The Shared Platform, Part 5 of 7 · Showing the value of a platform whose success looks like silence.
The busiest days on a platform I worked on were weekends, Christmas week and Chinese New Year. Mondays were the quietest. The load didn't come from our merchants. It came from their customers, all ordering at once.
When those peak days went smoothly, nobody noticed the platform. When they didn't, it was the only thing anyone talked about.
That's the measurement problem for every platform team. Success looks like silence, and silence doesn't get budget approved.
Engineers' dashboards answer a different question
We watched the platform in Datadog. It told engineers what broke and where. It didn't tell a CFO why the platform deserved next year's budget, and it was never meant to.
A platform team needs a second set of numbers: ones that show what the platform makes possible for everyone else. I look at four.
The first is reliability: uptime, how often incidents happen, and how long recovery takes. Measure these on peak days, not as averages. An average across a quiet Monday and Chinese New Year hides the only day that matters.
In payments, add one more check. Some failures never appear on a dashboard. They show up in reconciliation, weeks later, as money that moved when it shouldn't have. A platform that reconciles cleanly is reliable in a way uptime can't show.
The second is adoption. What share of product teams use the core services, and how many are quietly building their own? Low adoption is the most honest feedback a platform gets. If teams route around you, the platform is making their lives harder, whatever the roadmap says.
The third is lead time: how long a change takes to go from commit to production. This is where friction in the platform shows up first. Every manual approval, flaky test and waiting-for-platform ticket lands here.
The fourth is cost: cloud cost per product or team, and engineering hours saved by using shared services instead of rebuilding them. The second number is harder to measure, but it's the one that makes the case for a platform team existing at all.
Look ahead, not just back
Reliability, lead time and cost tell you how last quarter went. To see where next quarter is heading, run a short developer-experience survey every quarter and track the adoption trend. When developers say the paved road is getting harder to use, velocity and retention follow a few months later.
Say it in the language of the room
Don't show a CEO a graph of API latency. Show them a sentence that ties the platform to an outcome they already care about, such as: "Because of the new CI/CD pipeline, we shipped X% more features this quarter without adding headcount." Put in the number you actually measured, not an estimate.
A platform's value is how much faster and safer it makes everyone else. If you don't measure that and say it out loud, the only time leadership hears about the platform is when it fails.