← Back to blog
Data strategy

The problem with the best data in retail

EPOS data is the most trusted evidence in the industry, and it deserves to be. It is accurate, granular, and universally understood. When a buyer and a supplier sit down to argue about a category, EPOS is the common language they argue in.

The problem is not the data. The problem is what it can and cannot describe. EPOS data shows a singular historical scenario. It tells you, in precise detail, what happened under one set of conditions: one range, one price architecture, one promotional calendar, one shelf. It says nothing about the ranges you did not run, the prices you did not test, or the shoppers who walked past without buying.

Yet most of the commercial calendar is built on it. Ranges are built on last year’s data. Commercial calls run on instinct where the data goes quiet. Millions are wasted on expensive mistakes that looked sensible in the review deck, because the review deck could only describe the past.

This is the gap predictive retail intelligence exists to close. Rather than describing the one scenario that already happened, Breakthrough’s predictive infrastructure models the many that have not: how a range performs if you extend it, delist from it, or reprice it, and how shoppers respond when the category around them changes. The system draws on a product graph spanning 4M+ products and 56,000+ attributes, and grounds its modelling in 28 behavioural science principles, so the futures it simulates reflect how shoppers actually decide rather than how a spreadsheet extrapolates.

None of this replaces EPOS data. Historical performance remains the record of what the category did. Predictive infrastructure answers the question that record cannot: what the category becomes next, not what it did last year.

Explore how the data works.

See the seven dimensions →