How to run a monthly problem review that reduces repeat incidents
Agenda, metrics and inputs for a review board that changes something
A monthly problem review is the meeting where an operations organisation looks at the last month's Sev1s and Sev2s together, finds the causes that repeat, checks that corrective actions are closing, and decides what to fix structurally. Run well, it is where repeat incident rate actually falls. Run badly, it is a slideshow of last month's outages read from memory. The difference is inputs and agenda.
The review is only as good as its inputs. If root causes are free text, corrective actions are untracked and Time to RCA is unmeasured, the meeting cannot do its job. Fix the data before you fix the agenda.
What metrics should a monthly problem review track?
A short, stable set, reported the same way every month so the trend is visible. Definitions stated on the slide.
Repeat incident rate
Share of Sev1/Sev2 matching a prior known error
Time to RCA
Sev1 closure to accepted RCA, vs committed window
RCA SLA attainment
% delivered inside window, by severity
MTTI and MTTR
Trend, with stated definitions
Change-induced share
% of problems with a change as root cause
Open corrective actions
Count and age, by owner
Abstention rate
% of RCAs where cause could not be determined
Worst-repeat services
Top services by recurrence
What is the agenda?
Sixty minutes, in this order. The first half is data; the second half is decisions.
- 1
Attainment and trend, five minutes
The metrics above, this month against the last six. No narrative yet. If a number moved, note it and come back to it.
- 2
Repeat incidents and known error matches, fifteen minutes
Every incident this month that matched a prior known error. For each: was the known error's fix deployed? If not, why not, and who owns it? This is where repeat rate is reduced.
- 3
Top causal patterns, fifteen minutes
Group the month's root causes by cause code and by service. Look for a pattern that spans teams: a class of config change, a shared dependency, a gap in change control. Decide whether it warrants a structural action.
- 4
Open corrective actions by age, ten minutes
Everything past due, with the owner in the room or named. Close, re-date with a reason, or escalate. Do not let actions roll silently.
- 5
New RCAs of note, ten minutes
One or two RCAs from the month worth the board's attention: an abstention that exposed a telemetry gap, an unusual causal chain, a customer-facing RCA that generated follow-up. Not a read-through of every incident.
- 6
Decisions and owners, five minutes
Structural actions agreed, each with an owner and a date, recorded in the CAPA tracker. The meeting ends with a list, not a feeling.
What inputs does the review need in advance?
Circulated at least two days before, so the meeting starts from a shared reading rather than a presentation.
Metrics pack
The tracked metrics with six months of trend and definitions stated. Generated from the problem records, not assembled by hand.
Known error and recurrence register
Each incident this month linked to any prior incident or known error it matches, and the fix status of that known error.
CAPA tracker extract
Open corrective and preventive actions with owner, due date, age and closure evidence where closed.
The month's RCAs
Full reports with evidence indexes and confidence statements, so anyone can check a claim raised in the room.
Who should attend?
The problem manager chairs. Service owners for the services with incidents this month, or their delegates with authority to accept an action. The change manager, because change-induced share is usually the largest slice. A representative from customer success or support where customer-facing RCAs are owed. An SRE or platform lead. Executives attend the quarterly version, not the monthly; the monthly is a working session and should be small enough to make decisions.
Onepane produces the monthly problem review pack.
Onepane's Monthly Problem Review Pack contains recurring-incident analysis, top causal patterns, MTTI and Time to RCA trend, SLA attainment, abstention rate and the services with the worst repeat rate, generated from the month's accepted RCAs and delivered ahead of the meeting. See /artifacts/monthly-problem-review-pack.
The bottom line: A monthly problem review reduces repeat incidents when it runs on structured inputs, a stable metric set, an agenda that moves from data to decisions, and a CAPA tracker that carries the decisions to the next meeting.
Onepane delivers the Monthly Problem Review Pack from the month's accepted RCAs: repeat analysis, causal patterns, Time to RCA trend and SLA attainment.