Hiring researchers first is the expensive way to do this
Put researchers into a company with no recruitment pipeline, no repository and no consent model, and you get expensive interviews that nobody can find six months later. Each study is commissioned, run, presented once, and lost. The researchers are good. The function never compounds.
Build the operations first and the first two quarters look slower than hiring would have. From the third quarter it compounds, because research stops being something a team commissions and becomes something a team can simply do.
I ran my first user interviews in 2011 and built my first research operation ten years later. At CHIP as the person doing them. At CHECK24 as the one arguing for them in an organization that judged every change by conversion. At Smart InsurTech, FTI Group and Conrad Electronic as the one accountable for whether the company could do them at all. The sequence above is not a management preference. It is what you conclude after being the researcher who could not find last year’s study.
Two companies, the same sequence
Conrad Electronic. A B2B electronics platform serving professional buyers. The panel grew from 200 people to more than 7,000 in two years, the Dovetail repository has processed more than 300 interviews, and seven product teams run their own studies. The team is five.
FTI Group. A tour operator with more than 11,000 employees and no research function at all. The infrastructure went in before the team, and the team grew to eleven, hired for people who could teach the rest of the company how to work with customers.
A third, at Smart InsurTech, went in as research practice alongside the design foundations while the company was learning to work as a product organization. The test there was whether the product teams kept using it without being asked, and they did.
Each of those pages carries its own decisions: the two or three moments where the argument changed what was going to happen, with what was planned, what I argued and what followed.
What is the same every time
Operations before headcount. Three times, in three companies, and it looked slower each time for about two quarters.
One searchable place. A repository that everyone can read is what turns a study into an asset. Without it, research is a series of presentations.
Consent settled once. GDPR handling decided with legal at the start, not negotiated per study.
Trained practitioners. A service desk absorbing requests shows output faster and buckles the moment the company starts asking. Teaching product managers and designers to run their own research is the only arithmetic that works when the function is small against the organization.
A standing panel. The participant pool is the least glamorous part of this and the one that changes the speed of everything else.
What I would do differently
Put a number on decision quality in the first quarter, for example the share of roadmap items backed by repository evidence.
At FTI the design system had a return on investment model and that is exactly why it kept its funding through a difficult year. The research function had better stories and no equivalent figure, which made it the easier thing to question whenever budget got tight. Culture change that is not measured is culture change that has to be re-argued every quarter.