Skip to content

Introducing the index: free, open AI-visibility leaderboards

The index ranks which brands AI assistants recommend across eight categories, for free, with the whole scoring method published. See where you stand, and check exactly how we got there.

Published/

Updated/

Read/4 min

Tags/AI visibilitybrand monitoringChatGPTClaude

The index is a set of free public leaderboards showing which brands AI assistants recommend, category by category. It launches across eight categories, and the whole scoring method is published at /methodology so anyone can reproduce a number and check our work.

Most tools in this space will tell you where you stand and charge you for the privilege. Fewer will tell you how they measured it. We think that is backwards: a score you can’t verify is a score you can’t trust.

Why a free, open index

An AI visibility number is only useful if you can act on it, and you can only act on it if you understand it.

So the index inverts the usual arrangement. The measurement is free and the method is open, published in full at /methodology down to the weights and the arithmetic. What indexwire charges for is the harder part: doing the work to move a brand’s position. The finding is the marketing; the fixing is the product.

That split is what lets the index stay honest. When the thing you sell is the measurement, every score has to look valuable, and a flaw that inflates scores is a flaw you can’t afford to admit. When the measurement is free, we can publish whatever the data shows, including the runs where our own methods don’t move the needle.

What the index covers

The index launches with eight categories:

  • screenshot and scraping APIs
  • social scheduling
  • AI transcription
  • product analytics
  • transactional email APIs
  • help desk software
  • applied AI consulting
  • AEO/SEO SaaS

Six of these are categories indexwire sells nothing in. That is deliberate: a board is easier to trust when the people publishing it have no stake in who wins. The other two are categories indexwire competes in, and it appears on those boards under exactly the same method as everyone else.

Each board ranks every brand tracked in the category, including the ones no assistant mentioned. They sit at the bottom, at zero, rather than being quietly dropped. A leaderboard that hides its zeros flatters everyone left on it.

How a ranking is built

A run covers one category for one ISO week. It puts a fixed prompt set to the assistants, records every answer, and scores each brand on how it was named, where it appeared, and whether the mention carried a citation. Each answer contributes a brand’s single best recommendation, and the total is divided by every answer requested, so an answer that ignored a brand counts against it.

The result is a share between 0 and 1, comparable only against the other brands in the same run. The prompt set and the assistant set are frozen when a run starts, and the scoring reads no clock and uses no randomness, so the same answers always produce the same score. That is what makes a board checkable rather than merely published. The full method has the weights and the reasoning behind each one.

We rank ourselves, including the bad weeks

indexwire tracks its own position in the two categories it competes in, and publishes it whether it is going up or down.

This is the part a funded measurement business structurally cannot copy. Publishing a result that undercuts your own product only works when the product is the fixing, not the score. It is the one durable advantage of building the index as marketing rather than as the thing you sell, so the receipts, up weeks and down weeks alike, are the point.

What the index does not do yet

The launch set is eight categories, not a map of the whole software market. We will add categories over time, and each addition is a deliberate choice rather than a crawl, so coverage grows in a direction we can defend.

The prompts are generated rather than drawn from real user conversations. That is a real limitation: a generated prompt set can over-weight some questions and miss others. Freezing the set per run is what makes a score reproducible, and publishing the method is what lets you judge whether the questions represent your category. Where they don’t, tell us.

We publish on a weekly rhythm because a daily one would mostly measure noise. When a methodology change makes older numbers non-comparable, we say so and re-fold rather than editing scores in place.

How to use the index

If you’re a founder or marketer: find your category on the index, check where you rank, and see which brands the assistants reach for instead. The data is free and needs no account.

If you run AEO or SEO work: read the method, run the same kind of prompts against the same assistants, and compare. Because the scoring is deterministic, any gap between your number and ours has a findable cause, and we’d rather hear about it than not.

Frequently asked questions

What is the indexwire index?

The index is a set of free public leaderboards showing which brands AI assistants recommend in a given category. Each board is produced by a documented, reproducible scoring method rather than a proprietary black box, and the full method is published at /methodology.

Is the index really free?

Yes. The public leaderboards are free to read, with no account and no sales call. indexwire charges for the work of improving a brand's position, not for showing where it stands.

Which AI assistants does the index measure?

Three, through adapters for ChatGPT, Claude, and Perplexity. Each run records the set it asked, and in the current methodology version every assistant counts equally. Adding an assistant later changes the answers, not the arithmetic.

How is a visibility score calculated?

A run puts a fixed set of prompts to the assistants for one category over one ISO week. Each answer contributes a brand's single best recommendation mention, weighted by how it was named, where it appeared, and whether it was cited, and the total is divided by every answer requested. The result is a share between 0 and 1. The full method is at /methodology.

Does indexwire rank itself?

Yes. indexwire competes in two of the launch categories and appears on those boards under the same method as every other brand, including in the weeks its own position goes backwards.

Which categories does the index launch with?

Eight: screenshot and scraping APIs, social scheduling, AI transcription, product analytics, transactional email APIs, help desk software, applied AI consulting, and AEO/SEO SaaS. Six are categories indexwire sells nothing in; two are ones it competes in.

More articles