A business dashboard should help leaders notice variance, ask better questions and assign action. It should not become a gallery of every number the software can display. Twelve well-defined indicators are usually more useful than a hundred unowned charts.
The exact measures depend on business model, but a growing company needs a balanced view of cash, demand, conversion, customer value, delivery, quality, inventory and capacity. Every KPI needs a formula, source, owner, target, frequency and decision.
1–2. Revenue and gross margin
Revenue shows recognized sales for the period using the finance-approved definition. Compare actual, plan and prior period by meaningful segment. Gross margin equals revenue minus direct cost, divided by revenue, and indicates whether growth contributes economically.
Document currency, timing, returns and exclusions. Rising revenue with declining margin may reflect discounting, product mix, input costs or classification. Investigate the driver before rewarding volume.
3–4. Cash runway and operating cash flow
Cash runway estimates how long available cash supports current net outflow and is most useful for sustained-burn businesses. Operating cash flow shows cash generated or consumed by normal operations and differs from accounting profit.
Use finance-controlled balances and scenarios. Review receivables, payables, inventory and committed expenses. A precise-looking runway should not hide uncertain collections, spending changes or growth assumptions.
5–6. Pipeline coverage and conversion
Pipeline coverage compares qualified opportunity value with a target for a defined period, using approved stages and probability. Conversion measures movement from one meaningful stage to another or from qualified opportunity to win.
Weak CRM data makes both misleading. Track stage ageing, next action and lost reason. Segment by source, product and market so one strong area does not hide deterioration elsewhere.
7–8. Customer acquisition cost and retention
Customer acquisition cost divides relevant sales and marketing cost by new customers for the same cohort and period. Retention measures customers or recurring revenue kept from a starting cohort according to a defined method.
State whether salaries, tools and brand spend are included. Compare acquisition cost with contribution and payback, not revenue alone. Cohorts reveal whether recent customers behave differently from the historical average.
9–10. On-time delivery and first-time quality
On-time delivery measures completed orders or projects against the confirmed promise. First-time quality measures outputs accepted without defect or rework. Together they prevent teams from improving speed by sacrificing correctness.
Define the promise, completion event and exception rules. Track cause and owner. Late work should remain visible even when teams repeatedly change dates, and customer-approved changes should be separated from internal failure.
11–12. Inventory turns and capacity utilization
Inventory turns compares cost of goods sold with average inventory; pair it with stockouts and service. Capacity utilization compares productive demand with available capacity using a stable unit such as hours, machine time or cases.
High turns with poor availability and high utilization with growing queues are warnings. Segment bottlenecks, protect maintenance and improvement time, and do not treat permanent overload as efficiency.
Implementation checklist
- Choose decisions: List weekly and monthly management decisions.
- Select balanced KPIs: Cover finance, customer, process and capacity.
- Write definitions: Specify formula, unit, period, source, owner and exclusions.
- Set thresholds: Use plan, capacity, risk and historical evidence.
- Validate data: Reconcile totals and test missing or late records.
- Design hierarchy: Show executive exceptions with drill-down to drivers.
- Assign response: Define who investigates each threshold and by when.
- Run reviews: Use commentary, decisions and owners, not chart narration.
- Track actions: Begin each meeting with prior commitments.
- Control changes: Version formula, source and target changes.
- Review usefulness: Remove measures that never change decisions.
- Protect access: Limit sensitive financial and people data by role.
Worked example
Prepare
Data owners close the period, reconcile exceptions and add short commentary on material variance. The meeting is not the first time anyone discovers a broken data source.
Scan
Leaders review cash, demand, conversion, customer, delivery, quality, inventory and capacity quickly. Only indicators outside threshold or changing materially enter discussion.
Diagnose
For each exception, separate fact from hypothesis. Drill into segment, cohort, location or process. A generated narrative is checked against source records.
Decide
Record one accountable action, owner and date, or explicitly accept the risk. An action such as monitor requires a threshold and next review point.
Follow through
The next meeting starts with prior actions and whether the expected driver changed. Repeated actions without movement indicate poor diagnosis, insufficient authority or unrealistic targets.
Improve the dashboard
Quarterly, test definitions, targets and usefulness. Add a KPI only for a real decision and retire charts that consume preparation without changing action.
Operating questions
What decision does this KPI support?
If nobody can describe the response to a meaningful change, the measure belongs in analysis, not on the executive dashboard.
Who owns definition and action?
The business owner approves formula and interpretation and coordinates response. The analyst may prepare data but should not silently redefine it.
What is the source of truth?
Name the exact system, field, close time and transformation. Reconciliation should reveal whether dashboard totals match controlled records.
What should be compared?
Use plan, prior period, relevant cohort or capacity—not an arbitrary green line. A target and forecast answer different questions.
How fresh must the data be?
Match refresh to decision rhythm and source quality. Real-time display is wasteful or misleading when operational decisions occur weekly.
Can users drill to causes?
The executive page shows exceptions; supporting views should expose segments and transactions without requiring a new manual spreadsheet.
Metrics
- Data ready by the agreed review time.
- KPIs with definitions and named owners.
- Reconciliation and correction rate.
- Exceptions with decisions and due dates.
- Actions completed and expected driver changed.
- Measures retired for low decision value.
- Manual preparation hours and duplicate reports.
- Access appropriate to user role.
Common mistakes
- Showing revenue without margin or cash.
- Using averages instead of segments and cohorts.
- Changing formulas without control.
- Treating targets as forecasts.
- Displaying precise numbers from weak sources.
- Rewarding utilization while queues grow.
- Reviewing charts without decisions.
- Adding KPIs whenever a tool offers a widget.
Frequently asked questions
Should every company use the same 12 KPIs?
No. Use them as a balanced starting point and adapt definitions to model, stage, risks and decisions.
How often should a dashboard update?
Match the decision and data rhythm. Real time is unnecessary when action is weekly or the source closes monthly.
What is a leading indicator?
It signals a driver before the outcome, such as qualified pipeline before revenue. It requires an evidenced relationship and an owner.
Who owns a KPI?
A business role accountable for definition, interpretation and response—not merely the analyst who prepares it.
How many KPIs belong on the first page?
Only enough to show balanced health and material exceptions. Detailed drivers should be available through drill-down.
KPI definition template
A dashboard is trustworthy only when every displayed number can be reconstructed. Apply this template to all twelve indicators and preserve the approved definition with the report.
Business question
State the decision or risk the KPI helps leaders understand. Remove measures with no plausible response. Record the definition, source, owner, exception and review date so another person can reproduce the decision. Test the procedure with a realistic normal case and one failure case. If evidence is weak, keep the action manual and improve the underlying process before adding automation.
Formula
Write numerator, denominator, sign, unit, rounding and treatment of zero or missing values. Record the definition, source, owner, exception and review date so another person can reproduce the decision. Test the procedure with a realistic normal case and one failure case. If evidence is weak, keep the action manual and improve the underlying process before adding automation.
Scope
Define entity, location, product, customer, channel and transactions included or excluded. Record the definition, source, owner, exception and review date so another person can reproduce the decision. Test the procedure with a realistic normal case and one failure case. If evidence is weak, keep the action manual and improve the underlying process before adding automation.
Time
Specify event date, period, close time, comparison window and treatment of late postings. Record the definition, source, owner, exception and review date so another person can reproduce the decision. Test the procedure with a realistic normal case and one failure case. If evidence is weak, keep the action manual and improve the underlying process before adding automation.
Source
Name the system, table or report and every transformation between source and dashboard. Record the definition, source, owner, exception and review date so another person can reproduce the decision. Test the procedure with a realistic normal case and one failure case. If evidence is weak, keep the action manual and improve the underlying process before adding automation.
Owner
Assign a business owner for meaning and action plus a data owner for quality and reconciliation. Record the definition, source, owner, exception and review date so another person can reproduce the decision. Test the procedure with a realistic normal case and one failure case. If evidence is weak, keep the action manual and improve the underlying process before adding automation.
Target
Separate plan, forecast, capacity, risk limit and historical benchmark because they answer different questions. Record the definition, source, owner, exception and review date so another person can reproduce the decision. Test the procedure with a realistic normal case and one failure case. If evidence is weak, keep the action manual and improve the underlying process before adding automation.
Threshold response
Define who investigates green-to-amber or amber-to-red movement, by when and with which drill-down. Record the definition, source, owner, exception and review date so another person can reproduce the decision. Test the procedure with a realistic normal case and one failure case. If evidence is weak, keep the action manual and improve the underlying process before adding automation.
Data quality
Display freshness, missing records, corrections and reconciliation so precision does not hide weak evidence. Record the definition, source, owner, exception and review date so another person can reproduce the decision. Test the procedure with a realistic normal case and one failure case. If evidence is weak, keep the action manual and improve the underlying process before adding automation.
Change control
Version formula, source, target and visual changes and communicate when comparisons are no longer like-for-like. Record the definition, source, owner, exception and review date so another person can reproduce the decision. Test the procedure with a realistic normal case and one failure case. If evidence is weak, keep the action manual and improve the underlying process before adding automation.
Keep a short change log beside the dashboard so users know when definitions, sources, targets or refresh timing changed and why comparisons may differ.
Final takeaway
Build the dashboard around decisions. Track a balanced set of financial, customer, operational and capacity measures, define each precisely, validate the data and turn every material variance into an owned action.
Sources and further reading
