dropwin

Methodology — data accuracy disclaimer

Last updated: 2026-05-23.

Informational only, not financial advice. Revenue, traffic, and ad-velocity estimates are derived from public signals (Meta Ad Library, Shopify /products.json, sitemap, atom feed) and are surfaced with a 0-100 confidence score. They are not business-grade accounting numbers and should not be used as the sole basis for purchasing decisions.

Where estimates come from

  • Monthly revenue estimate: derived from observed product velocity + catalog signals + ad scaling signals. Confidence drops for stores with fewer than ~30 days of observation or ads in fewer than 2 countries.
  • Ad velocity: computed from ad_metrics_history snapshots via the guarded-bounded ln-based formula documented in docs/superpowers/specs/2026-05-20-winning-products-saas-design.md §5.
  • Supplier confidence score (0-100): combines title similarity, image similarity, and price-band proximity per the M1 confidence rubric.
  • Active ads count: ads with last_seen_at within the last 30 days joined through products to the store.

How we compute the revenue confidence score

Every revenue estimate ships with a 0-100 confidence score. The score combines five categories of signal:

  1. Observation window — how long we have been tracking the store (newer = lower confidence).
  2. Ad-tracking breadth — how many distinct ads we have seen for the store's products (more ads = more independent velocity samples).
  3. Market coverage — how many countries the store is advertising into (multi-market = stronger signal of a sustained business).
  4. Catalog size — mid-sized catalogs (5-50 products) are the sweet spot; very small or very large catalogs lower confidence.
  5. Velocity stability — how consistently the store's ads scale over time (volatile = lower confidence).

We do not publish the exact weights — exposing them creates an incentive to game the score. The signal categories above are stable; weights get recalibrated against verified post-hoc revenue data as the dataset grows.

Concrete examples

  • High confidence (~90+): a store we have tracked for 4 months with 60+ distinct ads across 5 countries, a 25-product catalog, and a stable ad-velocity curve. The revenue figure is useful for shortlisting.
  • Medium confidence (~55): a store we have tracked for 6 weeks with 12 ads in 2 countries and a single product. Directional signal — useful for "this store is real and spending" but not a number to plan around.
  • Low confidence (~15): a store first seen 8 days ago with 3 ads in 1 country. We surface the figure with a "low confidence" badge — treat it as "exists and ad-spending" rather than a revenue claim.

Known limitations

  • Small stores (< $30k/mo MRR) have low-confidence estimates.
  • Supplier matches default to CJDropshipping when available; AliExpress matches are best-effort via public search.
  • Revenue timeline is weekly resolution; doesn't reflect intra-week promo bursts.

Questions: support@dropwin.app.