MentionMachine is a scanner for the question "does AI know my business exists?". The free part reads a website the way an AI crawler would and marks eleven things as fine, partly there or missing, such as whether GPTBot and PerplexityBot are allowed in, whether there is an llms.txt, and whether the address and opening hours are in structured data. Each gap comes with code to paste. The paid part is a one-off licence, $99 or $149, that asks ChatGPT, Claude, Perplexity, Gemini and Google AI Overviews real customer questions and saves the answers so you can see whether you are named.
Who it is for. MentionMachine is built for owner-run service businesses and the small agencies that look after them: plumbers, dentists, salons, detailers, accountants. These are businesses that get most of their work from people looking for a service nearby, and those people are starting to ask an AI assistant where they used to type into a search box. The owner usually has no idea what the assistant says, and no time to learn what structured data is. So every result is written in plain language, with the exact file to hand to whoever looks after the site.
What the free scan checks. It reads up to five pages over plain HTTP, the way a crawler does, and looks at eleven things: whether the answer fetchers used by ChatGPT and Perplexity are allowed in, whether training crawlers are blocked, whether there is an llms.txt, what structured data the site carries and which business facts it leaves out, whether there are questions and answers on the page, titles and descriptions, headings and image text, share tags, the sitemap, and the basics of https and mobile. The result is a score out of 100 and a short list of jobs, worst first.
What the paid scan adds. The licence is paid once and covers one domain. It asks five engines the questions a customer would actually type, records every answer with its date and model, and marks whether the business was recommended, named, cited or absent. It then generates the fixes from the site's own content: schema, an llms.txt, a robots.txt patch, FAQ answers and page briefs. Run the scan again after the changes and you can compare before and after.
What it has found so far. In October 2026 we ran the free scan over 1,902 local-business websites in the USA and UK. One in four already had an llms.txt, mostly written by the site builder and not by the owner. Fewer than one in a hundred blocked an AI answer engine in robots.txt. Nearly half stated nothing about the business in structured data, and seven in ten had no questions and answers on the pages read. The full report, with every count and the method, is on the site under Research.
