How I work
Three things I won't compromise on.
I've had the luxury of shipping in domains where the wrong call doesn't just cost money, it costs trust. These are the rules that kept me honest.
ROI outweighs novelty.
Technical excitement is cheap. Sustainable economics are rare. I've killed a $1M AI initiative during discovery when PHI/PII compliance blockers turned the ROI negative; I'd do it again. The hardest deliverable a PM produces is a well-argued recommendation against shipping. Build what can scale safely, not what demos well.
Behavior is signal. Opinions are noise.
Users tell you the checkout is "fine." The funnel tells you they abandon at step three. I trust instrumentation, Pendo telemetry and controlled experiments, over stated preference. That's how we cut enterprise cart abandonment 42%, and how I sequence roadmaps now.
In regulated domains, trust compounds.
In healthcare and GRC, adoption isn't a function of speed or feature count, it's a function of defensibility. The clinical MVP pivoted from speed-first to interpretability-first after clinician interviews told us they wouldn't sign what they couldn't defend. Controlled automation beat full autonomy; trust beat velocity. $500K+ ARR in six months followed.
Operating system