Optimising Retail Store Layouts with AI

The Challenge

Our client provides a retail space planning platform that enables retailers to design store layouts and analyse category performance. Their customers - major grocery chains - could see exactly how their current layouts were performing, but increasingly wanted to know what changes would improve profitability.

The fundamental question retailers face is deceptively simple: which product categories should go where, and how much space should each get? Everyone knows certain products sell better together, but quantifying these relationships and determining optimal space allocation remains complex.

The platform needed to develop a solution that could analyse existing store variations to identify improvement opportunities, without requiring years of historical change data.

The Solution

We partnered with the client to develop a proof-of-concept optimisation tool through intensive collaboration with their retail planning experts.

The solution developed machine learning models to address two core challenges:

Understanding Category Relationships

The system quantifies the sales impact when specific categories are placed next to each other. It can determine what portion of a category’s sales comes from complementary placement versus its baseline performance.

Optimising Space Allocation

The system analyses the critical question of space trade-offs between categories. By comparing stores with different space allocations, the model identifies where reallocating space between categories would improve overall profitability.

Solving the Aisle Structure Challenge

A critical data challenge emerged: while the system knew which categories were adjacent, there was no data showing which categories belonged to the same physical aisle. We developed an algorithm that infers the physical aisle structure by analysing adjacency patterns. This was essential - optimisation must happen at the aisle level where retailers actually make changes.

The optimisation engine finds optimal arrangements while respecting real-world constraints: maintaining total aisle lengths, keeping complementary categories together, and ensuring minimum space requirements. Rather than suggesting impractical wholesale changes, it identifies specific, implementable adjustments.

Results and Benefits

The proof-of-concept successfully demonstrated the value of AI-powered layout optimisation across multiple major retail partners. The system can process hundreds of stores rapidly, generating granular recommendations for every aisle.

The Scale of Opportunity

Analysis across a grocery chain revealed:

• The majority of stores had profitable improvements identified
• Meaningful profit improvements achievable across entire estates
• All achieved without adding products, changing prices, or major renovations

Practical Implementation

The system’s recommendations were remarkably practical:

• Minimal bay changes required per store to achieve results
• Limited aisle modifications needed on average
• Surgical repositioning of categories rather than wholesale reorganisation

This demonstrates the AI’s ability to identify high-impact adjustments that store teams can implement during regular resets rather than requiring special projects.

Actionable Insights

Through an intuitive web interface, store planners can:

• View current versus recommended aisle layouts visually
• See predicted improvements for evaluation
• Understand which specific categories should be adjusted
• Assess implementation complexity before committing to changes

Technical Validation

The adjacency relationship model was rigorously validated against actual store performance data, demonstrating strong predictive accuracy. The system shows consistent precision in its recommendations, giving retailers confidence that identified category relationships will translate to real improvements.

The model intelligently identifies which stores would benefit from changes and which are already well-optimised, avoiding unnecessary disruption.

Looking Forward

The client is now validating the proof-of-concept with retail partners, tracking real-world implementation results to refine the models further.

For retailers operating on thin margins, even small percentage improvements represent significant value. This tool demonstrates that AI can unlock that value by turning the complex interplay of space and adjacency into clear, actionable recommendations.