PURE · Seller Intelligence Quality
Likely-List AVM Quality

PURE · Prospecting · Model health

Seller Intelligence Quality

One scoring engine predicts who lists next and which features move price & DOM — this is its health: precision vs. who actually listed, coverage, and the loop that keeps it sharpening.

Sample metrics — sign in for live model health
Top-decile precision
Households scored
Predicted listed
Signals tracked
Farms to refresh

The engine family — one score, many surfaces

Each surface reads the same seller-propensity model, so a smarter score lifts all of them.

Continual-improvement loop

Nightly the model re-scores the farm, backtests against who actually listed, and re-weights the signals that predicted it. Run any step now:

Model health by farm

Per farm/zip — scored households, how many were predicted-to-list, how many actually listed (backtest), and the resulting precision.

Farm · zipScoredPredictedListedPrecisionRefreshed