Why we rehearse cardinality before teaching dashboards

Teams often arrive wanting prettier panels. That impulse is fair; graphs are the visible edge of observability work. We still spend early hours on cardinality and label discipline because every ugly incident thread we have anonymized started with metrics that were technically correct and operationally hostile.

In the first week of Signals that stay useful under load, participants delete work on purpose. They trim labels, re-home metrics, and argue about recording rules before anyone opens a color picker. The sequence is deliberate: if you skip straight to layout, you polish a story that is too long to read at night.

The second paragraph of this note is the encouragement part. Once the foundations breathe, dashboard work feels faster, not slower. Panels stop competing for attention because each one owes its existence to a question someone actually asks during an outage. We keep the rehearsal stage because it saves your future self from becoming the person apologizing in a postmortem about a metric no one dared to delete.

We publish notes like this for alumni who want language to explain the pacing to teammates who only see the syllabus titles. If you want the hands-on version, the cohort pages list the next start dates we have confirmed with lab partners.