- You cannot grow AI visibility you cannot see. Most brands are flying blind while AI answers quietly reshape their traffic, and the tool vendors want you to believe measurement starts at a paid dashboard.
- It does not. AI visibility is measurable across four layers, crawl, retrieval, citation, and traffic, and every layer has a free instrument: Microsoft Clarity, Bing Webmaster Tools Citation Share, a prompt library, and GA4.
- This guide sets up the full free measurement stack in about an hour, and gives you the weekly scoreboard to run it. The free AI Citation Monitor and Citation Share Auditor skills do the analysis for you.
Ask most brands how they are doing in AI search and you get a shrug. Rankings they can quote to the decimal; whether ChatGPT names them or a competitor, no idea. That blind spot has a cost: AI answers are already deciding purchases, and the brands being cited are compounding while nobody else is even keeping score.
Measurement is the E in StudioHawk's SOURC-E framework, Evaluation, and it is deliberately the last letter: AI visibility = (O x U x R x C) / S, measured by E. The other five pillars are the work; Evaluation is how you know whether any of it is working. Skip it and you are optimising on vibes. This guide builds the full measurement stack with free tools, layer by layer.
The AI Visibility Funnel: Four Layers
AI visibility is a funnel, and each layer answers one question. Measure all four or you will fix the wrong thing.
The layers narrow for a reason. Every AI answer starts with a crawl, only some crawled pages get retrieved, only some retrieved pages get cited, and only some citations turn into a visit. A weak number at one layer explains every number below it, which is exactly what makes this measurable at all.
Layer 1: Are AI Bots Crawling You?
If GPTBot, ClaudeBot, and PerplexityBot never fetch your pages, nothing downstream can happen. This is the first number to pull, and it is free.
Two instruments. First, Microsoft Clarity's AI bot reporting shows which AI crawlers and assistants hit your site, at what volume, on which pages. Second, if you are behind Cloudflare, its bot analytics break out verified AI crawlers by name. Before either, run the free AI Crawler Access Checker to confirm you are not blocking the bots at robots.txt without knowing it, which is more common than you would think.
Layer 2: Are You Being Retrieved?
Crawled does not mean used. The retrieval layer asks whether AI systems actually pull your content into the answer pool for the queries you care about, and Bing is currently the only engine that hands you this data.
Bing Webmaster Tools' AI Performance report shows the grounding queries where your content was pulled into Copilot and Bing AI answers, grouped by topic and intent. Our Bing Citation Share guide walks the whole report. Do not dismiss it as "only Bing": it is the same retrieval mechanics every engine runs, and it is the only free window into them.
Layer 3: What Is Your Citation Share?
The metric that matters most: of the AI answers in your topic space, what share cites you rather than a competitor? This is the AI-search equivalent of share of voice, and you can measure it two free ways.
First, Bing's Citation Share figure gives you the exact percentage for grounding queries, and the free Citation Share Auditor skill reads the export, buckets every query as owned, contested, or absent, and gives one move per opportunity. Second, run a prompt library: pick ten queries a real buyer would ask, ask them monthly in ChatGPT, Claude, Perplexity, and Google's AI Mode, and log who gets named. The free AI Citation Monitor skill structures the whole protocol for you. Ten prompts, four engines, once a month: that is a real citation-share trendline for zero dollars.
Layer 4: Is AI Sending You Traffic?
The bottom of the funnel is revenue-adjacent: visits that arrive from an AI answer. GA4 has the data, but it hides some of it.
Referral traffic from chatgpt.com, perplexity.ai, and claude.ai is visible once you know where to look; our claude.ai referral guide shows the setup, including the desktop-app gotcha that silently undercounts. Then there is the tell most analysts miss: a rise in unexplained direct traffic to deep informational pages often is AI traffic, because several AI surfaces strip the referrer. If direct entrances to a guide nobody types by hand are climbing, an AI is quoting it.
The Weekly Scoreboard
Four layers, four numbers, one sheet. This is the whole operating rhythm, and it takes about twenty minutes a week once the stack is set up.
| Layer | Metric | Free instrument | Cadence |
|---|---|---|---|
| Crawl | AI bot sessions by crawler | Microsoft Clarity, Cloudflare | Weekly |
| Retrieval | Grounding queries by topic | Bing AI Performance report | Fortnightly |
| Citation | Citation Share, prompt-library hit rate | Bing + the AI Citation Monitor skill | Monthly |
| Traffic | AI referrals, direct-traffic tell | GA4 | Weekly |
Read it top down. Citation share flat but crawl volume collapsing? The problem is upstream, at Structure. Retrieval strong but citations absent? Your content gets considered and loses the final cut, which is a Uniqueness problem. The funnel does the diagnosis for you.
Where Evaluation Fits in SOURC-E
Evaluation is the pillar that makes the other five honest. In SOURC-E, AI visibility = (O x U x R x C) / S, and every improvement you make to Offsite, Uniqueness, Relevance, Credibility, or Structure should show up somewhere in the four-layer funnel within weeks.
That is the real reason to build the scoreboard before you build anything else: it turns the framework from a belief system into a feedback loop. Read the full SOURC-E Framework guide for the other five pillars, run the SOURC-E Auditor skill to score all six at once, and check your entity foundations with the Entity SEO guide, because a machine cites what it can measure and trusts what it can identify. You now have the instruments. Most of your competitors are still guessing.
FAQ
How do I measure my brand's AI search visibility?
Measure four layers with free tools: AI bot crawl volume (Microsoft Clarity), retrieval into answers (Bing's AI Performance report), citation share (Bing plus a monthly prompt library across ChatGPT, Claude, and Perplexity), and AI referral traffic (GA4). Together they give you a full picture for zero dollars.
What is citation share?
Of the AI answers in your topic space, the percentage that cite you rather than a competitor. Bing Webmaster Tools reports it directly for grounding queries, and a prompt library extends the same measure to other engines. It is the AI-search equivalent of share of voice.
Do I need a paid AI visibility tool?
Not to start, and usually not for months. The free stack, Clarity, Bing Webmaster Tools, GA4, and a prompt library, covers all four layers of the funnel. Paid platforms add scale and automation once you outgrow a weekly twenty-minute routine, not capability you lack on day one.
Why is my direct traffic rising with no explanation?
Often it is AI traffic in disguise. Several AI surfaces strip the referrer, so a visit from an AI answer lands as "direct". If direct entrances to deep informational pages are climbing, an AI is likely quoting those pages.
How often should I check AI visibility metrics?
Crawl and traffic weekly, retrieval fortnightly, citation share monthly. AI answers move slower than rankings, so monthly trendlines beat daily anxiety. The point is a consistent scoreboard, not constant monitoring.