Multi-banner retailer: 3,400 assortment decisions per week
A global multi-banner retailer deployed xyner across merchandising and assortment decisions — 3,400 weekly decisions made with full traceability, 11% margin lift in the targeted categories.
What the customer was up against.
A multi-banner retailer was making ~3,400 weekly merchandising decisions across 8 banners and 22 countries with inconsistent rigour, limited cross-banner learning and no central audit. Margin performance varied 800 bps across similar categories in different banners; promotion outcomes were not systematically captured.
What xyner built.
Deployed xyner with specialist agents for assortment optimization, price recommendation, promotion design and supplier negotiation prep — fully integrated with the merchandising system, the POS data lake, supplier portals and the promotions platform.
Measured impact.
11% margin lift in pilot categories; assortment decisions now traceable across all banners; cross-banner learning meaningfully accelerated; supplier-negotiation outcomes improved as buyers entered conversations better prepared.
At a glance.
Situation
A multi-banner retailer was making ~3,400 weekly merchandising decisions across 8 banners and 22 countries with inconsistent rigour, limited cross-banner learning and no central audit. Margin performance varied 800 bps across similar categories in different banners; promotion outcomes were not systematically captured.
Intervention
Deployed xyner with specialist agents for assortment optimization, price recommendation, promotion design and supplier negotiation prep — fully integrated with the merchandising system, the POS data lake, supplier portals and the promotions platform.
Outcome
11% margin lift in pilot categories; assortment decisions now traceable across all banners; cross-banner learning meaningfully accelerated; supplier-negotiation outcomes improved as buyers entered conversations better prepared.
Retail · Merchandising & assortment
A global multi-banner retailer (grocery, apparel, hardgoods)
Global — 22 countries across 8 banners
Assortment optimization, category pricing, promotion design, supplier negotiation prep
12 weeks pilot, 8 months full rollout
From contract signature to full rollout.
What the deployment actually looks like.
Multi-banner retail requires careful balance between cross-banner learning and per-banner autonomy. The deployment uses banner-scoped data planes coordinated by a cross-banner learning layer that shares patterns, not data.
Assortment Optimization Agent
Continuously evaluates per-store assortment performance against demand, margin, space and SKU-level economics; recommends additions, deletions, range expansions.
Price Recommendation Agent
Recommends price actions per SKU and store cluster based on margin, competitive position, demand elasticity and category strategy.
Promotion Design Agent
Designs promotions optimised for ROI, footfall, basket-build or competitive defence — depending on category goal — with full cost-of-promotion accounting.
Supplier Negotiation Prep Agent
For category-management meetings, assembles supplier-specific briefing materials — performance history, share-of-shelf, joint-business-plan delivery, competitive supplier alternatives.
Cross-banner learning layer
Shares anonymised patterns and pattern outcomes across banners without sharing raw customer or supplier data; respects competition-law concerns.
POS & supply integration
Real-time integration with POS data lake, inventory positions, supplier portals and the promotions platform; respects supplier-specific data-sharing agreements.
How the rollout sequenced.
A 12-week pilot in one banner and one category before phased rollout across categories and banners.
Foundations
Deploy data plane with banner-scoped isolation; integrate merchandising system, POS data lake, supplier portals and promotions platform.
Agent configuration
Configure four agents against the pilot banner's pilot category; load category strategy, supplier agreements and competitive intelligence into RAG.
Shadow mode
Agents recommend in shadow; category managers review; thresholds calibrated; cross-banner learning layer tested.
Pilot category live
Live in pilot category with category-manager approval on every action; metrics reviewed daily; weekly business review with CMO and Chief Merchandising Officer.
Pilot evaluation
Pilot outcomes reviewed; expansion plan agreed.
Multi-banner rollout
Banner-by-banner rollout with per-banner variations; competition-law review per region; autonomy thresholds calibrated by category.
How the deployment is governed.
Multi-banner retail at this scale carries competition-law, supplier-relationship and consumer-fairness governance — all of which the platform respects by design.
Competition-law boundaries
Cross-banner learning shares patterns, never raw competitive data; legal review confirmed compliance with competition-law constraints in every market.
Supplier-data discipline
Supplier-specific data shared only with the agreed party; supplier-confidentiality terms enforced at the platform level.
Consumer fairness
Price recommendations tested for fairness implications; vulnerable-customer indicators (geography, store profile) trigger review.
Audit-grade decision trail
Every recommendation, every category-manager decision, every action captured with rationale for internal audit and supplier-relationship transparency.
Category-manager authority
Category managers retain final authority on every assortment, price and promotion decision; agents recommend, humans decide.
Three transferable lessons.
Three lessons for other multi-banner retailers considering agentic merchandising.
Cross-banner learning is the win — design for it from day one
The pilot's biggest unexpected win was cross-banner pattern learning. Designing the cross-banner layer upfront made it possible; bolting it on later would have been hard.
Competition law is not a blocker — it's a design constraint
Treating competition-law boundaries as architectural constraints (rather than legal afterthoughts) made the deployment defensible without limiting its value.
Supplier preparation is an underestimated lever
The biggest individual margin improvements came from better-prepared category-manager / supplier conversations. Underinvesting in agent-prepared negotiation briefs is leaving real money on the table.
Could the same outcome work in your environment?
Tell us your sector. A senior xyner partner will walk you through a tailored plan.