Claude prompt to create a dashboard

Help me design a decision-ready dashboard for [DECISION MAKERS] to manage [BUSINESS QUESTION]. The available data source and tables or sheets are [DATA SOURCES], with these field definitions and known data issues: [DATA DEFINITIONS]. The reporting period and refresh cadence are [TIMEFRAME]. The dashboard will be built in [TOOL]. These are the metrics or targets already approved: [APPROVED METRICS].

Do not start by choosing charts. First, state the decisions this dashboard should support, the action each decision owner can take, and the grain of the data needed. Separate leading indicators from outcome metrics and distinguish counts, rates, averages, and cumulative values.

Produce:

1. A metric dictionary table with metric name, business definition, numerator, denominator, inclusion and exclusion rules, date field, data grain, owner, target, and calculation. Use explicit formulas, spreadsheet formulas, SQL, or pseudocode appropriate to [TOOL]. Do not calculate a rate by averaging already-aggregated percentages unless weighted logic is correct.
2. A data-quality checklist covering missing values, duplicates, late-arriving records, timezone/date boundaries, join multiplicity, test data, and denominator-zero handling. State which checks can materially change each metric.
3. A dashboard wireframe in text: top-level KPI cards, trend section, breakdowns, detail table, filters, and drill-down path. For each visual, name the question it answers, chart type, dimension, metric, and a reason it is preferable to alternatives.
4. Build instructions in a numbered sequence, including query or formula logic, filter behavior, default date range, definitions tooltip, access considerations, and refresh or alert rules.
5. A plain-English readout template that explains what changed, likely drivers supported by the data, caveats, and the next operational question to investigate. Do not claim causation from a dashboard pattern alone.

Do not invent field names, targets, data, query results, or business rules. Flag any metric that cannot be defined from the supplied data. Before answering, check that every chart maps to a stated decision and that all percentage metrics have a valid denominator and time basis. Ask up to 3 clarifying questions only if a required input is missing.

Design an executive scorecard for [LEADERS] covering [BUSINESS AREA]. Use these approved KPIs and definitions: [METRICS]. Data is available in [DATA SOURCES] for [TIME PERIOD]. Return no more than eight KPIs, each with formula, target, comparison method, owner, and action trigger. Include a text wireframe and a monthly narrative template. Exclude metrics that have no decision owner or reliable denominator. Flag data-quality, attribution, and target-setting limitations; do not invent benchmarks or performance explanations.

Create a Google Sheets or Excel dashboard plan using [SHEET TABS] to track [QUESTION]. The key columns and definitions are [COLUMNS], and the required KPIs are [KPIS]. Provide sheet structure, named ranges or tables, exact formulas, pivot-table setup, chart specifications, and error handling for blanks and zero denominators. Include a reconciliation check from dashboard totals to the raw-data tab. Do not assume fields exist or use volatile formulas when a simpler approach works. Flag calculations that need row-level data not supplied.

Write a dashboard-ready SQL plan for [DATABASE DIALECT] using [TABLES] to answer [QUESTION]. Define the desired reporting grain, joins, date logic, filters, and approved metric definitions: [METRIC RULES]. Produce a commented query or CTE sequence, a data-quality query for duplicates and null keys, and a validation query that reconciles totals. State assumptions about timezone, late-arriving data, and zero denominators. Do not invent schemas, fields, or join keys; flag missing definitions before writing speculative SQL.