Power BI for Lending: Origination, Portfolio, and HMDA Dashboards

A lender builds Power BI dashboards. The CEO reviews them and notices the origination funnel shows one number for pull-through, the board package shows a different number, and the investor report shows a third. Each is calculated correctly within its own logic, but the definitions differ: pull-through based on applications received vs pull-through based on completed applications vs pull-through based on pre-approvals. The CEO asks for a single source of truth. The data team agrees it's needed. Six months later, the three numbers still don't match because the semantic layer was never built and each dashboard was built by a different analyst pulling from the LOS with different calculations.

Lender Power BI done right locks definitions in the tabular semantic model. One pull-through definition. One vintage. One delinquency status bucket mapping. One HMDA LAR calculation. Sourced from the governed data layer, documented, version-controlled. All dashboards consume from this model. The CEO, CFO, CRO, and compliance officer see the same numbers because they're sourced from the same governed definitions. Done this way, the data team maintains one model; every team uses it. Done without the semantic discipline, every new dashboard creates a new version of the truth.