AboutPragueWorking with clients globally
A measurement bureau for the AI layer
A new layer appeared between companies and their customers, and almost everyone measuring it is guessing. AI assistants answer millions of buying questions a day. We record those answers, measure them properly, and help companies act on what they say.
Founder and lead analyst
Martin Dráb has spent fifteen years on the strategy side of marketing. He co-founded and leads Grizzlink, a strategy and communications agency, and Meat & Friends, an investment vehicle that builds marketing operations into early-stage companies. Before that he worked at Publicis Groupe on brands such as Microsoft and Nestlé. Along the way he built the AI stack that runs his agency's daily operations, which is where the question behind Kesterley came from: what do these systems actually tell our clients' customers?
He holds a master's degree in communication from Charles University in Prague, studied marketing at Hong Kong Baptist University, and completed Mark Ritson's Mini MBA in Marketing. He designs the measurement, reads every audit before it ships, and takes the walkthrough calls himself.
Where the method comes from
From marketing science. The Ehrenberg-Bass research tradition showed how brands grow: through mental and physical availability across the situations in which buyers enter a category. Kesterley applies that framework to machines and calls the result AI availability. The buying situations become the questions we put to AI assistants; mental availability becomes a brand's share of their answers; physical availability becomes whether an AI agent can read your pages and do business with you.
How we work
Fixed scope, fixed price, written findings. An audit takes five business days and ends in a report your team can act on the same week. The program takes those findings to shipped fixes in 30 days, with an owner for every item and progress re-measured quarterly. We work with marketing leaders, founders and product teams directly, and with agencies that want the measurement under their own brand.
Why the numbers can be trusted
AI answers are unstable, and most tools in this market report precise-looking numbers anyway. We took the opposite path: designs frozen before measurement, a confidence interval on every number, our own reliability disclosed, and a public, versioned methodology. When a difference is within noise, we say so. It costs us dramatic headlines and earns the only thing that matters in measurement: being right.
The fastest ways to see how we work: run the free check on the homepage, read an industry report, or read the methodology itself.
Read the reports