Discuss a small first task →

Ivan Matiushkin

Why CRM, ERP and spreadsheet revenue totals disagree

Different revenue totals are not automatically broken data. A won CRM deal, a posted invoice and a historical revenue row represent different events. Reconciliation starts by defining which question each total answers.

Independent working demonstration using synthetic data. Not a client deployment.

Define the metric before comparing totals

Booked Revenue uses accepted closed-won CRM deals. Invoiced Revenue uses accepted posted ERP invoices. Reported Revenue combines accepted history through June 2026 with ERP invoices from July 2026. CRM deals are not added to that total. Booked minus invoiced can reflect timing; a won deal without an invoice is a business variance, not necessarily a data defect.

Choose the authoritative period

In the working example, history is authoritative through 30 June 2026 and ERP from 1 July. Two July historical records total €4,050 and are excluded as overlap. They remain visible in the evidence. Adding every historical row to every invoice would count an overlapping period twice.

Explain excluded amounts, including dirty records

The raw historical EUR total is €87,000. Initially €64,200 is accepted; €18,400 awaits Northstar customer mapping, €4,050 is overlap and €350 has an invalid period. ERP contributes €36,800 in accepted invoices. A duplicate and unsupported product are excluded under explicit rules; an unknown currency remains in a separate bucket with no invented FX conversion.

Correct the mapping, then run reconciliation

Initially Reported Revenue is €101,000: €64,200 + €36,800. Map Northstar to CUST-1042 and explicitly re-run reconciliation. Accepted history becomes €82,600 and Reported Revenue becomes €119,400. The €18,400 change is explained by one customer mapping; other source records are unchanged.

Read PASS within its definition

Reconciliation PASS means monetary records are accounted for as included or explicitly excluded, controls agree and no blocking configuration exception remains. Source Data Quality can still show WARNING because duplicate, date, product and currency defects remain in the original data. PASS does not erase them.

Trace the result before trusting the dashboard

Open Revenue BI, inspect the sources, run reconciliation, correct Northstar and reconcile again. In the dashboard, trace July revenue to its ERP invoices and excluded historical rows. Download the HTML or JSON evidence. For your own data, a first assignment can define one metric, reconcile two sanitised exports and explain every difference before dashboard development.

Related practical examples