In this guide
An incident update explains what’s affected, what the team is doing, and when to expect the next update. During a production incident, you can confirm the impact on the service while the cause is still under investigation. The examples below follow a fictional incident in which some customers have trouble opening their order history.
Write the first update while you’re investigating
On a public status page, start with what the customer was trying to do. Then say the team is investigating and give a specific time for the next update.
Some customers are unable to load their order history.
We’re investigating the issue.
We’ll post another update by 14:25 UTC.
Some customers is useful when the impact is limited to part of the customer base. By 14:25 UTC tells readers when to expect the next update. Share facts as you confirm them and post by the time you promised, even if the investigation is still underway. Google SRE Workbook
Report mitigation and monitor recovery
We’ve rolled back a recent configuration change tells readers the team reversed a change. Say what you observed afterward and what the team is watching.
We’ve rolled back a recent configuration change.
The error rate is back to normal, and we’re monitoring recovery.
Use mitigated when the team’s action has reduced or stopped the impact. Monitoring tells readers you’re watching the service after the intervention. Once you have evidence the service has recovered, close the incident and give the window when customers were affected.
Order history is loading normally again.
The issue affected some customers from 14:10 to 14:23 UTC.
Adapt the update for an engineering channel
In an internal channel, include a start time, an endpoint, or a metric that helps the team investigate. If you’re testing a hypothesis, We’re checking whether… makes it clear you’re still investigating the connection.
The orders API started returning more 5xx errors at 14:10 UTC.
We’re checking whether a recent configuration change contributed to the increase.
For public updates, follow your team’s approval process. GitHub’s published availability reports offer examples of how an engineering team describes customer impact and service recovery. GitHub Blog
Practice writing two updates
Imagine uploads started failing for some customers at 09:05 UTC and the team is still investigating. Write a two- or three-sentence customer update with the impact and a commitment to post again by 09:20 UTC. Then record a version for your internal channel and add a detail someone on the team could investigate.
Keep practicing with how to ask for clarification in English and how to give a daily stand-up update.