Sainsbury's rebuilds data capability to improve automation and decision-making

Phil Jordan, the retailer’s Group CIO says that the use cases for AI and ML are ‘endless’

Sainsbury's is rebuilding its data ecosystem in a push for better decision-making, and to better exploit Artificial Intelligence (AI) and Machine Learning (ML).

Phil Jordan, the retailer's Group CIO, recently told Computing that data at the organisation's was previously very siloed.

"We've been on a journey for four years to rebuild our data ecosystem," Jordan began. "We realised a long time ago that our data was very siloed and stuck in legacy areas or under individual brands. So we made big punt to use Snowflake as our enterprise cloud warehouse running on AWS, together with a heavy use of MicroStrategy.

"We're now a long way into that journey, moving and closing legacy warehouses, rebuilding in the cloud, as part of a programme we call ‘Aspire'. One of the goals is the creation of a single version of truth. So all of our reporting is now revisualised and comes through MicroStrategy."

The organisation is also retiring legacy systems and making a bigger push into the cloud.

Jordan added that whilst the journey can never be said to be ‘complete', much of the work has been done.

"Lots of the heavy lifting is complete. We're now exploiting the value of having a cloud-based enterprise analytics system. So we can now re-engineer our decisions, with many of them now being taken as automated machine-based decisions."

One example of this is the substitution of items for online orders, which is now done by algorithm.

"If you think about substituting items for online orders where the selected item isn't available, previously the picker would just choose what they thought was relevant. Now we have algorithmic suggestions based on history and customer reaction. That's so successful it's actually driving sales."

Another example is replenishment, where staff restock shelves as customers empty them.

"When we do replenishment in stores, we have algorithms to tell colleagues the most efficient route to walk the store. And if we can get AI and ML to play in those spaces the use cases are endless.

"For instance we helped our store managers during the pandemic by running algorithms over our data to predict staff absences by area. That really helped managers with resource scheduling.

"We were also the first supermarket to use our data to infer whether someone is elderly, disabled or vulnerable. That allowed us to protect their delivery slots during the pandemic, so we didn't allow all the slots to be consumed and leave them without options. That work was hugely valuable."

In Jordan's view, the retail industry is perfect for automated decision-making, and he intends to make full use of its benefits.

"We will have the opportunity to automate every decision we make on a daily basis. That's because it's an intensely transactional business, so it's perfect for analytics and re-engineered decisions. All of our pricing, the way we look at other retailers, the way we do picking, in all these areas if we can bring much richer data set to the decision-making process we will create massive value. That will work both for driving the business' top line and taking cost out. It's a very highly repetitive, highly intensive business, so it's a strong use case."

However, that doesn't mean that automation is finding its way into every area of the business. Jordan explained that recruitment and personalisation are still very much left to humans.

"We're very conscious of personalisation and of course of privacy and ethics, so anything needing ethical judgement we wouldn't leave to a machine. It's the same when it comes to people; there are lots of things we can do to help with great people management, and that's still a human activity. What we're trying to do is to put great tools in the hands of great people.

"We also largely keep automation out of recruitment. We're still very much reliant, even when moving quickly as we are now, on the judgement of people. So we don't tend to use AI in that regard."

Jordan earlier said that IT recruitment is currently harder than ever.