Do not commission another dashboard first. Agree metric contracts, reconcile the source data and make customer-acquisition decisions only against contribution—not revenue alone.
01 / THE MANAGEMENT PROBLEM
More reporting had not created one trusted decision.
At fictional Northwind Retail, revenue, acquisition cost and repeat-purchase rate appeared in separate management views. The stated demand was for a consolidated KPI dashboard. That is a solution request, not yet a problem definition.
The real decision: should leadership invest in customer acquisition, retention or neither until the economics are reconciled? A dashboard that presents conflicting denominators faster would accelerate the wrong decision.
Commercial / finance leadership (proposed role for this exercise).
Monthly resource allocation, supported by weekly exceptions.
Illustrative 12 months, October 2025 to September 2026.
No validated contribution margin, paid-channel attribution or cohort profitability in the source sample.
02 / THE DIAGNOSIS
Revenue increased. So did customer-acquisition cost.
The original Northwind dataset contains monthly revenue, order count, average order value (AOV), customer acquisition cost (CAC) and repeat rate. The comparisons below use the first and last month—not an A/B test or causal estimate.
| Metric | Oct 2025 | Sep 2026 | Change | What it does not prove |
|---|---|---|---|---|
| Revenue | ₹435.7 lakh | ₹663.2 lakh | +52.2% | Incremental profit |
| Orders | 26,093 | 39,726 | +52.2% | More unique buyers |
| AOV | ₹1,670 | ₹1,670 | Flat at endpoints | Stable product mix |
| CAC | ₹429 | ₹546 | +27.3% | Worsening ROI without LTV |
| Repeat rate | 52.7% | 65.2% | +12.5 percentage points | Proof retention caused growth |
Source: existing ReadSigil synthetic data/cases.json. Endpoints may reflect seasonality and different customer mixes.
Exhibit 1 · Endpoint movement, indexed to 100 in October
SYNTHETICThree hypotheses to investigate
- Volume effect: transaction count, rather than basket size, explains the endpoint revenue change mechanically.
- Acquisition-quality effect: increased CAC could still be rational if retained contribution per acquired customer improved; LTV and channel mix are missing.
- Definition effect: repeat rate may use a different customer cohort or lookback window than revenue. Alignment is necessary before attribution.
03 / THE OPTIONS
Another dashboard is the wrong first investment.
| Choice | Advantage | Binding weakness | Call |
|---|---|---|---|
| Build a new executive dashboard immediately | Visible progress | Conflicting inputs survive | Reject first |
| Outsource a new analytics platform | More capability | No proven problem with the existing platform | Defer |
| Agree metric contracts and reconciliation controls first | Decisions share the same denominator | Needs business and finance ownership | Recommend |
| Do nothing | No delivery cost | Undetected divergence continues | Not viable long term |
04 / RECOMMENDED INTERVENTION
Build a governed metric layer, then a decision pack.
- STEP 01Contract definitions. Revenue net of returns? Paid orders or placed orders? CAC as channel spend divided by new paying customers? Repeat within what cohort and period? Agree and version each definition.
- STEP 02Reconcile source systems. Daily order, payment, return and advertising aggregates feed a reconciled table. All exceptions have owners and a documented waiver path.
- STEP 03Expose trade-offs. Revenue growth, CAC, contribution after variable cost, repeat behavior, channel mix and customer cohorts sit together with provenance.
- STEP 04Establish the forum. Finance signs off on margin; commercial leaders assign actions; BI owns lineage, freshness and unresolved breaks—not the business decision.
05 / VALUE & CONTROL
The dashboard is not the outcome.
Proposed acceptance thresholds below are design targets for a hypothetical pilot. They are not measured results for Northwind.
| Success measure | Proposed acceptance test | Evidence required |
|---|---|---|
| Metric agreement | 100% named critical KPIs approved | Versioned definition ledger |
| Source reconciliation | Variance within ±0.5%, exceptions documented | Automated source-to-report controls |
| Freshness | Before the agreed morning decision cutoff | Automated load timestamps and alerts |
| Productivity | 30% fewer manual report corrections | Pre/post effort and correction logs |
| Business value | Positive incremental contribution after acquisition expense | Channel cohorts, variable margin, valid holdout |
Value equation: incremental contribution = incremental net sales × contribution margin − incremental acquisition spend − incremental operating expense. A richer-looking dashboard gets no credit by itself.
06 / DELIVERY PLAN
A 90-day roadmap with actual decision owners.
- DAYS 01–30Finance and Commercial freeze KPI contracts; Data/BI documents source lineage and establishes the discrepancy baseline.
- DAYS 31–60Data Engineering creates reconciled tables and tests with at least three conflicting legacy reports. Owners sign off on exceptions.
- DAYS 61–90Commercial runs the decision pack with channel/cohort economics; independent finance review validates contribution and meeting actions.
Stop rule: do not approve scaled media reallocation based on this dashboard while CAC attribution, customer cohorts or variable contribution remain unreconciled.
07 / EVIDENCE LEDGER
What the case establishes—and what it cannot.
| Claim | Classification |
|---|---|
| Revenue, orders, CAC and repeat-rate endpoints | Computed from synthetic exercise |
| A metric layer is preferable to immediate dashboard duplication | Recommendation / judgment |
| Better retention caused the revenue change | Not established |
| The proposed control targets were delivered | Not claimed / not tested |
Decision: commission definition and lineage work first, then spend decisions against verified contribution. This is a worked analytical case, not an account of client results.
Underlying synthetic KPI/MIS report (PDF) ↗ · Original sample view ↗