MLCompoundingHigh complexityLevel 3 Advanced decisioning
Bespoke Recommendations
A recommender built on your full first-party data, tuned to your catalogue, margins and strategy, going beyond what Braze's out-of-the-box engine can see.
Recommendations you own and control, tuned past the Braze defaults for the products and margins that matter.
Who it's for
Brands where Braze Recommendations has proven the value and now want control: their own logic, margins and catalogue nuance. The upgrade step.
The engagement
What you get
- A bespoke recommender model over your first-party data
- Recommendation scores written to Braze
- Ranking logic tuned to catalogue, margin and strategy
- Model monitoring and periodic retraining
What Fuse does
- Assemble the purchase, catalogue and margin data
- Train and validate the recommender against Braze Recs as the baseline
- Write recommendations to Braze and wire them into journeys
- Monitor and retrain on an agreed cadence
What we need from you
Data & access
- Full purchase history and product catalogue
- Margin / strategy inputs for ranking
Your responsibilities
- Provide full purchase history, catalogue and margins
- Confirm the ranking objective (revenue, margin, retention)
Outcomes and proof
- Recommendations tuned past the out-of-the-box defaults
- Ranking aligned to margin and strategy, not just clicks
- Lift proven against Braze Recs as the baseline
Assumptions
- Enough first-party history to train a recommender
Out of scope
- The Braze Recommendations switch-on, which is the Level 1 service
- Catalogue data engineering beyond the agreed feed