From scattered projects to a £76m question pipeline

The retailer had no shortage of talent, ambition or data. What it had was four problems that compounded each other. Data science capability was fragmented across the business. There was no consistent pipeline of business questions, so demand arrived through relationships rather than a process. Multiple data platforms were emerging in parallel. And there was no reliable way to connect data investment to business value, so every funding conversation started from scratch.

The ambition was clear: turn data and analytics into a repeatable engine for business impact. The difficulty was that nothing in the existing setup was repeatable.

We made the business question the unit of work. Not the project, not the use case, not the initiative. The question.

A question has an owner, a decision attached to it, and a number that moves if you answer it. It can be ranked against other questions. It can be right-sized: some deserve a product team and a year, most deserve a day. And a hundred questions with owners and values attached is a pipeline you can plan against, not a queue you're failing to clear.

Every question was mapped to one or more of the company's ten strategic jobs, so the line from corporate direction to an individual data scientist's sprint was visible end to end.

A defined analytics strategy and an operating model built loose enough to allow freedom but defined enough to drive consistency and re-use. Platform direction settled on a single cloud, with common data models. A live, transparent pipeline of business questions with owners, value and status, sitting above the delivery tooling so prioritisation happened at the question level. And a four-stage delivery model, Explore, Create, Realise, Enable, that could answer a question in a day or build a product over a year without treating them the same way.

Around 80% of questions were designed to be solved in Explore and stop there, as fast, high-quality work that never became a project. Work only moved to the next stage where everyone agreed the value justified the investment.

We ran every workstream in parallel: vision and strategy, contracting, tooling, recruitment and ways of working all started in week one. Design sprints to establish the pipeline began before there was a full team to work it, so the team's first day was spent on real questions. The value measurement approach was agreed before there were any results to measure.

Hub operational in week ten. Discovery and exploration sprints running from week ten. Roadmap and measurement framework in place by week twelve.

Year one: 28 business questions run through the hub, delivering £76m+ of realised value. The pipeline grew to 100+ questions and kept growing. Products reached 11,000+ users across 30+ countries in under a year, with value tracking extending to €1bn of value supported across the wider programme.