The AI visibility tab

The AI visibility tab measures whether AI assistants cite your site. You declare the questions your customers would ask an AI chat; Snorklee actually asks them to the engines, at the frequency you choose, and checks each time whether your site appears in the answer. Crossed with the AI presence card in the Analytics tab (the real visits coming from AI chats), it closes the loop: from citation to visit.

The tab is restricted to the site owner.

#The concepts

• Tracked question — a natural-language question, 8 to 300 characters, phrased the way a user would ask it (e.g. "which GDPR-compliant analytics tool would you recommend?").
• Check — 1 question asked to 1 engine, 1 time. A question tested on 5 engines therefore uses 5 checks. Your plan grants a monthly check allowance (see Plans, trial and checks).
• Cited — your domain appears in the sources the engine cites, or is mentioned in the answer text. Subdomains count (blog.example.com cites example.com).

#The 5 engines

Each check queries the available AI answer engines: ChatGPT (OpenAI), Claude (Anthropic), Perplexity, Gemini (Google) and Mistral. The first four browse the web and return a source list; Mistral answers without browsing, so detection there relies solely on your domain being mentioned in the text. An engine marked inactive is temporarily unavailable on the service side — the others keep running.

Only the question you declared is sent to the engines. No visitor data ever transits through these calls.

#Adding questions

Two ways, from the "Add questions to monitor" block:

1. Suggest from Search Console & Bing — Snorklee reads your real search queries (external sources to connect from the Integration tab) and rephrases them as questions people would ask an assistant. Each suggestion shows the original query and its clicks; click Track to adopt it. Limited to 5 suggestion requests per hour.
2. Add manually — type your question and click Add.

The maximum number of tracked questions depends on your plan (the "{used} / {max} tracked questions" counter at the top of the tab). Remove deactivates a question without erasing its history; re-adding the exact same question reactivates it.

#Per-question frequency

Each question has its own automatic check frequency: Daily, 2×/week, Weekly (default) or Monthly. This is your allocation lever: a strategic question daily, background questions monthly. The "≈ n checks/month scheduled" line estimates what your current setup will use.

#On-demand checks

• Check now (per question) — runs the check immediately on all active engines. Anti-burst: a 10-minute cooldown between two runs of the same question.
• Check all — launches, in the background, every active question not tested in the last 10 minutes. Results stream in row by row.

#Reading the results

Each question shows its latest result per engine:

• Your site is cited — with the position among the sources and an excerpt of the sentence that mentions you.
• Your site does not appear in the answer — where applicable with the domains cited instead of you: that is your competitive worklist.
• Engine not configured / error — the check could not complete on that engine.

At the top of the tab, four indicators aggregate everything: AI visibility (the percentage of checks that cite you), Average position among cited sources, Tracked questions and Active engines. Below: the "Visibility over time" curve (90 days), the "Who gets cited (share of voice)" card — the domains dominating the answers to your questions — and the citation rate by engine.

#The monthly allowance

The "Checks this month: used / plan" gauge tracks your allowance, packs included. When it runs out, checks pause until next month, until a plan upgrade or until you buy a pack — details here. Audience measurement itself is never interrupted.