Mining Common Purchase Sequences

The monday Data Analytics team wants to understand which actions most often lead to a purchase in monday Work Management. Given ordered user events, write a PostgreSQL query that identifies the most common sequence of the three actions immediately preceding a purchase.

Requirements

1. Partition event ordering by user_id and session_id, and order events chronologically using event_time and event_id as a tie-breaker.
2. Consider only purchases that have three preceding non-NULL actions in the same session.
3. Represent each sequence in chronological order, such as view_board > open_item > invite_member.
4. Count purchases and distinct users for each exact sequence.
5. Return every sequence tied for the highest purchase count, including its rank, ordered by purchase count descending and sequence ascending.