How mixed-methods research turned 1,500 survey responses into three actionable trader personas — and a platform direction OKX could actually build against.
Engagement
Product Research, Persona Development & Data Analysis
Client
OKX — Global crypto exchange, retail trading for professionals
Duration
8 weeks · mixed-methods research
§01 · At a Glance
A research engagement that replaced guesswork about "the pro trader" with evidence about three of them.
Read time · 5 minutes Published 2026 · FutureProof
Research Scale
1,500
pro-tier customers surveyed across regions and trading styles.
Depth Interviews
24
structured follow-ups with active high-volume traders.
Personas Delivered
03
evidence-based trader archetypes — each with a distinct decision loop.
Outcome
01
platform direction OKX could build against — not five competing ones.
§02 · The Challenge
Every roadmap was right. None of them agreed.
OKX's pro trading platform was growing fast, but product direction had become contested. Every team had a plausible roadmap — advanced order types, copy trading, better charting, derivatives depth — and every team had anecdotes to back it up. Leadership needed a way to decide without defaulting to whoever was loudest in the room.
The missing ingredient was a shared, evidence-based picture of who the pro trader actually was. The community wasn't one audience — it was several — and the team didn't have the data to prove it or act on it.
The problem beneath the problem
"Pro trader" was shorthand for at least three distinct user mindsets — and the platform couldn't serve all of them equally well.
01
High-volume discretionary traders, semi-automated market makers, and cross-venue arbitrageurs have fundamentally different workflows — but were being pitched the same feature set in the same UI.
02
Churn was concentrated in a specific segment the team didn't have a name for. Without a persona framework, nobody could say whether a new feature would help or hurt retention until after ship.
03
Roadmap debates kept relying on individual anecdotes ("my friend who trades $50m/month says…") rather than segmented behavioural data.
“
"Pro trader" is a segment of segments. Until you name them, every roadmap decision is a guess — and every new feature is a bet against whichever one you ignored.
FutureProof thesis
§03 · The Approach
Three moves that turned 1,500 responses into a product direction.
8-week engagement · mixed-methods · segment before you ask
01
Move · Quantitative Scan
Segment the user base before you ask the first question.
We designed a 1,500-respondent survey stratified across volume tiers, geography, and primary trading style. The instrument was built to surface decision drivers — not feature preferences — so we could cluster on mindset rather than usage patterns that shift with market conditions.
Build the personas from observed behaviour, not self-reports.
We ran 24 structured follow-up interviews against the cluster centroids — watching real sessions, mapping decision loops, and surfacing the moments where traders abandon a flow or reach for a competitor's platform. The three personas that emerged were grounded in what traders actually did, not what they said they wanted.
Collapse five roadmaps into one defensible direction.
The final deliverable wasn't a persona deck. It was a prioritisation framework leadership could use in roadmap meetings: which persona does this feature serve, at what cost to the others, and where does it sit on the retention curve we actually measured? Product direction stopped being about loudest voice and started being about which persona OKX chose to build for first.
Persona × feature matrixRetention-weighted prioritisationDecision framework
§04 · AI-Era Persona Research
What a traditional persona study would have missed.
Classic persona work tends to produce archetypes that look clean on a poster but collapse under roadmap pressure. The details feel generic. The decisions feel unchanged.
In financial platforms, the personas that matter aren't defined by demographics — they're defined by decision architecture. What is this trader optimising for? What counts as a win? What's the thing that makes them switch venues?
We combined quantitative clustering with behavioural observation so the personas had both statistical weight and operational texture. Each one came with a named decision loop, a stated switching trigger, and a retention signature — the things product teams actually need to reason about in a planning meeting.
"A persona isn't useful until someone can use it to kill a feature request."
§05 · Impact
Three personas. One direction. A roadmap that finally agreed with itself.
Research Output
03
Evidence-based personas — each with a named decision loop, retention signature, and switching trigger.
Decision impact
One.
Clear platform direction leadership could build against — not five competing ones.
What the engagement delivered
01
A quantitative segmentation of 1,500 pro-tier customers, cross-cut by volume, geography, and trading style — the baseline OKX can now re-run on any release.
02
Three operational personas, each tied to measured behaviour in the platform — not qualitative archetypes that can't be verified.
03
A persona × feature prioritisation matrix leadership uses in roadmap meetings to resolve cross-team disputes.
04
A retention signature per persona — product can now model feature decisions against the retention curve segment-by-segment.
05
A re-runnable instrument. The survey + interview protocol is OKX's going forward — not a one-off deliverable.
§06 · Work With FutureProof
Have a segment everyone's arguing about? Start with evidence, not a persona workshop.
If your team keeps debating "who the real user is" and the roadmap keeps drifting, that's a research and segmentation problem — not a design one.
We start with a $500 AI Readiness Audit. Five days. You'll come out with a clear view of where the decision keeps getting stuck and what it would take to unstick it.