If I Were Your Founding Researcher Today

If you're about to make your first research hire, or you are that hire, this is for you.

I've been on both sides. At Optym, I was the founding researcher: the only one, working directly with executives. At Meta, I worked inside a Reality Labs research org of 300+.

Indexing it on size is the first assumption I'd challenge. Big orgs have deep expertise, but their impact gets diluted by distance from decisions. Solo researchers sit right next to the decisions, but their rigor is fragile and nobody checks their work.

It's tempting to read that as big vs. small. It isn't. It's maturity, and maturity is the first thing I'd diagnose.

Maturity, not headcount

UXinsight's research maturity framework describes four stages. Here's my shorthand for each:

Emerging: research is heroic and episodic
Developing: a cadence exists, but it's reactive
Established: a roadmap exists and research shapes decisions
Pioneering: research is infrastructure, not a function

A 10,000-person company can sit at Emerging. A 30-person startup can sit at Established. If you're hired to "build research," the job is moving the organization up a stage, not shipping 50 studies. And maturity isn't a ladder you climb once. Reorgs and acquisitions push companies back down.

The second assumption I'd challenge: AI gives a solo researcher enterprise-level throughput. That looks like the gap closing, but it isn't.

AI makes the solo traps faster

AI solves the easy problem, which is speed, and it accelerates the three traps every team of one falls into:

Reactive tasking. Faster synthesis means you say yes to more requests you should have questioned.
Stakeholder bias. AI will happily polish whatever framing the loudest stakeholder handed you.
Insight fragmentation. Ten quick AI-assisted studies scatter faster than three slow ones.

What I'd do

So what would I actually do as your founding researcher today?

I'd build your company's ability to learn, not a research backlog, in three moves over 90 days. Each move closes one trap, and the goal of all three is to stop being the only person doing research.

The plan assumes Emerging or Developing, which is where most first research hires land. The field notes at the end cover every stage.

Move 1 · Days 1–30

Earn trust by anchoring to decisions, not requests

I'd start with a listening tour. I wouldn't ask "what research do you need?" I'd ask, "What decisions are you making in the next two quarters, and what would change your mind?" Stakeholder empathy maps turn the answers into what each person is measured on, what they fear, and what evidence they actually trust. They also show me which stage the org is really at.

Then I'd pick one or two quick wins: a live decision, a real evidence gap, delivered in weeks. Credibility pays for everything after.

Trap closed: reactive tasking. Once research is anchored to decisions, every new request gets weighed against what's at stake instead of joining a queue.

Move 2 · Days 31–60

Build the rails before you need them

Enterprise research runs on infrastructure that founding researchers postpone until it hurts. I'd build the lightweight version now:

A research roadmap tied to the product roadmap, so leadership sees what's coming and why.
A participant panel, so recruiting stops costing two weeks per study.
Two lanes for democratized research. "Self-serve with review" handles fast, evaluative work, and "researcher-led" handles generative, strategic work. The two never blend.
AI I've validated. I run it side by side with my own analysis and check where it drifts before handing anything off. I evaluate research tools the way I evaluate AI products.
Quarterly UX health checks against heuristics, so quality is a habit, not a heroic effort.

Trap closed: stakeholder bias. Rails give you a second opinion when you don't have a team to give you one.

Move 3 · Days 61–90

Scale through other people

A team of one has no peers to challenge the thinking, catch duplicate work, or carry insights into rooms you're not in. So you build that network on purpose. That means a shared repository where insights compound instead of dying in decks, feedback loops that show teams what their research changed, and teaching.

At Optym, my lunch-and-learn program spread to our offices in India and Armenia. At Meta, I led Collab Club, which deduplicated research across a 300-person org. The lesson was the same both times: influence without authority can be built into a system.

Trap closed: fragmentation. Knowledge spreads because other people carry it, not because one researcher remembers it.

How I'd measure it

Not studies shipped. I'd track four signals, borrowing the four maturity themes from UXinsight's framework:

Efficiency: share of research hours spent on judgment vs. logistics
Visibility: decision-makers who can name a study that changed a decision
Quality: studies with a brief, and democratized work that passes review
Connectedness: teams reusing insights they didn't commission

And the one that matters most. At Optym, success was two more researchers hired onto other product teams.

A founding researcher succeeds by making research bigger than themselves.

Challenge	What works
"Can you just run a quick survey?"	A three-question intake: What decision will this inform? What happens if we don't know? By when? Many requests dissolve at question one.
Ten requests, one of you	A visible queue, ranked by decision impact and blast radius, not loudness.
Stakeholder insists on their hypothesis	Write it into the brief as a falsifiable claim. They feel heard, and rigor stays intact.
Insights keep getting re-litigated	One-page insight briefs with evidence links. If it can't survive one page, it wasn't ready.
Execs nod at research, then ignore it	Be in the meeting where the decision is made. Readouts after the decision are archaeology.
Nobody trusts the AI synthesis	Pair every AI draft with traceable evidence: clips, quotes, timestamps. Trust is the product; AI is the factory.

Stage	Defense protect rigor	Offense create leverage
Emerging	A brief before every study, even quick ones. Intake, so "no" is a process, not a personality. A health check to test "we already know our users."	The Days 1–30 plan: listening tour, plus 2–3 quick wins tied to an executive's pain point.
Developing	DECIDE as the default study scaffold. Protected research slots, so urgent ≠ unplanned. Monthly audit of democratized work.	A panel and templates to cut logistics. A roadmap synced to quarterly planning. Train 1–2 PMs or designers in evaluative methods.
Established	A quality bar for democratized research (brief, ethics, synthesis review). Health checks run by 3–5 trained evaluators, not just you.	A roadmap synced to annual planning, so research shapes what gets built. The repository becomes the org's knowledge base. Mentor people doing research on adjacent teams.
Pioneering	Guard against maturity decay from reorgs and acquisitions. Rotate audit ownership so quality isn't personality-dependent.	Continuous discovery loops, embedded metrics, and research shaping strategy at the top. The job becomes scaling judgment, not headcount.