Scaling research coverage by teaching PMs to run their own studies — safely

Overview

The program was a mandate from the VP of Product: elevate UX understanding org-wide — building PM empathy for user needs and pain points while gating for research quality — in the absence of a fully-staffed research team that could cover every PM's needs.

Not meant to produce researcher-grade work. Meant to raise the org's baseline and make real progress from wherever it currently stood, for a 5-person team supporting 95+ PMs and 1,000+ engineers.

Approach

Time-boxed, tiered review gates rather than one large review — deciding deliberately which elements of a study plan actually need researcher vetting.

Heavy, guided templates as the main scaffold: they walk PMs through the process and surface the right signals at the right time for researchers to assess quality without redoing the work.

AI-parseable guidance material so PMs could move faster through planning using AI tools, while the process still forced a deliberate slow-down on what mattered most.

Three review gates: (1) high-level research plan — confirms audience and feasibility, gates recruiting; (2) detailed study materials — discussion guide, prototype readiness, gates scheduling; (3) execution support — coaching on early sessions plus an AI moderator-coaching tool built to scale beyond 1:1 capacity.

Hard line: the final deliverable requires the PM to state a recommendation, and a researcher reviews whether it's actually supported by the evidence and its strength — that judgment stays with research.

Impact

4 PMs fully certified, 10 in active training, and 3 studies live within the first months of launch.

PM-led studies directly shaped product decisions for Segmentation and Data Catalog features.

Sponsored by the UXR Director and SVP of Product — reflecting executive investment in scalable, high-quality research.

Paired with AI-powered research operations (Gemini + Claude Code across the research lifecycle) that cut screener development and review from 6+ hours to ~1 hour.