Inside Snagdar's demand score
Snagdar rates products on the stores it watches with a 0-100 demand score. Every point comes from a named signal with evidence you can read, so the score is explainable rather than a black box.
The signals
| Signal | Max | Evidence |
|---|---|---|
| Sell-through velocity | 40 | Observed going from live to fully sold out (within 2 minutes earns the maximum), or sizes selling out within 30 minutes |
| Resale premium | 25 | Resale value after eBay fees (~13%) and $8 shipping compared with retail (double or more earns the maximum). Estimates built only from asking prices count at most half, until sold prices confirm them |
| Scarcity language | 15 | Limited edition, exclusive, collaboration, numbered, purchase limit or raffle wording |
| Pre-release | 10 | Listed as unavailable before launch, or "coming soon" / pre-order wording |
| Freshness | 10 | Newly appeared or recently published |
| Cross-store | 10 | The same product showing up on several watched stores |
| Restock churn | 10 | Restocks that sell out again |
Scores are capped at 100. Tiers: very high 75+, high 55+, moderate 35+.
Relative to each store
Some stores sell out new listings in minutes as a matter of routine. Snagdar measures that from each store's last 14 days of launches and restocks: velocity, freshness and listed-elsewhere points are scaled down when fast sell-outs are common there (no change below 20%, down to a quarter of the points at 60% or more). On such a store a product also needs a second sign of demand to rank above moderate: sold-price resale premium, strong scarcity wording, or selling out fast on another watched store. Only stores where fast sell-outs are rare count as corroboration, and selling out on one earns the full cross-store points. Retail products and stores watched only with focus keywords aren't adjusted.
Only what was actually observed
The first time Snagdar scans a store, everything it sees is a baseline: it doesn't know when those products launched, so it can't claim they sold out fast. Speed is only measured from changes it watches happen afterwards. After downtime, anything that went live in the meantime gets no start time, so an outage can never make a drop look faster than it was.
From score to forecast
For products that haven't launched yet, Snagdar adds history: how the same brand did at the same store (then the brand elsewhere, then similar products, then the store overall), how fast those drops sold out, how often they restocked and what they resold for. The result is a forecast with a heat score, expected profit, a confidence level and a recommendation: stage now, watch or skip. Confidence rises as more comparable drops are recorded.