Context
An in-store expert, on every product page.
Currys is a leading omnichannel retailer of technology products and services, with brands including Currys PC World and Carphone Warehouse, and the market leader in the UK & Ireland.
In its stores, sales staff suggest the things that go with a purchase: a case for a laptop, an HDMI cable for a large-screen TV. Online, bundles were set up by hand on commercial grounds, and their uptake lagged the stores. As customers moved online, add-on sales, a key driver of profitability, were at stake.
The eCommerce team wanted the same journey online, with bundles personalized to each shopper and served in real time to drive incremental sales and margin.
The constraint
Manual bundles, and data that wasn't ready.
- Bundles were built by handThe merchandising team set up a bundle for every product and every add-on, with no personalization.
- Models stuck in the labThe data science team had models, but no way to put them into production on the website.
- Clickstream data was hard to useAdobe Analytics data was not in a shape for analysis at the level of one customer, and there was no large data engineering team.
- Traffic peaks were extremeVolumes rise as much as 10× on days like Black Friday and Boxing Day, so the models had to scale with demand.
- Data had to stay in placeEverything ran inside Currys' own Google Cloud, with no third-party copy of the customer data.

