In about thirty minutes, with nothing but a browser and a terminal, you can get an honest first read on whether AI engines can see, understand, and recommend your business: ask five money questions in three engines, check robots.txt and llms.txt, fetch a page the way GPTBot does, view source for schema, and give your Google Business Profile a once-over. Here's the walkthrough, with a time budget for each step.
Minutes 0–10: Ask five money questions in three engines
Write down five questions a ready-to-spend customer would ask — not your brand name, but category questions where an assistant has to choose whom to recommend: "best [your trade] in [your town]," "how much does [your core service] cost near [your area]," "who can fix [urgent problem] today."
Ask all five in ChatGPT (with web search on), Perplexity, and Google — watching for AI Overviews or using AI Mode. For each of the fifteen answers, jot three things: were you mentioned, were you linked, and who was recommended instead. Read the competitors' wins closely; the sources engines cite for them — directories, review platforms, service pages — are a map of where visibility comes from in your market. This step usually produces the day's biggest surprise, in one direction or the other.
Minutes 10–15: Check robots.txt and llms.txt
Load yourdomain.com/robots.txt in a browser. You're scanning for explicit blocks on AI crawlers — GPTBot, ClaudeBot, PerplexityBot — and for a blanket "Disallow: /" that someone left behind after a redesign. Either can quietly remove you from AI answers regardless of how good the site is.
Then try yourdomain.com/llms.txt. Most sites won't have one, and its absence isn't an emergency — it's a proposed standard: a short, curated summary of your site written for AI readers. If it takes under an hour to add one later, it's worth doing; today, just note whether it exists.
Minutes 15–20: Fetch your homepage as GPTBot
This is the step that catches the silent killers. In a terminal, request your homepage with GPTBot's user agent using curl, and look at two things: the status code and the content. A 403 or a "verify your browser" page means a firewall or CDN is blocking AI crawlers — likely by default, without anyone at your business ever deciding it. A 200 whose body doesn't contain your actual text (search the output for your phone number) means your content is rendered by JavaScript that AI crawlers never execute.
No terminal? A lighter proxy: right-click your homepage, choose "View page source," and search for a sentence you can see on screen. If it's missing from the source, most AI crawlers can't see it either.
Minutes 20–25: View source for schema
While you're in the page source, search for "application/ld+json." You're hoping to find structured data — ideally LocalBusiness or a more specific type, with your name, address, phone, hours, and service area, all matching what the page displays. Schema is machine-readable confirmation of facts the engines are otherwise left to infer, and it serves your classic SEO and your AI visibility at the same time. Finding none, or finding a stale address from two offices ago, goes on the fix list.
Minutes 25–30: Google Business Profile once-over
AI answers about local businesses lean heavily on the local data ecosystem, with your Google Business Profile at the center of it. Spend the last five minutes as a skeptical stranger: is the primary category right and specific? Hours current, including holidays? Phone and website correct? Are recent reviews getting responses? Any user-suggested edits pending that you never approved? A neglected profile undermines everything else, because it's often the most authoritative source engines have about you — this is where local SEO and AI visibility turn out to be the same discipline.
What can't a DIY audit tell you?
Being honest about the limits: this is a snapshot, and AI answers are probabilistic — the answer you got today isn't guaranteed tomorrow, and five questions in three engines is a taste, not a census. A DIY pass can't tell you your share of voice against competitors across the hundreds of question variants customers actually use, can't watch for the regression when a plugin update quietly reinstates a crawler block in October, and can't tell you whether the fixes you make actually move citations over the following months. That ongoing layer — continuous monitoring rather than a Saturday-morning checkup — is what a platform like Speak Local is for.
But don't let the limits talk you out of the thirty minutes. Most local businesses have never run even this much, which means the ones that do start with a real advantage: a concrete list of what's broken, what's missing, and who's eating their lunch in the answers. Set a timer and go find out.
