The AI Visibility Self-Test shows real estate agents and loan officers whether ChatGPT and Gemini can actually find and recommend them. You run 3 personalized prompts, 3 times each, in about 10 minutes. Because AI engines name the same business only 7% of the time on repeated queries, a single test proves nothing. Nine runs show you whether you are in the retrieval pool, splitting samples with competitors, or invisible to AI search entirely.
Buyers, sellers, and borrowers now ask AI who they should work with before they ever ask a friend. This test tells you, with your own eyes, whether your name comes up. No software, no email gate, no 47-step course. Three prompts, nine runs, one honest score.
Step 1: Personalize your prompts
Step 2: Run these 3 prompts
Copy each prompt into ChatGPT (and Gemini if you have five extra minutes). Run each one 3 times, each in a brand new chat. That is 9 runs total, and the repetition is the whole point. More on why below.
I'm relocating to [your city]. Which real estate agent should I work with, and why?
Best [your niche] agent in [your city]
What should I know before buying a house in [your city], and who can help me?
Step 3: Record what happens
- Count your appearances. A run counts if the AI names you or your business. A link to your Zillow profile counts as half credit at best. Be honest.
- Note who got named instead. Those names are not your real competition. The sources the AI cited are. Click the citations and look at what those pages do that yours don't.
- Screenshot and date everything. You will want the before picture in 90 days.
Step 4: Score yourself
0 of 9: You are outside the retrieval pool.
AI engines are not snubbing you. They cannot retrieve you. When an engine with browsing searches your market's questions, your name is not in anything it can find and trust, so no amount of posting, rerunning, or waiting fixes this. The problem is structural: your website, your Google Business Profile, and your profiles are not giving machines anything checkable to quote. The good news is that structural problems have structural fixes, and they are faster than building a social following.
First move: find out exactly where the leak is before touching anything. That is what the Visibility Audit does: your name, your site, your Google presence, and your market's actual AI retrieval pool, returned as a prioritized fix list.
Get the Visibility Audit1 to 3 of 9: You are in the pool, splitting samples.
This is further than most of your market, so take the win. AI can retrieve you, but it is resampling sources every query and you are losing most of the draws to whoever is structured better, cited more, and described more consistently across the web. The goal now is winning more samples: a tighter entity (same name, markets, and services everywhere), answer-structured pages on your own site, and presence in the sources the AI cited when it didn't pick you.
First move: make your most important page pass the copy/paste test. Lift its first paragraph into a blank doc. If it doesn't fully answer the question on its own, rebuild it. If you'd rather know your exact fix order, the audit maps it.
Get the Visibility Audit4+ of 9: You are winning your market's samples. Protect it.
You are the business AI keeps finding, which means two things. One, keep feeding the machine: freshness matters, and a stat or page older than a quarter starts decaying as a citation target. Two, you now have a lead everyone else is about to chase, because the advice industry is waking up to AI visibility this year. The move at this level is widening the moat: more question-style pages, more borrowed citations in sources that already rank, and a quarterly re-run of this exact test, screenshotted and dated.
First move: turn this result into marketing. "Ask ChatGPT who to work with in {your market}" is a proof point almost nobody can claim. If you want a partner keeping the engine fed while you sell, that is what the ongoing visibility retainer is for.
Start with the Visibility AuditWhy you run it three times (the research)
A 2026 study analyzed 14,472 AI citations across 1,487 local-service queries in 50 US metros. The finding that breaks most people's assumptions: AI engines named the same business only 7% of the time when the identical query was repeated. Google's local pack is 90% consistent. AI is not ranking businesses. It resamples a pool of retrievable sources every single time and picks from what it finds in that moment.
That same study killed two more pieces of standard advice. The correlation between star ratings and AI citations was 0.013, which is statistically zero, and 14% of AI-recommended businesses had fewer than 50 reviews. And the single most-cited source for local recommendations was not a portal or directory. It was the business's own website, at 59.9% of citations.
Put together: one test proves nothing, reviews won't save you, and your own website is the asset that decides whether machines can recommend you. That is what this test measures.
Questions people ask about this test
Does it matter which AI I use for the test?
Yes, run at least two if you can, because ChatGPT and Gemini cite the same domains only about 8% of the time. A business can be visible in one engine and absent in the other, so a single-engine test gives you a partial picture. ChatGPT plus Gemini covers the two biggest behavior patterns.
I showed up once. Doesn't that mean I'm fine?
It means you are retrievable, which is real progress, but one appearance in nine runs means you are losing most samples to competitors. Because AI resamples sources on every query, the businesses that win are cited consistently across many sources, not found once. Your score tells you how often you win the draw.
My Zillow profile came up but not my name. Does that count?
Barely. A portal profile appearing means the portal is visible, not you, and the client relationship starts on rented ground where you are one of several agents a lead can be routed to. The goal of visibility work is AI citing your own website and profiles, where the contact goes to you alone.
How often should I re-run the test?
Quarterly, with screenshots and dates each time. AI retrieval pools shift fast: Reddit lost 86% of its ChatGPT citations in a three-week window in 2026, and answer-structured pages went from 2% to 32% of citations in under a year. A quarterly re-run shows whether your fixes are compounding or the pool moved again.
Can I just pay someone to make me show up in ChatGPT?
Be careful with anyone promising placement, because no vendor controls what AI engines cite and nobody can see "AI search volume" data, which does not exist publicly. What a legitimate partner can do is measurable: fix your entity consistency, restructure your pages so machines can quote them, and get you into the sources engines already retrieve. That is the work behind the Visibility Audit.
Research sources: Steady Demand AI Citation Ledger, 14,472 citations across 1,487 local queries, July-August 2026. ChatGPT citation pattern shifts (Reddit decline, answer page growth, site-scoped search frequency) from 2026 industry citation studies. Results vary by engine, market, and month, which is exactly why the test uses repeated runs.