Search is becoming a machine-readable layer of the internet

For twenty years, being found meant ranking. Now a machine reads the web on somebody’s behalf and hands back an answer with three companies in it. We build the software that tells you whether one of them is you.

Get Started Why we built it

The industry is changing. The software should change with it.

For more than two decades, search was organised around pages, keywords, links and ranked results. You published something, it was indexed, and a person chose from a list.

That model has not been replaced — it has been wrapped. People search through generated answers, assistants and agents that retrieve, compare, summarise and increasingly act on their behalf. The list is still under there. Fewer people reach it.

This changes what visibility means. Ranking for a query is no longer enough. A company also has to be understood correctly, retrieved in the right context, supported by sources the machine trusts, and picked when an answer is assembled.

None of that appears in the tools most companies already own. Analytics does not run for a crawler, so the visit is never recorded. Rank trackers measure a results page fewer people see. And the answer itself — the sentence that names your competitor — is written once, read by one person, and gone.

Growth Mention exists to keep it. What the assistants say about you, every day, in the interface a person would use — and which machines read your site in order to be able to say it. Being named in an answer and being readable to the thing writing it are the same problem seen from either end.

You cannot argue with something you cannot see

What an assistant says about a company is written once, read by one person, and gone. Nothing keeps it. So the first thing worth building was not a strategy — it was a record: the same questions asked every day, the answers kept, and a line you can point at.

Somebody has to stand where they all meet

No single assistant will tell you how you are described by the others, and none of them owes you an explanation. So our whole job is to sit at the point where all of it arrives — five of them, every day — and keep the record long enough for it to be worth something.

What we think this work is

Four positions the company is built on. They are the reason the product looks the way it does, and the reason it does not do some of the things it could.

001

Measure the thing itself

It is always cheaper to measure something adjacent and call it the same. We would rather read the answer a person would actually get than infer it from something correlated with it, even when that makes the work slower and the product smaller. A convenient number that is not the number is the most expensive thing you can put on a dashboard.

002

Be able to say where it came from

Every figure we show should survive the question “how do you know that?” If we cannot point at what it was derived from, we would rather not print it. A figure nobody can stand behind is worse than none — it does not leave you uninformed, it leaves you confidently wrong.

003

Keep the record

Most of what happens here is written once and disappears: an answer given to one person, a page fetched at four in the morning. We think the record is the product. A number tells you where you stand today; the thing it was calculated from is what lets you understand why, months later, when the question finally gets asked.

004

One problem, not two products

Being named in an answer and being readable to the thing writing it are the same problem seen from either end. Splitting them into two tools that never meet would be easier to sell and less useful to own, so they live under one login and describe the same brand, the same rivals and the same pages.

Why we started it

We came to this from the other side of the table — doing the work of getting companies found, and watching the ground move under it. The rankings held. The traffic did not. And the explanation was somewhere none of the tools could look.

The uncomfortable part was how ordinary it felt from the inside. Nobody was doing anything wrong. An assistant had simply read something, formed a view, and repeated it to people who never saw a search result at all — and there was no way to watch any of it happen.

So we built the thing we wanted: something that stands where the machines meet, reads what they say, records what they took in order to say it, and keeps both long enough to be useful. Not a score. A record you can argue with.

It is early, and the ground is still moving. That is rather the point. The companies that will be described well by machines in five years are the ones paying attention now, while it is still possible to see what is being said and do something about it.

Find out what the assistants already say about you

Add your prompts and the first answers arrive within minutes.

Get Started Free