£7M
expected annual profitability improvement
Entire UK store estate reviewed with keep/fix/move/exit recommendations
Board-ready financial scenarios delivered in weeks
Generated an expected £7M annual profitability improvement for an international confectionery retailer through a data-driven property strategy covering the entire UK store estate. The work combined geo-demographic catchment analysis, cannibalisation modelling, and four-wall economics to produce store-by-store recommendations (keep, fix, move, or exit) with sensitivity-tested financial scenarios.
The Problem
An international confectionery retailer with a large UK store estate was making portfolio decisions on incomplete information. The headline metrics available to leadership (sales per square foot, upcoming lease events, basic footfall counts) told part of the story, but none answered the questions that actually mattered: which stores are genuinely profitable on a four-wall basis? Where are we cannibalising ourselves? Which catchments are underserved, and which are saturated?
The Solution
We built an end-to-end property strategy combining geo-demographic analysis, store performance modelling, and financial scenario planning. The work moved through three phases: understand the catchments, model the drivers of performance, and translate both into actionable, finance-ready scenarios for every store in the estate.
Every store was assigned a catchment based on drive-time isolines, profiled with resident and workplace population, household income, age structure, and lifestyle segmentation. Overlapping catchments were identified and quantified through a cannibalisation matrix using a gravity model weighted by store size, accessibility, and local competition. This was one of the most valuable outputs: several stores that looked healthy in isolation turned out to be splitting a catchment that could support only one location, while a handful flagged as weak were actually suppressed by a nearby sibling.
We clustered the estate into mission-based groups (impulse, destination, convenience, gifting, tourist) so that each store was benchmarked against peers with comparable economics, not judged against a flagship when it served a fundamentally different purpose. A sales driver model (regression supplemented by gradient-boosted trees) identified which local factors most influenced revenue: catchment affluence, competitor proximity, footfall quality, and store condition. Stores trading significantly below expectation were flagged for investigation.
Each store was then rebuilt on a true four-wall P&L and run through a financial scenario model covering four actions: keep, fix (refit), move (relocate), or exit (close). Every scenario included sensitivity testing on key assumptions, producing base, downside, and upside cases that gave leadership a realistic range rather than a single point estimate.
Results and Impact
| Metric | Value |
|---|---|
| Expected annual profitability improvement | £7,000,000 |
| Stores reviewed | Entire UK estate |
| Scenarios modelled per store | 4 (keep, fix, move, exit) with base, downside, and upside cases |
| Cannibalisation pairs identified | Multiple overlapping catchments quantified for the first time |
| Clusters defined | 5 mission-based store types with tailored benchmarks |
| Time to board-ready recommendation | Weeks, not months of circular discussion |
| Stakeholder sign-off | Property, operations, and finance aligned on a single model |
The £7m annual improvement reflected a mix of actions: exiting unprofitable locations, resolving cannibalisation by consolidating overlapping stores, redirecting refit capital to stores where the modelled uplift was strongest, and pursuing relocations in catchments with clear unmet demand.
Key Takeaways
-
Portfolio strategy becomes actionable when it names a specific action per store. A heatmap of "good" and "bad" stores is not a strategy; a costed action plan (keep, fix, move, exit) with stated assumptions is.
-
Cannibalisation is the hidden variable in most retail portfolios. Until you quantify overlap, you cannot tell whether a weak store is genuinely weak or simply starved by a neighbour. Several recommendations flipped once cannibalisation was made visible.
-
Sensitivity testing builds confidence, not doubt. Showing a range of outcomes (base, downside, upside) made the board more comfortable committing, not less. They could see that even the conservative case justified the recommended actions.