We were deciding on a new API architecture and I had 3 engineers with conflicting recommendations. I stopped pretending I could adjudicate the technical tradeoffs and handed that call entirely to the team lead. He picked one, we shipped on schedule.
Explicitly hands the decision to engineering — PM abdicates rather than facilitates.
We were deciding between two rendering architectures. I ran three sessions with engineers, pushed hard on latency trade-offs, and chose the more complex path. It added six weeks to the schedule. Adoption metrics hit target, but the team was exhausted.
Engaged genuinely but never examined whether the complexity tradeoff was worth the schedule cost.
We were deciding between two database architectures for a new feature. I ran four structured sessions with the engineering leads, asking them to walk me through failure modes specifically. They aligned on one option, we shipped it, and latency dropped 40 percent.
Structured sessions around failure modes — technically humble, productively directive, and repeatable.