Skip to main content
Retrospective processes are easy to configure and easy to quietly stop following. Nothing breaks when a retrospective is skipped, no alert fires, and the incident closes anyway — so completion decays without anybody deciding to let it. Tracking it takes three pieces, each of which already exists.

1. The Status You Measure

Every incident carries a retrospective progress status: Count skipped separately from not_started, but do not read it as a deliberate decision. Rootly also skips a retrospective when a skip preference matches, when Skip retrospectives for cancelled incidents is on and the incident is cancelled, or when the incident is closed in the web app, Slack, or Google Chat before a non-mandatory retrospective is completed. To keep a retrospective from being skipped this way, make it mandatory in Skip and Mandatory Preferences. It is available in Liquid as {{ incident.retrospective_progress_status }} — see Incident Variables.

2. Where to See the Backlog

Open Post-Incident from the sidebar and select the Retrospectives tab. The numbers at the top cover the whole organization: Incidents with incomplete retrospective steps, Incidents with overdue retrospective steps, and Avg time to complete a retrospective. Each one compares the last 30 days with the 30 days before. The filters on the page do not change these numbers. Below them, each incident is listed with its retrospective steps. Three views are usually enough:
1

The backlog

Filter Status to Todo and In Progress, and Incident Severity to the severities your process actually requires a retrospective for. This is the list to shrink.
2

The backlog by team

Add the Incident Team filter. Completion is a team habit, and an organization-wide average hides the two teams carrying the problem.
3

Overdue steps

Watch Incidents with overdue retrospective steps. A retrospective written six weeks after the incident is an archaeology exercise. The people involved have forgotten the detail that mattered, and the action items get vaguer the longer it waits.
Filter by the severities your process requires, not by all incidents. A view that counts every SEV4 as an outstanding retrospective produces a number nobody believes, and a number nobody believes gets ignored rather than argued with. The numbers at the top count every severity, so add a skip preference for the severities that do not need a retrospective.

3. Reminders That Do the Chasing

Chasing by hand is what stops after two weeks. Step reminders do it for you. Each step can send up to three reminders to its owner: before the due date, on it, and after it. Rootly sends them as a Slack direct message when the owner’s Slack account is linked, and by email otherwise. Steps with no owner send no reminders. See Configuring Process Steps.

Reviewing It On a Cadence

A backlog nobody opens tracks nothing. Review the Retrospectives tab under Post-Incident on a fixed cadence with the people who own the outcome, not with everyone. Weekly is usually the better choice: a month is long enough that a slipping retrospective is already stale by the time anyone looks.

Configuring Process Steps

The steps in a retrospective process, which determine what “completed” means, and their reminders.

Configuring Processes

Skip and mandatory preferences, which decide which incidents need a retrospective.