A platform team reports 90% of services are on the platform, and delivery has not improved. What is being measured wrongly?

What is being measured wrongly

Coverage rather than value. A service can be "on the platform" while the team still builds its own deployment tooling, maintains its own observability, and works around the platform for everything that matters. Coverage counts registrations, not usage.

It is also frequently a mandated number, which makes it a compliance measure rather than a signal.

What to measure instead

• Time from an idea to a running service in production, which is the platform's actual promise.
• The proportion of teams using each capability voluntarily, per capability rather than in aggregate — because the aggregate hides that everyone uses the deployment tooling and nobody uses the templates.
• How much a team must know to operate their service. A platform that reduces cognitive load is doing its job; one that adds a layer while teams still need the underlying knowledge is not.
• Support load from platform users, split by "how do I" and "it is broken". A high rate of the first is a documentation and abstraction problem, and it is a leading indicator of eventual bypass.
• Delivery flow metrics for teams on the platform versus those not, which is the only comparison that tests the value proposition directly.

The signal that matters most

Voluntary adoption of new capabilities. When the platform ships something and teams adopt it without being asked, the platform is a product. When adoption requires a mandate and a migration programme, it is infrastructure teams are made to use — and the difference determines whether the next thing it ships is adopted.

Why the distinction matters commercially

Platform teams funded as cost centres optimise for cost; funded as products they optimise for adoption. That difference determines whether documentation, onboarding, support and developer experience get built at all — and those are exactly what turns coverage into value.

The uncomfortable reading of 90% coverage with no improvement

The platform is solving the platform team's problem rather than the product teams'. The most productive next step is not more coverage but interviewing the teams whose delivery did not improve.