How it works

The old game: someone types into Google, gets ten blue links, and scrolls. You fight for position. That's SEO, and an entire industry spent two decades playing it.

The new game: someone asks ChatGPT or Google's AI “who's the best option near me?” and gets one answer, built from two or three sources. Nobody scrolls. The citation is the new ranking, and in most industries, nobody has claimed it yet.

Here's what makes the timing matter: trust compounds with tenure. The websites that grabbed the top of Google 15 years ago are still very hard to unseat; domain age itself became a trust signal. Verified records work the same way. The record you start today will have years of history by the time engines check these signals at scale. The identical record started a year from now won't.

→Can I read you? Your facts have to exist in a form AI tools actually parse: structured data and a machine-readable guide, not just nice-looking pages.

→Can I verify you? Your name, address, and phone need to say the same thing everywhere they appear. Contradictions kill trust; the engine moves on to someone consistent.

→Can I trust you? Fresh, provably-maintained facts beat stale scraps. A verified record with a maintenance history is the strongest signal a business can send.

An llms.txt file is a one-page guide to your business that lives on your website, written for AI tools instead of people. What you do, where you are, what matters, which pages count. It's the file AI crawlers fetch first, and most businesses don't have one. That single missing file is often the difference between being read and being skipped.

You can write one yourself in about 20 minutes; here's the complete guide. On every paid tier, we generate and maintain yours automatically.

Building the record is half the job. The other half is finding out whether it worked, and that is one screen rather than two subscriptions that disagree with each other.

→Measured, from Google.Connect Search Console and sixteen months of Google's own clicks and impressions arrive on day one, and we keep collecting past the window Google itself keeps. Every paid plan.

→Modelled, and labelled as modelled. Search volume and AI-prompt demand are estimates. They sit beside the measured numbers and are never averaged with them or added to them. A term with no estimate shows a blank, not a zero.

→The AI side. Which engines your robots rules let in, and, for the questions you chose, whether your domain was the source and which domains were instead. Untested stays untested; an engine we could not reach stores nothing rather than a false miss.

→The search side. Ranking footprint over time with your tracked competitors on the same axes, the spread of your positions in bands you can click to filter, and a drill-down into any single word: position history, measured clicks, demand on both channels, a twelve-month interest curve with rising queries, and a live search check.

Then you change something and run it again. Because the history is append-only, the comparison is a record of what happened rather than a chart redrawn to agree with the present.

There are two kinds of records, and we're straight about the difference. A self-declared record is you saying who you are: real, useful, free, and you can make one yourself in five minutes. A verified record means someone independent checked your facts and stands behind them: on our top tiers, a human, against multiple sources, on a schedule.