Skip to content
← All principles

Reliability is part of the product

Users do not separate the product from the platform that delivers it. To them every silent failure is the product. Neither should you separate them.

Principle 5 of 11
Above, the maker’s model: the product and the platform that delivers it, two separate things. Below, what the user meets: one continuous object, joined at a seam they feel but never see.

There is a habit of treating reliability as something outside the product: owned by infrastructure, measured in uptime, defended in postmortems. Users do not experience it that way. To them, every silent failure, every retry that didn’t happen, every state that quietly drifted, is the product. The feature you shipped and the platform that failed to deliver it are one object in their hands.

This is the premise of site reliability engineering: reliability is a feature, with a budget and a backlog, not a footnote beneath the real work. You don’t get credit for the demo. You get credit for the second year.

Reliability is not a non-functional requirement. It is a feature, and often the first one a user actually notices, because they only notice it once it is gone.

All principles