cold
Open ChatGPT, Claude, Perplexity, and Google's AI mode in separate tabs. Use a fresh chat with no history so your own past prompts do not bias the answer. Paste each question, one at a time, into each model.
Perplexity is the one to watch closely, because it shows its sources. When it answers, look at the citations. Those linked pages are literally the pages the AI trusted enough to build its answer from. If none of them are yours, you now know exactly where the gap is.
3. Log four things per answer
Open a plain spreadsheet. One row per question, per model. Record only four columns:
Mentioned: did your product name appear at all? yes / no
Cited: was your own site linked as a source? yes / no
Sentiment: if mentioned, was it positive, neutral, or a throwaway line?
Competitor: which rivals got named instead?
Do not overthink it. Twenty minutes of honest yes/no logging tells you more about your AI visibility than any dashboard.
Reading the result
Once the grid is filled, you will land in one of three situations, and each means something different.
Not mentioned anywhere. The model does not know you exist as an answer to that problem. This is almost never about your homepage copy. It is about the model never having seen your product described, in plain language, on pages it trusts. The fix lives off your own site as much as on it: comparison pages, roundups, forum threads, and your own content written so a machine can lift a clean, self-contained answer out of it.
Mentioned but not cited. The model says your name but links a competitor or a third-party roundup as the source. You are in the conversation but you do not control the framing. Here the work is making your own pages the most quotable source about you, so the citation points home.
Competitor named, you are not. The most useful and most annoying result. Look at who keeps appearing. Then go read the pages Perplexity cited for them. That is a map of exactly where they earned their visibility, and exactly where you need to show up.
Make it a baseline, not a one-off
Run the same list next month. AI answers drift as models retrain and as new content gets indexed, so a single snapshot is a photo, not a trend. Keep the sheet, date each run, and watch the "mentioned" and "cited" columns move over time. That drift is the only number that tells you whether anything you do is actually working.
Where this goes next
This manual audit is the diagnosis. It shows you the gap with no special tooling, and for a lot of makers, just running it once is the wake-up call.
Turning it into a repeatable system, knowing the specific pages to build so an AI cites you, and verifying in Search Console that AI crawlers are actually reading them, is the longer game. I put that full verify-and-monitor playbook into a field guide called Crawled, Indexed, Cited https://nataliiap.gumroad.com/l/seo-monitor, part of a three-book set on AI search visibility for web apps. But run the twenty-minute check first.
