Skip to content

Data & Research

How we measure, in the open.

A measurement company's method should itself survive measurement. So we publish how the numbers are made — the definitions, the confidence, the sources — before we publish any number.

01/Methodology

The method, stated plainly.

What Share of Model measures

Share of Model is the share of category answers that name, cite, or trust a brand — read by engine, intent, language, and region. It is not a rank and not a vanity count: it only means something paired with confidence and correctness. We define the query universe first, then measure against it, so the denominator is explicit rather than implied.

Why every number carries a confidence interval

Answer engines are stochastic — ask twice, get two phrasings. A single reading is a sample, not a fact. We report every metric with a confidence interval and a sample size, and we treat a movement as real only when it clears its interval. The CI cuts both ways on purpose: it makes the number diligence-proof, and it makes it harder to oversell.

Source citation over source guessing

A grade without a source is an opinion. For each answer we trace which pages, partners, and threads appear to be teaching the model, and who owns them — so a recommendation points at a specific, checkable cause rather than a hunch.

02/The index

The India Share-of-Model index.

Our first public study: a quarterly, open read on which brands AI names across India's categories, engines, and languages — the scoreboard for the answer era, with its method attached. We're finalising the cadence and first release; we'd rather publish it late than publish it wrong.

First release forthcoming

No figures are published yet — this page describes the method, not results. Subscribe to Machine Readable to get the index the moment it's live.

Want the method applied to your category?

A founder-reviewed scan runs this methodology on your brand and sends the read within 48 hours.

Get your scan