A minimum viable range: cutting SKUs without losing revenue

A ten-week range review cut a private-equity-backed UK pet superstore chain's dog food assortment by 46%, from 835 in-scope stock-keeping units (SKUs) to around 454 core lines, while retaining 95% of revenue. QuantSpark's recommended range was projected to add £58k in weekly margin plus £146k in conserved or transferred revenue, a rare case of a retailer cutting nearly half its range without giving up the sales that range was assumed to protect.

The starting point was a range that had grown for its own sake. The chain's dog food assortment had ballooned to around 1,930 unique SKUs stocked across 52 weeks of the year, yet 80% of sales came from fewer than 300 of them. The long tail added real operational cost while giving customers only marginal incremental choice. Worse, many of the high-selling SKUs that justified the range's size carried below-average margin, and pricing and promotion decisions were not data-led, so the retailer was eroding margin on the products doing most of the work.

This is the trap that catches most range-rationalisation exercises: SKU counts are easy to cut, but nobody knows in advance which customers will simply buy a near-identical product from the same retailer and which will walk to a competitor. Cut blind and a retailer risks losing exactly the revenue it was trying to protect.

QuantSpark's answer was a fixed-fee, 10-week engagement built in three stages. First, exploratory data analysis established the baseline: SKU counts, sales concentration and margin by product, refined during the engagement to show that 95% of revenue in fact came from just 55% of SKUs, an even tighter concentration than the client had assumed. Second, a substitutability analysis built a customer choice model to estimate how demand would transfer if a given SKU were delisted rather than simply disappear. That model found loyalty customers switch to an alternative product around 48% of the time when their preferred choice is unavailable, with food type, product lifecycle stage and brand identified as the attributes that most influence which alternative they choose. Third, an optimisation algorithm used those substitution estimates to incrementally build a recommended range for small, medium and large store formats, optimising first for revenue and then refining for profit and business logic, producing the minimum viable range: the smallest assortment that preserves the revenue and choice customers actually value.

The method was designed to be repeatable and extensible, so the same three-stage workflow, baseline analysis, substitution modelling, optimisation, can be reapplied to other categories beyond dog food.

The commercial case rested on translating the optimised range back into pounds. Because the substitution model estimated whether a delisted product's revenue would be lost outright or transferred to a retained near-equivalent, QuantSpark could aggregate the retained and transferred revenue and the associated margin across the whole recommended range. That aggregation produced the projected £58k weekly margin uplift and £146k in conserved or transferred revenue, delivered outputs of the engagement itself, distinct from two earlier, indicative figures produced at proposal stage (a modelled £830k per year margin-uplift opportunity) and in internal collateral (a cited £3m per year gross-profit opportunity), which were exploratory estimates rather than outcomes of the delivered analysis and are not carried forward as facts here.

For a retailer, the headline result is not simply that the range got smaller. It is that a 46% cut in core SKUs was achieved while holding onto 95% of revenue, evidence that most of a bloated range's size was complexity rather than genuine customer value, and that with the right substitution model it is possible to find and remove that complexity with precision rather than guesswork.