By Aleksey Klempner, CTO.LA · Last updated: October 20, 2026
Our panel: 12 questions × 4 engines
We test our own site the same way we suggest you test yours.
- 12 questions in four groups: brand (3), Los Angeles (2), "best in LA" (1) and topics (6), such as AI website audits, agentic commerce and e-commerce CTO help.
- 4 engines: ChatGPT, Perplexity, Gemini and Google AI Overviews. That's 48 answers per run.
- Logged out where possible (exception: on Oct 16 one engine, Perplexity, was signed in), a new chat for each question, with a screenshot of every answer.
- A hit means CTO.LA is named in the answer text. Being listed only as a source is recorded separately.
Results: baseline and first re-run
| Question group | Oct 16, 2026 (4 engines) | Oct 16, same 3 engines | Oct 19, 2026 (3 engines) |
|---|---|---|---|
| Brand | 11/12 | 8/9 | 9/9 |
| Los Angeles | 8/8 | 6/6 | 4/6* |
| Best in LA | 2/4 | 1/3 | 2/3 |
| Topics | 0/24 | 0/18 | 0/18 |
On Oct 19 one engine asked us to sign up before answering, so its 12 answers are n/a and we compare only the same 3 engines. *Neither run was confirmed as located in LA: on Oct 16 Google placed the connection near Washington, DC (Hagerstown, MD), and on Oct 19 the connection was placed there too, so the 8/8 isn't LA-confirmed either. The two Oct 19 LA misses may come from that location or the headless browser setup, so we don't count them as a drop until a run from LA confirms them.
What the data says
- Brand is strong, topics are not. Engines know who we are, but named us in none of the topic answers in either run.
- "Source" isn't "named". On the e-commerce question, two engines showed our page as a source in both runs (three on Oct 16), but didn't name us in the answer.
- Location changes answers. Test local questions from the place your customers are.
- Method matters. Changing browser, login state or location changes results, so keep them fixed and note any change.
How to run your own AI visibility test
- Write 10–15 questions your buyers ask: your brand name, your service + city, "best … in [city]", and your main topics.
- Pick your engines and use them the same way every time (logged out, new chat per question).
- Record each answer: named in the text, listed as a source, or absent. Save a screenshot.
- Score by group (brand, local, topic), not as one total.
- Fix the gaps: direct answers, structured data, consistent facts and outside mentions. See answer engine optimization.
- Re-run on a schedule with the same method and compare like with like.
FAQ
How do I check my brand visibility in ChatGPT?
To check your brand visibility in ChatGPT, ask it logged out, in a new chat, what your company is and who it serves. Note whether it names you, describes you correctly and links to your site.
What is an AI visibility score?
An AI visibility score is the share of test questions where an AI engine names your business. It only means something if the questions, engines and method stay the same between runs.
How often should I test AI visibility?
Test AI visibility on a regular schedule, such as weekly or monthly, and after big site changes. AI answers vary, so one run is a snapshot, not a trend.
Is there a free AI visibility audit?
Yes. The CTO.LA AI website audit is free. It checks how AI engines can read your site and where to improve.
Why does my site show as a source but not in the answer?
Engines sometimes use a page as evidence without naming the business. Clear answers, consistent company facts and outside mentions make it more likely you're named.
Find out how AI engines see your site. Run the free AI website audit
Data: CTO.LA citation panel, Oct 16, 2026 baseline and Oct 19, 2026 re-run (first-party, screenshots on file).