Data-Driven F&B Delivery: Jahez as a Leading Example
summary
Search queries generate 55% of orders so a pipeline batches one week of queries, sessionizes fast keystrokes, filters item searches, applies Levenshtein distance grouping and a spelling-substring dictionary, then uses soundex Arabic-English matching to collapse pronunciations; results feed a missed-opportunities dashboard. Hundreds of restaurants join within two months, conversion rate from flagged leads doubles, and recent joiners contribute about 20% of their cohort revenue; embeddings are explored but not yet replacing the rule steps.
AI-written summary · about
chapters (7)
- 0:00Jahez and data-driven growth
- 1:09Jahez ecosystem overview
- 5:24External data products
- 10:17Internal projects and search misses
- 11:03Search Misses technical deep dive
- 17:53Impact of Search Misses project
- 18:15Q&A
related talks
27mOur experiments with food recommendations
08mAI-Driven Innovation for Enhanced Food Delivery Operations
38mDriving Customer Loyalty with Predictive and Conversational AI
35mData Science for Dinner: Recommending Recipes
20mJeremy Schiff, Senior Manager, Data Science, OpenTable
41mHow Deliveroo improved the ranking of restaurants
20mDataEngBytes 2023 - PER-T1-06 - Kamal Hossain
26mPersonalizing User Content - The Power of a Vector & Search DB