Sales Pipeline — CRM API to Power BI

B2B pipeline report fed from a CRM API: an incremental extractor, a DuckDB warehouse that keeps every version of every deal, and a Power BI model with incremental refresh. Shows how much of the open pipeline is real.

Preview

Highlights

An apparent 'fast deals lose' finding turned out to be a source-data quirk in three months; the chart says so instead of selling it
Built and validated as code; not yet opened in Power BI Desktop

Problem

The headline pipeline looked healthy, but nobody knew how much of it was deals that would never close, whether win rate differed by person or product, or whether each monthly API sync had loaded cleanly.

Approach

Incremental REST extraction with retry and throttling logs; warehouse keeps deal history so the pipeline can be rebuilt as of any sync date
Beginner to intermediate: KPI cards, funnel, deal-age bands, what-if sliders for the stale threshold and forecast win probability, field parameters
Advanced: decomposition tree, key influencers, small multiples, window functions (OFFSET, WINDOW, INDEX), ISINSCOPE share of parent, many-to-many funnel bridge
Expert: DAX user-defined functions, calendar time intelligence group, visual calculations (running sum, moving average, rank), object-level security, mobile layout, CI checks

Stack

Power BIDAXTMDLPBIRREST APIDuckDBPython

Outcome

71% of the open pipeline ($3.51M of $4.97M) is stale: older than the 104 days in which 90% of deals close
Win rate barely differs between managers (62–64%) or products, so revenue gaps come from team size and deal mix, not closing skill
Model: 19 tables, 46 measures, 6 functions; 8 report pages, 191 visuals; both validators pass