The Answer Engine Doesn't Know You Exist: A Visibility Audit for the AI Search Era
Photo: Dietmar Rabich, CC BY-SA 4.0, via Wikimedia Commons
The click that quietly stopped happening
Picture a homeowner in Denver at 9:40 on a Tuesday night, standing next to a furnace that's making a noise it shouldn't. A few years ago she'd have typed something into Google, scrolled past the ads, and clicked three tabs. Tonight she opens ChatGPT and asks for the best commercial HVAC shop in the metro that handles rooftop units and answers the phone on weekends.
The model names three companies. None of them is yours. Not because your rankings are bad — they're fine. Because the model has no confident reason to believe you exist in that category.
That's the whole problem in one sentence: search used to hand out links, and now a growing slice of it hands out answers. Links reward pages that rank. Answers reward businesses that are legible, corroborated, and easy to quote. Those are related skills, not identical ones.
What an answer engine is actually doing
Strip away the branding and these systems run a short loop: retrieve relevant material, weigh whether it's trustworthy, then write a response that often names sources. When your name shows up in that response, it's usually because several independent signals agreed with each other. When it doesn't, at least one of them is missing.
The signals that matter most for a small or mid-sized business:
- Entity consistency. Your business name, address, phone number, hours, and service area should read identically everywhere they appear — your site, your Google Business Profile, Apple Maps, Bing Places, Yelp, industry directories, your own social bios. Models resolve an entity by cross-referencing; contradictions create doubt, and doubt means omission.
- Machine-readable structure. Organization and LocalBusiness schema, clean heading hierarchy, and content that lives in HTML rather than being injected by scripts after load.
- Citable expertise. Paragraphs that answer one specific question in plain language, in a form a model can lift with a citation. A 900-word essay on the philosophy of comfort is less useful to the machine than a clear 60-word answer to whether a heat pump works in a Colorado winter.
- Third-party corroboration. Directories, review platforms, local news, trade associations, supplier partner pages, podcast transcripts, and niche community threads. Models are conservative about naming an unknown company; they're much bolder about repeating something three other sources already say.
- Category clarity. If your homepage leads with poetry and buries the actual service list four scrolls down, you've made classification hard. Say what you do, where, and for whom, near the top of the page.
Pass one: the entity audit
Build a spreadsheet with one row per source and columns for name, address, phone, hours, and services. Fill in twelve sources: your site, Google Business Profile, Bing Places, Apple Maps, Facebook, Yelp, your two biggest trade directories, and four others your customers actually use.
You will find at least three problems. A discontinued suite number. An old tracking number nobody forwards anymore. Saturday hours you dropped in 2023 that still live on a directory you never claimed. Fix them all in the same week, and standardize the format — no more alternating between Street, St., and St without the period.
Then add JSON-LD structured data for Organization and LocalBusiness, including sameAs links to your verified profiles. Run the page through Google's Rich Results Test and validate it before you call it done.
Pass two: the question audit
Write down twenty-five questions your buyers actually ask. Not keywords — questions, in the words a real person would use at 9:40 on a Tuesday. Imported equipment availability. Financing. Whether you handle after-hours emergencies.
Now check whether any page on your site answers each one directly. Build answer blocks: an H2 phrased as the question, followed by a direct response in the first two sentences, followed by the nuance. This is good writing for humans and good raw material for machines, which is a rare alignment.
Pass three: check the doors
A surprising number of sites accidentally lock the bots out. Look at your robots.txt and confirm you aren't blocking GPTBot, Google-Extended, PerplexityBot, ClaudeBot, or Bingbot. That last one matters more than people realize, because several assistants lean on Bing's index. Also confirm your sitemap is current and that your key service pages render their content without JavaScript.
Pass four: earn mentions outside your own walls
Aim for ten new third-party mentions in ninety days. Join the local chamber and get on the member directory. Sponsor the association golf tournament if that's your scene. Pitch a supplier's partner page. Offer an expert quote to a journalist. Appear on a podcast and make sure the transcript goes up. None of this is glamorous, and all of it is the raw material an answer engine needs to feel safe naming you.
Proving it: tracking AI-referred traffic
Owners ask how to justify the work. Start with GA4: create a segment for sessions whose referrer contains chatgpt.com, perplexity.ai, gemini.google.com, copilot.microsoft.com, or claude.ai. You'll also see spikes in direct traffic landing on pages you never promoted, which is often an AI answer sending someone to your site without a clean referrer.
Second, add one free-text field to every form: how did you first hear about us. Include an AI assistant option. The response quality is imperfect and the pattern will still surprise you.
Third, keep a prompt log. On the first of each month, run twenty standard prompts across three tools and record whether you're named. It's crude, it takes fifteen minutes, and it's the closest thing you'll get to a rank tracker for a system that doesn't publish rankings.
What this is not
It isn't a reason to chase whatever trick is trending on LinkedIn this week. Stuffing pages with question headings and calling it AI optimization is the same mistake as keyword stuffing, one layer up. The systems reward clarity, consistency, and being talked about by people who aren't you. That was already the job.
The oracle only repeats what the village already says. Be legible, and it will say your name.