Book a call

Research method

One research method, four practices: how we point UX research at marketing and operations.

Sep 24, 2025 · 9 min read

Cover illustration: one research method across four practices

We describe ourselves as a UX research team and an AI application research team working across four practices. That reads like a diversification story. It's the opposite: one method, applied to four surfaces. The method is UX research's oldest discipline — watch what people do, distrust what they say, and make decisions on evidence rather than seniority.

Why companies fragment into dialects

In most organizations, marketing reads traffic, product reads feature adoption, operations reads spreadsheets, and the AI team reads benchmarks. Four dashboards, four vocabularies, and not one of them is looking at the same user for longer than a quarter. When each function optimizes its own numbers, the user experience becomes the accidental sum of four local maxima. It usually shows.

The same method, four aims

Point the method at marketing and search queries become user research at scale — thousands of people stating intent in their own words, no moderator needed. The gap between what they type and what your pages say is a usability finding, and it's also the raw material of SEO and GEO.

Point it at operations and a retention curve becomes a longitudinal usability study. The curve shows where habit fails to form; session observation shows why; the fix is designed, shipped, and read back off the next cohort. Growth work is research with a revenue line attached.

Point it at AI and evaluation becomes usability testing for models: real tasks, defined success, failures examined one by one instead of averaged away. A model without an eval is a design without a test — you don't know what you shipped, only that you shipped.

What this buys, honestly

One method across four practices means findings compound instead of evaporating at department borders: what we learn watching users struggle feeds the marketing language, the retention hypotheses, and the AI feature's success metric — because they were always the same user. The limit is equally honest: method transfers, domain knowledge doesn't, and each practice still has to be earned on its own terms. The through-line is a single sentence — what users actually do outranks what anyone thinks they should do. Everything we sell is that sentence, aimed at a different surface.

Let's talk

Let's find where intelligence moves your metric.

Tell us what you're building. We'll tell you honestly where intelligence moves the number — and where it doesn't.

Book a call