Results

A metric result carries a reporting window, dimension key, numerator, denominator, value, source row count and status. It can be Calculated, Suppressed or Invalid, with a reason where needed. Its links to a property, metric definition and operation run give an analyst the path back to the source and the operation that produced it. The value alone is a poor basis for a release decision.

Metric observations add readings tied to both the measured property and a source snapshot. A snapshot records its source system, capture time, coverage window, record count, source hash and freshness. Fresh, Stale and Failed help the team decide whether the source is current enough for the reporting period being discussed.

• Observation. The period, unit and snapshot tie a reading to its source capture.
• Definition. The metric grain, version and aggregation method explain the intended measure.
• Result. The value and status remain linked to the run and property used for review.

Define the reporting windows that the Analytics Lead will compare. A conversion value for one day should not silently sit beside a month of sessions. Agree how a stale snapshot is handled before a result is used in a published discussion. Decide who reviews a definition’s Approved status, because an older or Draft definition may describe a different measure.

When comparing result values, keep the numerator and denominator visible. A zero denominator needs a meaningful outcome rather than a fabricated rate. A Suppressed or Invalid result should retain its reason so the reader knows why it cannot be used as a normal observation. The source row count helps the team investigate a result derived from an unexpectedly small set of facts.

The Analytics Lead sees an apparent conversion increase for a landing page. They open the property, check its session and conversion values, then inspect the result’s reporting window and linked definition. The Analyst checks the associated run and the snapshot’s coverage and freshness. If the observed period differs from the property period, the team corrects the comparison before presenting a trend.

A Reviewer can follow the same trail from decision to source revision, then to result, run and underlying observations. A follow-up action can assign any discrepancy to the Property Owner. This keeps a reporting conversation focused on the specific source and period, rather than asking someone to trust a chart without its working record.

• Calculated Results — Compare value, status, reason and reporting window.
• Collection Schedule — Check snapshot coverage and freshness.
• Property Lifecycle — Return to the accountable measured source.

Use compare cohorts for the read-back of outcome fields and review and publish for the release decision.