AEO for local businesses
When someone asks an AI assistant who to call, the reply names a few local businesses and stops. This page explains why AI visibility behaves differently for local businesses, what is worth measuring, and the checks you can run yourself this week.
Why AI visibility behaves differently here
Assistant answers are a short list of named businesses, and you cannot see the answer someone else received. Absence is silent: nothing notifies you that a competitor across town was recommended instead.
The answer is assembled from surfaces you only partly control. Your own site feeds it, but so do business profiles, directories and review platforms, each with its own copy of your facts.
Small facts do the damage. Hours, phone numbers, services and service areas change, and an answer that repeats last year's hours sends a real person to a locked door.
One overall number hides the picture. Visibility differs by engine and by question, so an average can look fine while the one question your customers actually ask returns somebody else.
What is worth measuring
Each of these is a real measurement, not a modeled score. How each one is produced, sample sizes and intervals included, is documented on how we measure.
Citation share on local-intent prompts
Sample the questions people in your area actually ask across Perplexity, Google AI Overviews and ChatGPT, and record how often your business is named. Every share carries its sample count and a 95% Wilson interval, because thin local samples are exactly where noise masquerades as change.
Answer accuracy against your own facts
Declare your ground truth once (hours, phone, services, service area) and sampled answers are checked against it. A wrong opening time in an AI answer is a measurable defect, not a mystery.
Crawler access on your site
Read your robots.txt per named AI crawler and report who is allowed in. A site that blocks AI crawlers is asking assistants to describe it from third-party surfaces only.
The gap list
Local prompts where nearby competitors are cited and you are not, from the same sampled answers, so you can see which questions are being answered without you.
AI referral traffic
Visitors who arrive from AI assistants, tracked as their own referral class, so recommendations you cannot otherwise observe leave a measurable trace on your own site.
What you can do yourself, starting today
- 1
Ask an assistant what a customer in your town would ask, in their words, and do it more than once and on more than one engine. Note who is named each time.
- 2
Write your ground-truth facts down once (hours, phone, services, service area) and check every public profile and directory listing against that sheet. Fix the mismatches at the source.
- 3
Put the same facts on your website in plain text and LocalBusiness schema, not only inside an image, a PDF menu or a social post.
- 4
Keep the profiles that answers draw from current, and respond to reviews. Those surfaces feed answers alongside your own site.
- 5
When you want to know what assistants actually say about you, measured with sample sizes, run our free measured audit at /audit. No signup required.
Every number the audit returns is measured or marked not measured. Before you trust any tool with this job, ours included, read measured vs asserted.
AEO guides for other verticals: Shopify and ecommerce · SaaS