AI Strategy and Roadmap — Retail Group, Manchester, UK
The Situation
A UK retail group operating 180 stores across England and Scotland, with a growing e-commerce division accounting for 31% of revenue, had spent most of 2023 fielding AI vendor proposals. Eleven of them in 8 months. Each one arrived with a deck, a demo, and a number that looked impressive until you tried to connect it to an actual business problem the group had.
Leadership was not dismissive of AI. The opposite. The CEO and CFO both believed the technology was going to matter and were anxious about moving too slowly. The problem was they had no framework for deciding where to start, no way to evaluate vendor claims independently, and no internal team with enough AI experience to fill that gap.
Two vendor proposals had already been approved in principle before Amorisoft was engaged. Both were paused when the internal project sponsors could not answer basic questions from the board about data readiness, integration complexity, or how success would be measured.
What Amorisoft Did
Amorisoft came in as an independent advisor with no vendor relationships and nothing to implement. The engagement had one output: a strategy the board could act on.
The first 4 weeks were spent on discovery. Amorisoft conducted 34 structured interviews across store operations, e-commerce, supply chain, finance, and HR. The goal was not to find where AI could be applied. It was to find where the business had measurable problems, what data existed around those problems, and what the organisation had the capacity to actually deliver given its current technology infrastructure and internal skills.
From 34 interviews, 19 candidate AI use cases were identified. Each was assessed across 4 dimensions: business value, data readiness, implementation complexity, and organisational change requirement. The assessment was done without reference to any vendor's product.
19 candidates were narrowed to 7 for detailed business case development. 7 were narrowed to 4 for year-one recommendation based on a combination of projected value, speed to value, and the group's realistic delivery capacity.
The 4 approved initiatives were demand forecasting for the top 200 SKUs, AI-assisted markdown pricing for end-of-season stock, a customer churn prediction model for the loyalty programme, and an internal HR attrition risk model. Each came with a defined data requirement, a build-versus-buy recommendation, a vendor shortlist where applicable, and a success metric agreed with the relevant business owner before the board presentation.
The 3-year roadmap covered the remaining 3 use cases from the shortlist, sequenced by dependency and organisational readiness rather than by size of projected return.
Results
The board approved all 4 year-one initiatives at the presentation in week 14. Combined projected saving across the 4 initiatives was £2.1 million, with demand forecasting accounting for £900,000 of that figure. The 2 previously approved vendor proposals were formally stood down. The 3-year roadmap was adopted as the group's official AI investment framework. Implementation of the first initiative, demand forecasting, began 3 weeks after board approval using a vendor from Amorisoft's recommended shortlist.