A review gate is a service, not a checkpoint

Every architecture review board I’ve seen starts with good intentions and a calendar invite.

Six months later it has a queue. Teams wait for the next sitting, the template runs to pages, and the answer comes back as a list of concerns with no path forward. Nobody decides to bypass the board. They just notice that building first and asking later is faster.

That is how a control becomes theatre. The reviews still happen. They just happen to the projects that were going to behave anyway.

When I set up a review gate, I try to design it from the other side of the table. The person waiting at it has a date they promised someone. So three things matter more than the template.

A clock. The requester knows on day one when they’ll get an answer, and it’s days, not sittings.

Published guardrails. If the baseline is written down, most teams can check themselves before they arrive. The review becomes a conversation about the two things that are unusual, not the twenty that are standard.

“Yes, with conditions” as the default answer. A flat no is rare and comes with the alternative attached.

None of this lowers the bar. It moves the effort to where the risk is and makes the approved route the quick one.

A gate people route around protects nothing. Measure yours by how many projects walk up to it early, not by how many it stops.