Find where browsing stops becoming a purchase

Define what the events actually show

The source records views, cart events and purchases. Mehrab cleaned and grouped events into sessions, then compared journeys by category, product and behavioural segment.

Cart behaviour is the measurable purchase intent stage. The source preparation reuses the cart flag as a checkout intent proxy for its generic analysis pipeline. It records no separate checkout event, so it cannot identify where checkout itself failed or explain why an individual left.

Give the product team a place to start checking

The largest measured loss came between product detail and cart. Category comparisons helped identify where navigation and product information deserved review.

A propensity model explored high intent non buyers. It offers a review queue and testable hypotheses, not proof of increased purchases. Usability research and controlled changes would be next.

A concrete next check is whether important product information is missing or hard to compare on the product page. Another is whether items in the unknown category need clearer taxonomy. Each hypothesis would be checked with observation or a controlled test before claiming an improvement.