Missing a release, then fixing how the team planned

Context

I directed delivery of a global drug forecasting platform, leading a 20-person Agile Scrum team and owning scope, backlog, wireframes, and user stories. The platform launched in 70+ countries and supports more than $28B in revenue. Midway through, we brought on a new offshore vendor team shipping minor releases on a two-week cadence.

The mistake

My job was to set up Agile governance. The vendor’s tech lead pushed back hard on story point estimation: his team had worked together for years, and estimation was overhead that would slow them down. I wanted a good working relationship from the start, so I let them skip it.

For six weeks their status reports came back green, and green meant nothing because I had nothing to check it against. Days before a minor release, the tech lead told me we were at least a week out. We missed the release. Our engagement partner was close with the client, so it escalated straight to him, and I owned it: I had let the team run without a check.

The fix

I went back to the tech lead with a simple case: we’d tried it his way and it hadn’t worked, so we’d try mine. We started mandatory planning poker, so no ticket entered a sprint without points, and I tracked the team’s say-do ratio (points committed against points delivered) to measure reliability directly.

Result

The first sprint was painful. By sprint three the velocity data was stable enough to re-plan the roadmap, and the next major release launched to 70+ markets on time. On the product side, reworking the architecture and adding pricing by country lifted forecast accuracy 40% and extended coverage to 1.4B more patients.

What I took from it

Rapport built by skipping the hard parts doesn’t hold up once something slips. The say-do ratio gave both sides a shared number to argue from instead of my read on whether we were on track.