Daily Prompt Pack: Metric Dashboards – Build the three charts your board actually reads

Most dashboards are decoration. They have twelve tiles and nobody knows which two numbers matter. I build the opposite kind: three charts, each one tied to a decision someone can actually make this week. Today’s three prompts do the pruning, the auditing, and the wiring.

[Claude]

You are a skeptical board member. I’ll paste 10-25 metrics below with their current values. For each, answer: what decision would change if this number moved 20%? If none, mark it CUT. Output a table with columns Metric, Owner, Decision It Drives, Verdict (KEEP/CUT), then name the 3 that survive and the single chart type for each. My metrics: [paste]

[ChatGPT]

Audit this dashboard definition for vanity metrics and lagging indicators. Input: a list of tiles with name, formula, refresh cadence. For each tile flag: VANITY (only goes up), DOUBLE-COUNTED (derivable from another tile), or LAGGING (no leading companion). Output a prioritized fix list with the replacement metric and its formula for each flagged tile. Dashboard tiles: [paste]

[Cursor]

Given this schema, write one SQL query per chart for a weekly exec dashboard: new MRR by plan, net revenue retention, and activation rate within 7 days of signup. Use CTEs, one query per file section, comment the grain of each row. Return only runnable Postgres SQL plus a 3-line README explaining the join keys. Schema: [paste]

The full archive lives at aicofounderstack.com/tag/prompt-pack. New pack every morning.