PUBLISHED STUDIES
All case studies
Every one of these started the same way: a number nobody could explain, and a suspicion that the obvious answer was wrong. The interesting part is never the redesign — it's the moment the evidence contradicts what everyone assumed.
Four studies across social care, retail, consumer electronics and a product I'm building myself. One was commissioned and measured. Three were run on my own initiative, where nothing can be claimed and the reasoning has to carry it. Both are labelled, because a portfolio that reports them the same way isn't telling you the truth about either.
Reducing drop-off for users seeking social care
The client believed the site needed rebuilding. Before committing to that, I tested whether the evidence agreed — and which problems a redesign would actually solve.
Exits were happening on Services and About — not the contact form. The problem was structural, not the form.
Measured outcome · client engagement
They came in expecting a redesign. The evidence had a narrower answer.
Read the research → 8 min readACROSS CONTEXTS
The sector changes. The question rarely does.
These studies span second-hand retail, social care, consumer electronics and a consumer AI product. The industries have almost nothing in common — but the behavioural question underneath each one is close to identical: intent exists, and something in the experience is spending it before it converts. What changes is which signals carry weight, and that is what the research is for.
E-commerce & retail
Where the barrier is confidence, not price. The behavioural question is what a buyer needs to believe before they will commit — and which reassurance has to appear before the decision, not after it.
Trust · PDP & checkout · CRO
Product & SaaS teams
Where analytics show what happened and nobody has time to establish why. The research question is which of several plausible explanations the behaviour actually supports.
Onboarding · Activation · Retention
Service & care platforms
Where people arrive under pressure, often deciding for someone else. Reassurance carries more weight than persuasion, and a drop-off is somebody who did not get help.
Service design · Accessibility · Trust
Fixing a false bottom on a premium PDP
Behavioural benchmarks suggested most visitors never reach the persuasive material below the hero. Four isolated experiments, each with a defined win criterion, to test whether placement is the fix.
Before it was a content problem, it was a visibility problem.
Experiment design · proposed, not run
The hero wasn't too big. It was hiding the reason to buy.
Explore the experiment → 6 min read
Closing the trust gap on second-hand retail
Warranties, verification and return protection weren't visible at the point of decision. Eight participants across two cohorts showed where trust broke down — and where reassurance needed to move.
7 of 8 participants couldn't interpret the condition grades. Price wasn't the objection; the absence of a reason to trust it was.
Research scope · qualitative study
People didn't need more information. They needed a reason to trust it.
See the investigation → 7 min readDesigning an AI companion that transforms everyday laundry
Helping people move from guesswork to confidence by turning one of the home's most overlooked appliances into a trusted AI-supported experience.
The interface, not the person, is the source of the error.
Prototype scope · in progress
The hard part isn't designing the AI. It's making people believe it belongs in their kitchen.
Research in progress · not yet publishedWhy these belong together
Four industries. One way of asking.
Good research doesn't begin with answers. It begins with a question worth testing — and the willingness to be wrong about it in public. Every one of these projects ran the same three checks: who is this person right now, what has their industry taught them to expect, and where exactly does the flow break.