Internal Consumer SLO
A reliability commitment made to teams who cannot switch supplier, which is why it must be measured from their side rather than from the platform's.
Platform teams naturally measure their own components: control plane uptime, pipeline runner availability, registry response time. Consumers experience something different — whether they could deploy, whether the build finished, whether provisioning returned within a sensible time.
The gap between those two views is where trust erodes. A platform reporting 99.95% on every component while teams routinely cannot ship on a Friday has measured the wrong thing, and the resulting conversation is unproductive because both sides have data.
The commitments worth making are journey-shaped: time to provision a new environment, deployment success rate, pipeline queue time at the ninetieth percentile, time to first successful deploy for a new service. Each is measured end to end from the consumer's request. Attaching an error budget to them has the same effect internally as externally: it converts "the platform feels unreliable" into a number both sides can plan against, and it gives the platform team legitimate grounds to stop taking feature requests when the budget is spent.