Help me design a decision-ready dashboard for [BUSINESS QUESTION] for [AUDIENCE]. It will use [DATA SOURCES] and be built in [TOOL]. The reporting period, refresh cadence, and grain are [TIMEFRAME AND GRAIN]. First, identify the 5-8 metrics that directly answer the business question. For each metric, define its business meaning, numerator, denominator where relevant, inclusion and exclusion rules, date field, aggregation, and comparison period. Distinguish leading indicators from outcome metrics. Do not recommend a metric simply because it is available in the data. Then produce: 1. A one-paragraph dashboard purpose statement and the three decisions it should support.
2. A metric dictionary in a table, including exact SQL, spreadsheet formula, or calculated-field logic appropriate for [TOOL]. State assumptions and name required fields.
3. A wireframe described section by section: top-level KPI cards, trend views, breakdowns, filters, drill-throughs, and exception alerts. Put the most decision-critical information first.
4. Data-model and query guidance, including joins, date spine needs, handling of nulls, duplicate records, late-arriving data, and division-by-zero cases.
5. A plain-English explanation of how each chart should be read and what action a concerning result should trigger.
6. A short build order and acceptance checklist.
Prefer comparisons against a meaningful baseline, not isolated totals. Avoid vanity metrics, overloaded charts, misleading dual axes, and filters that change a KPI definition silently. Ask up to 3 clarifying questions only if a required input is missing. Before answering, check that every recommendation traces to the stated business question and flag formulas that cannot be computed from the listed data.Fill in
| Placeholder | What to enter | Example |
|---|---|---|
| [BUSINESS QUESTION] | State the decision or operational question the dashboard must answer. | Which acquisition channels are producing retained paid subscribers at an efficient cost? |
| [AUDIENCE] | Name the people using the dashboard and the decisions they make. | Growth lead and channel managers deciding next month's budget allocation |
| [DATA SOURCES] | List the tables, spreadsheets, systems, and important fields available. | Google Ads spend by campaign and date; Stripe subscriptions; product events keyed by user_id; CRM campaign mapping |
| [TOOL] | Enter the target tool, such as Looker, Power BI, Tableau, SQL, or Google Sheets. | Looker Studio with BigQuery |
| [TIMEFRAME AND GRAIN] | State the reporting window, refresh frequency, and row-level grain of the data. | Last 12 months, refreshed daily; one row per user event or subscription transaction |
How to use
- Paste the prompt and replace every bracketed input with your dashboard context.
- Include a sample schema or existing query when metric definitions depend on unfamiliar fields.
- Check that every KPI has an owner, a defined denominator, and an actionable threshold.
- Follow up with: "Turn the wireframe into the exact calculated fields and SQL views I should build first."
Variations
Executive scorecard
Use this for a short weekly leadership view.
Design a one-screen executive scorecard for [COMPANY OR TEAM] covering [REPORTING PERIOD]. Its readers are [LEADERSHIP AUDIENCE], and the decisions are [DECISIONS]. Using [AVAILABLE DATA], select no more than six KPIs. For each, define the calculation, baseline, target, owner, and a plain-language interpretation. Describe a layout with KPI cards, only essential trends, and a short risks-and-actions section. Flag any KPI that lacks a reliable denominator or cannot be refreshed on [CADENCE]. Do not invent benchmarks. Ask up to 3 questions only if needed.
Operations dashboard
Use this for a team managing daily workload or service levels.
Create an operations dashboard specification for [PROCESS], used by [TEAM] every [CADENCE]. We need to monitor [SERVICE LEVELS OR OUTCOMES] from [DATA SOURCES]. Define queue, throughput, aging, backlog, quality, and staffing metrics only where supported by the data. Give exact calculation logic, the right grain, alert thresholds labeled as proposed, and a layout for triage at the start of a shift. Include rules for reopened work, missing timestamps, business hours, and duplicate tickets. Ask up to 3 questions if required inputs are absent.
Spreadsheet dashboard
Use this when the dashboard will live in Google Sheets or Excel.
Plan a [GOOGLE SHEETS OR EXCEL] dashboard for [QUESTION] using tabs or tables named [SOURCE TABS]. Specify the raw-data structure, helper columns, pivot tables, formulas, validation rules, and charts. Give formulas using the correct spreadsheet syntax and explain where each formula goes. Design the summary tab for [AUDIENCE], with filters for [FILTERS]. Include protections against blank values, text-formatted dates, duplicate IDs, and divided-by-zero percentages. Do not assume fields that are not listed. Ask up to 3 questions only when essential.
Tips
- Define the grain before calculating rates; a campaign-day table and a user-level table usually cannot be joined directly without duplication.
- Show both the numerator and denominator behind a rate when a small base can create a dramatic percentage swing.
- Use a single canonical date definition for each metric, such as signup date versus payment date, and label it in the dashboard.
- Test dashboard totals against a trusted source before adding segmentation, filters, or cohort views.
FAQ
Should I ask AI to choose all of my dashboard metrics?
Give it the decision and available data, then use its suggestions as a structured starting point. Metric ownership and business definitions still need agreement from the team that acts on them.
Can this prompt generate SQL and formulas?
Yes. Include the target tool and enough schema detail for the calculation to be grounded in actual fields.
How many KPIs should a dashboard have?
For a primary decision dashboard, five to eight is usually enough. Additional diagnostic metrics can live in drill-down sections.