An AI visibility audit sounds like something you pay an agency six figures for. You don't. With a free checker and a fixed set of questions, you can snapshot how ChatGPT, Gemini, Claude and Perplexity see your brand in about ten minutes — and if you rerun it on a schedule, it actually tells you whether your work is moving the needle.
Here's the workflow we use, written so you can copy it.
Step 1 — Lock your "truth" fields
Before you run anything, write down three things and don't change them week to week: your exact business name, your category, and your city. These become the fixed inputs for every audit. If you change them mid-stream, you can't tell whether a score moved because of your work or because you changed the question.
Step 2 — Build a short, fixed query set
Grab the AI Visibility Checker and let it generate the buyer-intent prompts from your profile — that covers the "who are you, where are you, are you any good" questions automatically. On top of that, add five or six real questions your customers actually type. Pull them from call logs, email, or your search console. Write them down. Keep the same set each month.
The key discipline: a fixed query set. If you keep changing the questions, you're measuring chaos. Same questions every month, then you can see real movement.
Step 3 — Run the checker and record the baseline
Paste your URL and fields, run it, and note three numbers for each engine:
- Did it mention you at all?
- If yes, did it match you on your business name or your domain?
- What did it say — favourable, neutral, or ignored?
Also record the website score, the AI score, and the overall grade, plus the list of failed checks. That snapshot is your baseline. Everything after this is compared to it.
Step 4 — Turn the fails into a fix list
The failed checks are your to-do list, in priority order. Don't touch the passes. Your first audit session isn't about fixing everything — it's about identifying the five or six most likely blockers, usually things like missing structured data, inconsistent business info, or thin content pages.
Step 5 — Rerun the same check next month
This is the step everyone skips, and it's the one that makes the audit worth anything. At the end of the month, run the exact same query set and compare.
- Did a fail flip to a pass? Good — that change was real.
- Is an engine mentioning you that didn't before? That's attribution working.
- Did the website score climb but the AI score stay flat? That tells you the problem is content and entity clarity, not technical setup.
You're not looking for a vanity number. You're looking for which lever moved, so you know what to pull next.
The honest take
A ten-minute audit won't replace a deep dive, but it gives you something most brands don't have: a baseline they can actually act on. Run it once to know where you stand, then run it every month to see if the work is working.
At Launch at Dawn we turn this into a managed monthly process for brands in Montreal and Vancouver. Book a free website teardown and we'll run your first audit with you, show you the gaps, and leave you with a list of exactly what to change.