Audit the operating review like you would audit any other operating system: by asking what outputs it reliably produces.
Start with decision yield. In the last three reviews, what decisions were actually made? Were they the right level of decision for the room? Did they change priorities, resources, sequencing, risk posture, customer commitments, or ownership? If the yield is low, the review is probably a readout.
Check variance quality. Are agendas built around material gaps between expected and actual? Are positive and negative surprises examined? Or does the template force the same tour every time regardless of signal?
Check narrative quality. For major metric movement, does the packet separate fact, interpretation, confidence, and ask? Can the narrative be challenged? Does it name what would prove it wrong?
Check ownership. For every issue discussed, is there a clear owner with authority to act? If ownership is split, is the decision right explicit? If escalation is needed, is the escalation path known?
Check follow-through. What percentage of commitments from the last review were completed, changed intentionally, or left vague enough that nobody can tell? Low follow-through is often a design problem. The review may be creating weak commitments or decisions without authority.
Check surfacing speed. Are risks appearing early enough to act on them, or only after they have become incidents? A review that discovers problems late is not connected to the work. Look at the weekly signal layer, incentives, packet standards, and whether bad news gets punished.
Check room composition. Are the right deciders present? Are too many spectators present? If the room is too large, the review will drift toward performance. If the decider is absent, it will drift toward discussion.
Finally, check changed behavior. After the review, what is different? A priority changed. A decision was made. A risk was accepted. A resource moved. A tradeoff became explicit. A cadence changed. If nothing changes repeatedly, the review is documenting inertia.
The audit should produce one next improvement. Tighten the packet. Rebuild the agenda around variance. Add a decision log. Remove spectators. Clarify escalation rules. Start with follow-through. Pick the constraint and fix it. Operating reviews improve the same way operating systems improve: one honest bottleneck at a time.
