Made by Michael
Zest: a research-led MVP for a food-systems event.
End-to-end project from initial research through to an MVP that tested broad ideas and surfaced a user segment worth building for. Owners, managers, event organisers, and end users, all in the room.
Brief
An end-to-end brief: take a food-systems event concept from open question to a working MVP, with the research to back which segment to build for. The arc covered interviews, affinity-mapped synthesis, two primary personas, a UI flow, low-fidelity, and three final designed surfaces tuned to one of the segments identified along the way.
Problem
The concept had three plausible audiences and no way to choose between them. Owners, managers, and event organisers each pulled the product in a different direction; end users hadn't been spoken to at all. Building for everyone would have shipped a product for no one. The job was to put every voice in the same room, find the tensions, and come out with a sharp enough segment to design against.
My role
I led the project end-to-end as the sole designer:
- Interview script and recruitment across three operator audiences plus end users
- Affinity-mapped synthesis: qualitative observations against quantitative data on a single A1 board
- Two primary personas — one operator-side, one user-side
- UI flow, low-fidelity wireframes, and the MVP scope decision
- Final design across onboarding, landing, and venue surfaces
Research
Interviews ran across three operator audiences — business owners, managers, event organisers — plus the end users themselves. Each conversation tested the same set of pain points from a different angle, so the operator and consumer ends of the system could be read together rather than as separate problems.
Affinity map
Synthesis went onto a single A1 wall: qualitative quotes clustered against quantitative data points so the team could read both registers from the same artefact. The cluster that drew the most cross-audience agreement became the MVP scope.
Two surprises came out of the wall: end users described the value proposition in operator language (what was on offer, when, where), and operators kept describing friction in user language (the bit before they made a booking decision). The tension between those two readings shaped the personas.
Personas
Two primary personas came out of the affinity work: one operator-side, one user-side. They became the shorthand for stakeholder workshops and the screening filter for every flow decision after that — if a screen worked for one persona but contradicted the other, it didn't ship into the MVP.
One segment validated, two parked, three surfaces shipped.
UI flow
I mapped the flow before drawing any screen. Two entry points (operator preview, end-user first session) had to land on the same venue page from different starting states, which set the rules for what onboarding had to carry and what the landing page could afford to delay.
Low-fidelity
Low-fi went up on the affinity wall next to the synthesis, so the first round of wireframes could be read against the cluster they came from. That kept the MVP honest: every screen had to point back to a research note on the wall, or it didn't earn its slot.
Final design
Three named surfaces shipped in the MVP — onboarding, landing, venue page — each tuned to do one job for one persona. Operator-side previewed the look and feel before commitment; user-side reached the value proposition (which venue, what's on, when) on the first session.
Onboarding
Onboarding had to clear two thresholds: enough surface for an owner to assess the look and feel before they committed venue data, and enough logic for a first-time user to land on a venue page that meant something to them. The flow surfaces only the inputs both sides had asked for during interviews.
Landing
The landing page was scoped to one job: get the user to a venue worth opening within a single screen of scrolling. Anything that didn't move the user toward that decision came out of the MVP and went into a backlog for round two.
Venue page
The venue page is where both personas converged. User-side: the value proposition spelled out in plain language — what's on, when, where. Operator-side: the surface they'd be assessed on, so the layout had to read as a credible storefront before any real data went in.
What I'd do now
Zest is a 2017 project; I'd run it differently today. Three things stand out. I'd tighten the personas to one each side rather than two, and let the MVP do the persona-splitting work instead. I'd ship a quantitative validation step alongside the affinity wall, so the segment decision was backed by measurable signal, not just a dense cluster of stickies. And I'd use the venue page as the recruiting surface for round two of testing — the surface that showed best in interviews was the one I should have re-tested in the wild.