AI can generate a weekly metrics digest that says what changed, where the number came from, and what still needs explanation.

Build a reporting contract before you automate the summary.

List the metrics, source locations, date window, owner, target or comparison basis, and any calculation that happens outside the dashboard. If revenue is net of refunds in one source and gross in another, the report needs to say so.

This contract turns a request for “the weekly numbers” into a repeatable job. Strawberry can then use the selected dashboards and tabs to collect the defined inputs rather than guessing what belongs in the executive update.

Put the change beside the evidence.

A metric moving from 12 to 18 can be a 50 percent increase, but the base size still matters.

Show the current value, previous value, absolute movement, and percentage movement where both are useful.

Link the dashboard, report, or underlying table behind each material change. The person reading the digest should be able to investigate without starting a fresh hunt through six tools.

Use the digest to surface questions rather than manufacture explanations.

A change may come from product behavior, seasonality, a data delay, a campaign, a booking correction, or an instrumentation issue. Label confirmed drivers separately from hypotheses and unknowns.

The best weekly digest is short enough to read and specific enough to start the right conversations. It should name the metric owner who can validate a major anomaly.

Keep the weekly process stable as the company changes.

Save the reporting contract and output shape as a skill.

Run it as a routine only after the sources and validation steps work reliably, then review changes to the metric set as deliberately as you would change a financial definition.

A routine can do the collection and draft the narrative. It should flag a dashboard that moved, a metric that disappeared, or a calculation that no longer reconciles.

Frequently asked questions