Skip to content
Built for AI agentsConnect yours

About

RankHound: who builds it, and the rules we keep

RankHound tracks your AI mentions and helps you get recommended by ChatGPT, Claude and Google AI Overviews. It tracks brand mentions and citations by ChatGPT, Claude, Gemini, Perplexity, Google AI Overviews and Google AI Mode, tells you what to change, and shows whether it worked. We built it because we wanted to know why assistants recommended our competitors and not us, and what to do about it, and everything we tried either guessed or flattered.

Start 7-day free trialRead the methodology

Card required, nothing charged for 7 days

The one feature we will not build

Why we refuse the single score

It is the most requested feature. A single AI visibility score is easy to sell and easy to put on a slide. It is also close to meaningless, because the engines disagree with each other constantly, and averaging that disagreement produces a number describing no real surface.

It hides the one thing worth knowing.
Which engine you are losing, and why.
So every figure is per engine.
With the sample size and the collection method attached, so a number never hides where it came from.
And the rules are written down.
The methodology sets out every one of them, so you can hold us to them.

On the record

What we will not do

A short list, and we would rather publish it than be asked. You can watch the third one hold in the product: when a collection fails, the Answers screen says so, and that answer is left out of every denominator.

  • No invented reach

    We will not estimate how many people saw an AI answer. Nobody outside the assistant vendors knows, and a confident figure would be invented.

  • No percentages on tiny samples

    Below five answers you get the fraction, such as 3 of 4, never a percentage.

  • No blame for our outages

    We will not count our own collection failures as your absence. A probe that fails leaves the denominator.

  • No placement promises

    We show you where to act and whether it worked. The work is yours, and anyone guaranteeing placement in a generated answer is selling something they cannot control.

The founder

Why we built RankHound

It is built by Jamie Partridge, who also writes everything published here, so the site is held to the same rules as the product. He built it because he wanted it: something that shows which answers name a competitor, which pages the engines read, and the arithmetic behind every number it prints.

Jamie Partridge, Founder of RankHound

Founder, RankHound

Jamie Partridge

Builds RankHound. Spends most of his time reading AI answers that name someone else, and working out why.

About JamieLinkedInX

Before you sign up

Who it is for, and who should not buy it

Worth being direct, because a refund on day two helps nobody.

Brands that want to get recommended by AI.
Founders, marketers and agencies who want the evidence behind a number rather than a score.
Teams that want to read the answers.
Every rate opens onto the stored answers behind it, with the sample size beside it.
Agencies with several brands.
Every plan includes 5 brands, each measured on its own.

Not a fit if you need

  • One number for a board deck

    The product is built to refuse a blended score, so you will find it frustrating.

  • Agent traffic, shopping or content

    Agent traffic analytics, shopping visibility and content generation are not here.

  • History from before you signed up

    There is no backfill. Measurement starts the day you do.

  • The whole category from 2 prompts

    2 prompts is about 256 answers a month: plenty of signal on those, but a corner of the category rather than its shape.

Where we are

A UK company, measuring the market you choose

RankHound is a trading name of AIMENTIONTRACKER LTD, registered in England and Wales. AI answers vary by where the question is asked from, so a workspace measures one market rather than an average of several. You choose which during onboarding and can change it later; the default is the United States.

Any market, one at a time.
Any of 212 countries, in any of 128 languages. We would rather measure one market properly than blend two into a figure that describes neither.
A second market is a second workspace.
Never a combined number: a second market means a second brand, and every plan includes 5 brands.
We read everything.
If you want to talk to us, get in touch.

Company number 17456072, incorporated 14 September 2026. Registered office: 71-75 Shelton Street, Covent Garden, London, WC2H 9JQ, United Kingdom.

The honest state of things

Where things stand today

RankHound is new, and we would rather say so than dress it up.

New, and saying so.
AIMENTIONTRACKER LTD was incorporated on 14 September 2026.
Formerly AIMentionTracker.
RankHound was called AIMentionTracker, at aimentiontracker.ai, until 18 October 2026. Same product, same team, and every old address forwards to the same page here.
Live, with no waitlist.
Anyone can sign up today, and every plan carries all six engines.
Not built yet.
White labelling and historical backfill. Measurement starts the day you do.

This site is the proof

  • Plain HTML at build time

    Every page is real HTML when it is built, so an assistant can read it without running a script.

  • Every crawler welcome

    robots.txt allows every crawler and names the AI ones, rather than asking them away.

  • An llms.txt to quote

    It states what the product does, what it costs and what it does not cover, in sentences a model can repeat.

Read ours

7-day trial

Hold us to these rules on your own brand

Add your brand and the prompts your buyers type, see who AI recommends in your place, and open any number to read the answers behind it.

Start 7-day free trialRead the methodology

Card required, nothing charged for 7 days.

On every plan

  • Every answer stored in full, with its citations.
  • Every rate shown with the number of checks behind it.
  • Cited-but-not-named reported as its own state.