Real-time behavioral tracking

Know who's about to buy. Before they do.

Traceum streams user behavior into a live ML model and predicts purchase intent in real time — so you can show the right offer at the right moment.

f0–f11 Features

Cart value, scroll depth, session time, clicks — 12 live signals per user.

PSM Scoring

FastAPI model returns a propensity score per session within milliseconds.

Outcome Tracking

Conversion events close the loop for weekly automated retraining.

Live feature vector sample

session_id: "a3f9b2c1..."
feature_vector: [
  2,      // f0: cart_items
  84.99,  // f1: cart_value
  3,      // f2: categories_seen
  7,      // f3: product_views
  2,      // f4: search_count
  312,    // f5: session_duration (s)
  68.4,   // f6: scroll_depth (avg %)
  1,      // f7: wishlist_adds
  24,     // f8: click_count
  1,      // f9: checkout_starts
  0,      // f10: device_type (desktop)
  1       // f11: referral_encoded (organic)
]
propensity_score: 0.847