AI ASSORTMENT OPTIMIZATION & RANGE PLANNING

Assortment.AI
Simulate Strategic Assortment, Space and Merchandising Changes Before You Commit

When shelf strategies get built on best guesses or blanket rules, you lose sales, waste space, or miss what local stores actually need. or blanket rules, you lose sales, waste space, or miss what local stores actually need. Assortment.AI brings product mix and shelf limitations into one model that uses AI to simulate store scenarios and give you the best approach.

WHY TEAMS RELY ON ASSORTMENT.AI

Test every shelf strategy with AI before you roll it out.

→ Simulates assortment and merchandising changes with performance forecasts before they hit the shelf

→ AI forecasting of revenue, volume, and holding capacity based on store-specific realities — not banner-wide averages

→ Compares what-if scenarios across stores or regions without rebuilding everything from scratch

→ Understands space tradeoffs before they cause cuts, delays, or stockouts

→ Aligns headquarters and field teams around shelf-ready strategies that move business

Product.AI Demand.AI Assortment.AI Space.AI Execution.AI

01

Bring in your data

SKUs, movements, planograms, store details — everything your team already works with.

02

AI model trains on your reality

Bespoke AI model trained on your actual movement data, decision trees, and demand transfer rates.

03

Simulate scenarios

Simulate how changes in assortment, merchandising, or space would perform — by store or region.

04

Export the winning strategy

Push the chosen strategy to Space.AI for planogram generation, or directly to execution systems.

Continuous Learning Loop

How Assortment.AI works

Built-in AI strategy engine
Run realistic simulations across assortment, space, and merchandising decisions automatically.

Flexible scenario modelling
Compare strategies using your own rules, limits, and goals across any number of store clusters.

Execution-ready outputs
Transfer chosen strategies directly into Space.AI or sales team workflows.

Consumer decision trees

AI consumer decision trees built from real transaction data capability data — how shoppers actually choose.

Demand transfer accuracy

Shows actual transfer rates (typically 40–70% lower than assumed) — protecting revenue standard models write off.

Store-cluster precision
Creates store-specific assortments in minutes — not banner-wide rules that miss local shopper behaviour.

merchandising

Category Management Teams
Simulate range changes before execution. Protect revenue with accurate demand transfer models.

Marketing

Space Planning Specialists
Understand space tradeoffs before they cause cuts, delays, or stockouts.

Retail

Commercial Directors
Align HQ and field teams around shelf-ready strategies backed by AI simulation.

Platform context

Assortment.AI was acquired by Vision Group in 2024. Recognised by CB Insights as one of the world's top 100 Retail Tech companies and named to Fast Company's World's Most Innovative Companies list. Assortment.AI is now fully integrated as Layer 04 of the Vision Retail Intelligence Stack — receiving demand signals from Demand.AI and outputting assortment decisions to Space.AI.

Assortment.AI's AI engine originated inside the Coca-Cola Founders Program — with a specific challenge: how do you improve vending machine sales while reducing labor? Hospital vending machines were uniformly stocked, leading to frequent stock-outs and ballooning costs. The solution was to use actual sales data, not traditional retail research, to predict what each machine should carry.

The algorithms were developed in collaboration with CSIRO's Data61 — Australia's national science agency. They could predict sales for each item in every machine and recommend the ideal space allocation, enabling tailored restocking strategies and revenue optimisation.

That approach became the foundation of Assortment.AI — machine learning models trained on store-level data, similar to how Netflix recommends content: not what the average user watches, but what each individual user is most likely to choose. Applied to retail: not what the average store sells, but what each specific store's shoppers actually want.

The Netflix analogy
Netflix does not show everyone the same film. Assortment.AI does not recommend the same assortment to every store. Store-level machine learning reveals what each location's shoppers actually choose — and builds the optimal range from that signal. Data Has A Better Idea™
Shoppers vote with their wallets in stores every day. Assortment.AI makes this visible — surfacing the store-level truth behind every assortment decision, so retailers and CPG brands can act on what is actually happening rather than what they assume is happening. Hyper-local by design
110+ retailer-category combinations. The right products in the right stores — regardless of how many clusters a retail chain operates. Not banner-wide rules. Store-level intelligence.

01

Discovery

Analyse business objectives, challenges, and available data. Identify the crucial objectives that will lead to success. Understand your store estate, category dynamics, and current assortment baseline.

02

Pilot

Create and train bespoke AI models on your actual data. Run live experiments to confirm model value. Build a solid business case before committing to full rollout.

03

Deploy

Integrate Assortment.AI into everyday business operations. Full deployment with continuous support. Quarterly reviews and model adjustments to ensure alignment with evolving business goals.

ASSORTMENT PLANNING OPTIMIZATION · BUILT INTO ASSORTMENT.AI

Store by store. Not banner by banner.

Traditional assortment planning applies one strategy uniformly across hundreds of different stores. Assortment.AI uses real consumer decision trees and store-level demand data to optimise the range for each store cluster — recognising that the right assortment in a city-centre convenience store is different from the right assortment in a suburban supermarket.

New product forecasting
Predict how a new SKU will perform at store level before launch — which stores should range it, in what quantities, before a single unit is ordered.

Range rationalisation
Identify underperforming SKUs where demand transfers cleanly to alternatives — so deletions improve category efficiency without losing consumer demand.

Space-aware optimisation
Assortment decisions connect directly to Space.AI planograms — so the optimised range reflects real shelf constraints, not theoretical ideals.

Demand transfer modelling
Model where demand goes when a SKU is added or removed — in-brand transfer, competitor switch, or category exit — using real observed substitution behaviour.

Store-cluster strategies
Group stores by shopper behaviour, format, and demographics — then optimise the assortment for each cluster rather than applying a single banner-level decision.

CMA negotiation support
Build the commercial case for shelf space with assortment simulation data — showing the revenue impact of ranging decisions on the retailer's category, not just your brand.