MandM Direct manages models at scale

Dataiku announces its continued success with MandM Direct to support its data science practice. Working with Dataiku and a leading cloud platform provider, the partnership supports a wide range of use cases, from personalised marketing campaigns, to on-site experience and pricing recommendations.

MandM Direct is one of Europe's leading online discount retailers, with over three million active customers and seven dedicated local market websites across Europe. It is the second largest online fashion retailer in the United Kingdom and delivers to 25+ countries worldwide. The company scaled quickly in 2020, with accelerated growth and increased volume in customers and data magnifying the challenges it faced in scaling its machine learning environment in production.

MandM Direct’s rapid growth resulted in two big challenges. Firstly, its core data team sought to get all available data out of silos and into a unified, analytics-ready environment. Working across a hub and spoke model for its data centre of excellence, the team wanted to be able to leverage data to answer business questions in a way that didn’t always involve code. Secondly, the team wanted to scale its AI deployment in a traceable, transparent and collaborative manner so that they could prevent interruptions or failure of machine learning deployments.

MandM turned to Dataiku to help break the data silos and democratise data access across its teams. With Dataiku’s visual and collaborative interface for data pipelining, data preparation, model training, and MLOps, the company was able to easily scale out its models in production without failure or interruptions, with complete transparency.

Using Dataiku, the team is now able to scale its models in its production environment, with hundreds of live models doing everything from scoring customer propensity, to generating pricing models. The platform provides visibility into model performance metrics, clear separation of design and production environments, and many more MLOps capabilities.

Teams can now easily push-down and offload computations for both data preparation and machine learning. Using Dataiku means this capability is accessible to all user profiles across MandM, without knowing the underlying technologies or complexity.

“Having a platform like Dataiku allows our data scientists to focus on building cool things, not spending hours and hours on maintenance and making sure things are running,” said Ben Powis, Head of Data Science at MandM Direct. “With workflows deployed in Dataiku, we save literally days of work every month and we’re running ten times more models in production than we had been previously.”

The team has also built out a feature library ‘shop window for machine learning projects’ with Dataiku that contains more than 400 features specific to its business.

Florian Douetteau, Dataiku CEO said: “This is an exemplary opportunity for Dataiku to work with a dynamic, high-volume data science team whose work demanded the flexibility of AutoML and visual tools, while still maintaining the ability to work in code to produce models manually. It’s exciting to see how quickly the team has experienced the benefits of the solution: we’re seeing them add business value in a variety of areas, from adtech to customer lifetime value, via new dashboards, detailed analysis, and new machine learning projects deployed in production.”