The case-study page lies in the most flattering way.

The setup. A B2B SaaS engagement, late-stage. The brief came in as a "redesign of a tool that feels dated". Eight-week scope, mid-five-figure budget. The team was earnest and the brief was honest.

What actually happened. Six weeks in I had assembled enough conviction that the actual question was not visual. The team was disagreeing — quietly, in the work, not in the meetings — on what the tool was actually for. The redesign was the disguise. I had spent five weeks polishing the wrong thing because the symptom looked exactly like a redesign problem.

What I learned. The one-sentence-test (write what the product is for, by hand, three people independently) costs about an hour and would have surfaced the disagreement in week one. I now run it before anything visual on any engagement that started life as a "redesign".

The setup. An early-stage consumer product. Listening week as planned — three days of internal conversations, two days of user calls, a written diagnosis at the end.

What actually happened. The user calls were with the most articulate users I could find on short notice. They were also the three power users. The diagnosis I wrote was crisp, confidently delivered, and load-bearing for the next three months of work — and quietly wrong on the question that mattered most. The mediocre and quiet users would have told a different story.

What I learned. Build the user-selection rubric first, with the client, and bias toward the quieter end. Articulate is not representative. Three thoughtful calls with mid-engagement users beat ten with power users for diagnostic purposes — and I write that into briefs now.

The setup. A twelve-week diagnostic-plus-design contract. Mid-stage product, real revenue, capable team.

What actually happened. The answer was effectively in by the end of week two. The team needed to make a decision and stop, not run a redesign. I extended scope for another nine weeks because I had convinced myself the team needed me to "see it through". The team needed me to leave the room and let them act on the diagnosis. The work I shipped was fine and unnecessary.

What I learned. End engagements when the answer is in, not when the calendar says. I now structure listening-week contracts with an explicit "we stop here if the answer is enough" clause, priced as a separate fee. The client almost always picks that option, and the work that follows — if it follows — is sharper for it.