| Series | Source | Notes |
|---|---|---|
| Plan / goals | CFO Model — SW GROWTH_DTC NEW vs RETURN, live plan block (rows 4–8) for Sep–Dec; Aug held on its plan | Embedded from a live read on 2026-09-09 (v109F, modifiedTime 2026-09-04). Which block is live depends on the month — AUG FCST (rows 45–49) is an August snapshot and went stale for Sep–Dec once August closed. A closed month reads the same in both blocks; the last closed month is where they diverge. If the model re-versions, re-verify the row map before rebuilding. |
| Shopify actuals | Shopify connector — ShopifyQL FROM sales SHOW net_sales GROUP BY new_or_returning_customer TIMESERIES day | net_sales = gross − discounts − returns. NOT total_sales (runs ~3.6% hot). Ties to the model to the cent (Jan New $242,112.39). Empty-label bucket discarded. |
| Amazon actuals (total) | Windsor.ai → Amazon Seller Central, Sales & Traffic by date | Ordered Product Sales, filtered to the SW account A2HKM3NALUG752-US (a second Seller Central account is now connected to Windsor and is excluded). Ties to the model within ~1% most months. |
| Amazon New / Returning | AFN shipment warehouse (embedded monthly, Jan–Jul 2026); current month estimated at July's 44.6% new share | Shipment basis — won't tie exactly to Ordered Product Sales (Jul ran −10.7% on spillover). The CFO Model has no Amazon split; plan cells show —. |
Day-weighting: monthly plan is spread using SW's 2026 day-of-week curves (New CV 0.119, Returning nearly flat; Amazon flat) — never ÷4, never flat-week.
Reading variance: Jan–Jul 2026 plan columns are the model's own closed actuals, so variance should read ≈0 — a gap means the connector and model disagree (see reconciliation below). Only Aug–Dec is a forecast to beat. YTD vs plan mixes the two; treat it as directional.
Amazon reconciles Ordered Product Sales (connector) against the model's Amazon revenue row. Small gaps are timing (order vs shipment) and late returns.