A proprietary, point-in-time panel of marketplace demand — sales-rank (BSR) trend and a units proxy, review-velocity, price index and new-launch rate across a fixed product panel. Aggregated, ticker-mapped signals that read brand and marketplace momentum on AMZN and MELI before it prints.
We snapshot the same product panel on each marketplace every day and archive it point-in-time. A stable panel is what makes the rank-to-units mapping and the price and review series comparable through time — the signals are strongest measured on a consistent basket rather than a shifting catalog.
Each signal is an aggregated time series at the marketplace × category level, pre-mapped to the tickers it informs — engineered for demand nowcasting on a fixed panel.
Daily best-seller-rank snapshots converted, on a fixed panel, into a units-sold proxy via a category rank-to-units mapping. Aggregated, it nowcasts marketplace and brand demand direction ahead of reported growth.
Net new reviews per product per day, a second independent demand read that cross-checks the BSR-based units proxy. Divergence between the two flags rank manipulation or promotion-driven spikes.
A matched-basket price index across the panel, rebased to 100 — the buy-box and list price trajectory that drives marketplace take-rate and 1P gross margin. Reads promotional cadence and pricing power by category.
Live count of active listings plus the rate of net new-product launches and discontinuations. Accelerating launches signal seller and brand investment in the marketplace; contraction flags category rotation or share loss.
Each marketplace and region is normalized to one schema and mapped to the equity whose GMV and take-rate it drives, so the desk pulls a single blended demand read per name.
Engineered to pass a fund's data-diligence: point-in-time integrity, transparent aggregation, and honesty about where the units proxy is precise and where it is noisy.
Every BSR, price and review-count snapshot is archived with its capture timestamp on a fixed panel. Series reconstruct exactly as they stood — no survivorship from catalog churn, no look-ahead.
The rank-to-units mapping is a standard industry technique but inherently noisy at the single-SKU level. It is most reliable aggregated over a fixed panel and read as direction and rate of change, not an absolute unit count.
Derived only from publicly observable listing surfaces. No personal data, no reviewer identities, no seller-account records, nothing material and non-public. GDPR-aligned and DDQ-ready.
An illustrative slice: the aggregated BSR-derived units proxy for the fixed US panel, indexed, alongside median BSR and the independent review-velocity read that corroborates it.
| Month | Median BSR | Units proxy | Rev. velocity | MoM |
|---|---|---|---|---|
| 2026-03 | 2,140 | 110.4 | 1.92 | +1.4% |
| 2026-04 | 2,060 | 112.9 | 1.98 | +2.3% |
| 2026-05 | 1,980 | 114.6 | 2.03 | +1.5% |
| 2026-06 | 1,910 | 116.1 | 2.08 | +1.3% |
| 2026-07 | 1,840 | 118.6 | 2.14 | +2.2% |
| 2026-08* | 1,805 | 119.7 | 2.17 | +0.9% |
Request a point-in-time sample — a marketplace × category slice on the fixed panel, with the units-proxy methodology and ticker mapping. We'll send a data dictionary and a backtest walkthrough.
Request a Sample →