Setting direction for five PMs.

I set the direction. The teams own the delivery.

The five product managers report to the head of product, not to me. What I own is the direction they build against and the final call on what ships across seven products.

Priorities move every week at a company this size, so I hold the part that can stay fixed: the direction, and a review against it before anything is built. Within that, the people closest to the work own the decisions.

The prototype comes first, and engineering sees it before design does.

I prototype in Claude Code and send it to engineering with two questions: where am I overbuilding, and how does the flow feel? Only then does a design file start.

Review runs in three layers after that. A daily standup for pulse. A protected weekly project review, which keeps a shifting roadmap from becoming shifting work. And working sessions on whatever is riskiest that week, where engineering, product, marketing and executives end up in the same room.

Support reports to me, so a wrong call comes back as tickets.

A spec usually leaves its author for good. Mine comes back through my own support and community team.

Lounges are the clearest case. Messages sent inside a lounge stayed inside it, on purpose. What that did was bury them, because nobody looks inside a lounge for a message. It reached me as tickets, not through a research cycle, and the fix was to store lounge messages in chat as well.

Four steps passed review, including mine. Use failed them.

Email or phone, verify, set up a profile, pick interest tags. In use it failed three ways. The steps stood between a new user and the feed. The combined sign-up and log-in screen made returning users create duplicate accounts. And people tapped through the interest tags, so the data wasn’t worth the screen.

We split sign-up from log-in, cut the middle, prefilled the profile, and inferred interests from what people watched and followed. Duplicate-account complaints dropped and install-to-feed got shorter. I never instrumented either, so I can give the direction and not the size.

I delivered a consolidation I didn’t choose.

Five apps read as five different companies. The decision to consolidate the brand came from above me. The social app I inherited had eleven different search bars: five heights, five corner radii, and a shared component used in two of them.

Then the company rebranded partway through, with every product still shipping. Neither engineering nor my own team wanted the work, and I couldn’t fall back on it being my call. I carried it, and I didn’t undercut it.

Five different apps now stand on one shared foundation. Engineers now ship new screens against tokens and components that already exist instead of drawing them again.

Five designers hired, every demo run, the board presented.

I hired five designers from more than 40 interviews. One joined as a junior and now leads the team.

I led the investor presentation and the investor site, and presented both to the CEO, board and executive team.