Run your first AI visibility scan
AI visibility measures how often AI assistants mention and recommend your brand when people ask questions in your category. This playbook takes you from an empty project to your first real, statistically-honest visibility score in about ten minutes.
What you'll be able to do: A live AI-visibility score for your brand, measured across the AI engines your buyers actually use.
- 1
Create a project and set the brand website
Open AI Visibility, create a project for the brand, and set its website in the setup wizard. The website matters: it's how CrunchJunkie tells your own citations apart from competitors' and populates Source-performance metrics. Skipping it leaves several metrics empty.
- 2
Add prompts — or let Crunch discover them
Prompts are the questions you want to test, like "best CRM for small agencies". Add your own, or click Suggest prompts and Crunch reads your brand, site and competitors to propose a well-rounded set across Discovery, Brand and Competitor intent. Aim for 15–30 that mirror how real buyers ask.
- 3
Choose engines and a scan mode
Pick which of the 8 engines to track (ChatGPT, Gemini, Perplexity, Claude, Google AI Overviews, Google AI Mode, Grok, DeepSeek). Scan modes bundle sensible defaults: Lite (2 efficient engines, cheap), Balanced (4 key engines — recommended), or Thorough (all 8). The live estimate shows roughly how many checks each scan will use before you commit.
- 4
Run the scan
Click Run scan. Each prompt runs against every enabled engine, several times, so a single lucky answer never looks like a trend. Scans run in the background — you can close the tab and you'll get an email when it's done.
- 5
Read the score honestly
Your headline Visibility % is the share of runs that mentioned you, shown with the exact sample size ("27/50 runs") and a ± margin of error. Share of Voice is your slice of all brand mentions; Average Position is where you tend to appear. Every figure carries how it was measured — so you're never guessing.
- 6
Set a schedule
Turn on a daily or weekly schedule so the score keeps updating without you clicking Run scan. You'll get an email when each scheduled run completes, and the trend charts fill in over time — the slope is what matters, not any single day.
Frequently asked
How is AI visibility different from SEO rank tracking?
SEO rank tracking measures where you appear in a list of blue links. AI visibility measures whether an AI assistant mentions or recommends you at all when it synthesises a single answer — a different question, because the AI answer often ends the buying journey before a results page is ever seen.
How many prompts and runs do I need for a reliable number?
More runs mean a tighter margin of error. Under about ten runs the score carries a low-confidence note. A practical starting point is 15–30 prompts scanned across several engines with repeated runs; the ± figure on every metric tells you how much to trust it.
Does the scan use my personal ChatGPT history?
No. Scans call the providers' stateless APIs with just the prompt and a web-search tool — no account history, memory or personalization — so the measurement isn't biased by any one person's usage. That's also why your score can differ from what you see logged into your own ChatGPT.
What does a scan cost?
On your own API keys, the provider bills you directly and the in-app estimate shows the token usage. On Managed AI, each scan counts against your monthly check allowance. The scan-mode estimate shows the number of checks before you run anything, so there are no surprises.
Try it on your own brand
AI search visibility and white-label reporting in one console — put this playbook to work inside the 14-day free trial.
Start free