How to check if your site shows up in AI search
You can check your AI-search visibility in about 30 minutes with tools you already have. The 4 steps: read your analytics referrals for the AI domains, chatgpt.com, perplexity.ai, gemini.google.com, claude.ai. Read your robots.txt for blocks on each engine's search crawler. Run a site: search on Bing to confirm index presence. And set up a fixed monthly spot-check of real buyer questions. Before paying for a tracker, note the category's own critics ask the right question: “Where is this score coming from? What is it based on?” (Geeky Tech podcast).
Step 1: referral traffic, per engine (10 minutes)
In GA4, open Reports, Acquisition, Traffic acquisition, and filter session source for each AI domain. Then break down by landing page. This tells you 3 things no paid tracker can improve on: whether AI engines cite you at all, which engines, and which pages earn it. Record the monthly numbers somewhere boring, like a spreadsheet. Trend beats snapshot.
Step 2: crawler access, per bot (5 minutes)
Open yourdomain.com/robots.txt. Look for rules affecting OAI-SearchBot, PerplexityBot, Claude-SearchBot, and Googlebot, and for wildcard rules that catch them accidentally. The training-versus-search bot distinction trips almost everyone, explained in our robots.txt walkthrough. An accidental block here zeroes out an engine no matter what else you do. This is the highest-stakes 5 minutes in this protocol.
Step 3: index presence (5 minutes)
Search site:yourdomain.com on Bing and on Google. ChatGPT search leans on Bing, Gemini grounds in Google. Thin presence in either index caps your ceiling in the engines downstream of it. If Bing barely knows you, fix Bing indexation before anything fancier.
Step 4: the monthly spot-check (10 minutes)
Write 5 questions your actual buyers ask, in their words. Ask them in ChatGPT with search, Perplexity, and Google, in fresh sessions, and record who gets cited each time. Keep the wording identical month to month. Single runs mean little, because cited sources churn heavily between months per industry reporting (Search Engine Land). A logged series smooths the noise into a real trend, and seeing which competitor gets cited tells you what the engines currently prefer for your topic.
When a paid tracker earns its keep
Trackers sample engines at scale and chart share of voice. That is real value for enterprises reporting to boards, at prices running from tens into hundreds of dollars monthly, per engine in some cases, covered in our Brand Radar breakdown. For an owner, the free protocol above answers the practical question, and the money is better spent fixing what it finds. Visibility follows readiness: crawler access, rendering, liftable answers, clean entity data. That readiness layer is what our $10 audit checks, finding by finding, with the fix attached.
Frequently asked questions
What counts as good AI referral traffic?
There is no honest benchmark. Baselines differ wildly by niche. Measure your own trend: this month versus last, and which pages move. Anyone quoting a universal number invented it.
How often should I run the spot-check?
Monthly. Weekly amplifies noise, quarterly misses shifts. Calendar it with the same 5 questions and log the results in the same file.
Can AuditLamp track my AI mentions?
No, and we say so plainly. We audit readiness, the site-side half you control. The measurement half is your analytics, which is free and more truthful than anything we could resell you.