Monitor health, traffic, and usage

Use metrics to find patterns and logs to explain individual requests. Always check the environment, time period, and measurement before comparing a chart with a user report.

Find the right view#

Compare like with like#

Choose the same time range and environment when comparing charts. Check whether a value counts requests, sessions, users, pages, or failures. A single page load can make several API calls.

Look for a change relative to normal behavior: increased latency, a rising error proportion, a new traffic pattern, or a schedule that stopped producing runs. A lower request count can indicate either reduced demand or a broken client.

Investigate a signal#

1. Identify the affected interval and action or app.
2. Check recent releases and configuration changes.
3. Open matching action logs and inspect representative failures.
4. Compare a successful run using the same path.
5. Record the likely cause and the verification needed for a fix.

Treat an unavailable data source differently from a genuine zero. Avoid drawing conclusions from a dashboard that has failed to load or has no retained evidence for the selected period.

Review reports and warnings#

Confirm a generated report's period, status, and data scope before sharing it. For a warning, inspect the underlying evidence and determine whether action is needed; acknowledging a signal is not the same as fixing its cause.

Use Maintenance for recurring failures and downloads for available mobile monitoring tools.