Most of them are about timing: the right question at the right moment. Ask it too late and it becomes an afterthought the whole design has to bend around. Ask it in order, and the compromises never pile up.
I work at the level of systems and information architecture, not just screens. In practice, that comes down to a few consistent moves:
Those structural calls are invisible when they are right and catastrophic when they are wrong. On the largest programs I am also the constant, holding the structure and intent steady across years as teams rotate, the way I did across GM's multi-year transformation.
Most of what matters in a service experience is never said out loud in an interview. On GM I anchored the work in 20-plus hours shadowing live calls, listening for the uncertainty beneath each one, because the brief is rarely the real problem. See it on GM →
On Ford's Build & Price, we were stuck between organizing by model or by vehicle until research settled it: global buyers did not know Ford's model names, they decided on type and budget. I let the data set the structure, because sometimes the fastest path is to stop debating and go ask. See it on Ford Build & Price →
On myQ, the most-used feature failed 100% of the time in testing, so I deleted the elegant, unusable model and rebound it to the group it describes, with no loss of capability. Simplicity is not fewer features, it is the same power expressed the way people think. See it on myQ →
On myQ, the billing page was a mess of 12 plans, but the real problem was a pricing model built like a catalog. I worked with the business to make it hardware-agnostic, collapsing 12 plans into 1 and a week-long setup into minutes; you cannot redraw your way out of an upstream problem. See it on myQ →
On GM's guided workflows, the fast path was to tell advisors what to do and hide the logic. I held the opposite line and showed the reasoning behind every step. You can't always open the model, but you can surface the logic and keep a person in control, and no one trusts, or learns from, a system they can't understand. See it on GM →
Most people use AI like a search engine: ask, take the answer, move on. I design the collaboration itself like a product.
On one self-initiated project, that meant a full UX engagement in three days, built on a 3,000-word operating brief with five specialist lenses engineered to disagree. AI should buy back the thing that is hardest to protect: time to think. Read the Prompt Design study →
Interviews, shadowing, and system audits. Map current state to future so everyone sees the same gap.
Turn research into "how might we" framings, and agree on a few principles before designing.
Flows, then IA, then interface. Modular parts that scale across markets and devices.
Prototype, test light, iterate. Stay in through engineering and rollout.
I question the right things at the right time, with clarity as the goal, and I will defend the harder right call. I mentor the designers on my tracks and bring calm to large, political, fast-moving programs.
My growth edge is the choreography, not the architecture. The structure comes easy; the timing and alignment across teams and months is the part I am still sharpening, because the right rhythm is different for every team and culture, and learning to read that is the craft.
Happy to walk a hiring team through any of these decisions in detail.