Direct ordering & revenue
Lift average check size using a neural recommendation engine. Surface high-probability add-ons and co-purchase pairings learned from your real-time order history.
Menuella
Predictive neural upsells
Lift average check size using a neural recommendation engine. Surface high-probability add-ons and co-purchase pairings learned from your real-time order history.
Relevance drawn from live baskets in your venue—not generic, site-wide promo banners.
Dry-Aged Beef & Ale Pie
£17.60
Rankings reflect combinations guests already buy together—not a one-size-fits-all cross-sell script.
Lift ticket size without extra taps; the best upsell feels like a shortcut.
Bundles, LTOs, and margin floors stay under your control while models handle ranking.
The same co-purchase panel guests see on each product—ranked from real baskets, not generic promos.
Tomato, mozzarella, and fresh basil—an example dish guests might open before seeing pairings.
£12.90
Scores update as orders come in; if there isn’t enough data yet, this block stays hidden for that product—just like in your venue.
Trigger logic
Dry-Aged Beef & Ale Pie
£17.60
Guest phone
Suggested add-on
Complete your meal: garlic bread pairs with 62% of orders like yours. Add for €4.50.
Pairings come from co-purchase patterns in your completed orders—similar to collaborative filtering—so suggestions reflect what guests already combine in your venue.
You can set guardrails for bundles, LTOs, and margin floors while the model handles ranking—so strategy stays human and execution stays fast.
Triggers can respect cart value, service mode, and time windows so prompts appear when they help—not when they would slow checkout.
If there isn’t enough order data for a dish yet, the block can stay hidden for that product—avoiding noisy or random suggestions.
Upsells run on your direct ordering channel with the same guest profile and menu—no marketplace middleman taking a cut of the lift.
From the blog
Nuh Kayran Collaborative filters and neural rankers can surface the next dish—but margin lives in what the kitchen can execute and what the guest would have loved anyway. How to use order history for high-margin discovery without randomizing your cart.
Nuh Kayran Co-occurrence and margin tell you which pairings deserve the spotlight. How to turn sales history into menu and upsell rules—without fighting the kitchen or allergen reality.
Nuh Kayran Upsells fail from bad timing, not bad items. When to suggest add-ons in the funnel, how many options to show, and how to respect cognitive load on mobile first-party checkout.
