Specify a KPI Dashboard Before Anyone Builds It
Define the metrics, thresholds, layout and data prerequisites for a business dashboard — so the build is assembly, not archaeology.
When to use it: Use before building any dashboard — the spec forces the metric definitions, sources and triggers that make one worth glancing at.
You are a business-intelligence designer for an Australian small business. Write the full specification for a KPI dashboard BEFORE any tool is opened — a dashboard is a set of decisions rendered on a screen, and the decisions come first.
<context>
WHO READS IT: [OWNER ONLY? TEAM ON THE WALL? — honesty here changes everything]
DECISION CADENCE: [WHEN IT GETS LOOKED AT AND WHAT GETS DECIDED — e.g. Monday planning, monthly pricing review]
CANDIDATE METRICS: [EVERYTHING BEING CONSIDERED, WITH WHERE EACH NUMBER LIVES — e.g. weekly sales (Square), quote conversion (spreadsheet), review rating (Google), cash buffer (Xero)]
TARGETS: [ANY EXISTING TARGETS OR THRESHOLDS, AS THEY STAND]
LIKELY BUILD TOOL: [e.g. Google Sheets, a BI tool someone mentioned — or undecided]
</context>
Before specifying, cull the candidates: every metric must pass 'if this number moved sharply, would we DO something different?' — metrics that inform no action are decoration, and eight is the ceiling.
<task>
1. For each surviving KPI (max 8): a plain-words definition and formula (numerator, denominator, period — precise enough that two people compute the same number); source system and the export/lookup that produces it; owner; refresh cadence matched to how fast the number can meaningfully change; target or threshold from MY inputs (else [NEEDED: set after 4 weeks of baseline]); and the action a red state triggers — a threshold without an owner and an action is a colour, not a control.
2. Layout spec, tool-agnostic: glance order (top-left carries the metric that most changes decisions), every number shown WITH its comparison (vs target, vs same period last year — name which), trend sparkline vs point-in-time for each, and a text sketch of the grid.
3. Cadence trap check: flag any KPI whose refresh is slower than the reading cadence (a monthly number on a weekly dashboard reads as 'no change' three weeks out of four — show it monthly instead).
4. Data prerequisites: for each source, what must be true for the number to be trustworthy (consistent categories, up-to-date reconciliation) and the one-off clean-up needed first.
5. Anti-vanity note: the 2-3 candidate metrics you culled and why — so they don't creep back in.
6. Build handoff: the open questions the builder must not answer on your behalf ([NEEDED] items, threshold sign-offs).
</task>
<output_format>
KPI spec table → layout sketch → cadence check → prerequisites → culled list → open questions.
</output_format>
Rules: no metric without a source I stated; no invented targets or industry benchmarks; financial thresholds that carry tax or compliance weight (e.g. GST registration turnover) are facts to confirm with the accountant, not numbers to assert. En-AU spelling.
Copy the block above straight into Claude — anything in [BRACKETS] is yours to fill in.
Want it tuned to your business? Bring it to the free weekly call and we'll adapt it live.
Join the free callMore data analysis prompts
Decide Which Customer Questions Your Data Can Actually Answer
Triage your customer-behaviour questions against the data you really hold, and get spreadsheet-level recipes for the answerable ones.
Match the Right Simple Analysis to Your Business Data
Pick analysis techniques that fit your question, data shape and sample size — with spreadsheet steps and misread warnings.
Use Sales and Feedback Data to Back Your Best Offerings
Combine sales, margin and feedback evidence into a keep/grow/fix/kill call on each product or service.