Roberto Sorin - stock.adobe.com
Readers of a certain age will remember taping the top 20 on a Sunday night. And will also remember – if they had failed to catalogue what they were, possibly illegally, recording – not being able to find that particular New Order track among a stack of cassette tapes.
Or UB40 track. This is the particular example that Steve Pimblett, chief data officer (CDO) at The Very Group, tends to give when speaking at data management conferences about the company’s data strategy. Not so necessary for delegates to such conferences, but very useful for board members of a certain vintage when signing off on data catalogue technologies – in Very’s case, from Alation.
Very is formerly Littlewoods, of printed catalogue fame – a retailer with an annual turnover of almost $2.5bn delivering approximately 49 million products online across Europe every year.
Its roots go back to the football pools and mail order catalogues in the 1920s and 1930s in Liverpool and Manchester. The company opened its first bricks and mortar store in 1937, in Blackpool. Littlewoods merged with Shop Direct in 2005, and rebranded as The Very Group in 2020.
The company’s makeover from high street and mail order mainstay to online retailer left it with decades of data. However, the siloed nature of that data and limited automation were hindering it.
Pimblett has been driving a cultural change towards data, aiming at building a foundation of trusted data to enable Very’s workers to turn data assets into actions for its customers. He joined the company in 2020.
Families on a budget are the firm’s primary focus, he says, adding: “That came out of data, because what we were looking was doubling down on our core customers: Who are they? How do we better serve them? We came up with seven segments. And the top segment is families on a budget. Why? Because they really value our full, comprehensive offer. They can come to a single place and shop for everything. They value that we have got clearance and discounts as well. And we offer flexible ways to pay so they can spread the cost.”
Pimblett says he joined the company, in part, because of the “scale of customers – four and a half million of them”.
“And we’ve got multiple categories, 2,000 brands, over 200,000 SKUs at any one point in time, new SKUs coming on, and old SKUs coming off. The hottest thing at the minute is air fryers, in case you didn’t know. And we’ve got a financial services business, offering credit, like the old catalogue business did,” he says.
“We’ve also got a robotic warehouse, we’ve got every digital media imaginable. It’s a data paradise, to go after deeper insight and action across all the different business domains to help them achieve their business goals.”
Pimblett divides the data strategy at Very into five buckets. These are “usage”, including search technology and chatbots; “curation”, namely how data is categorised and stored for more efficient discovery; “risk”, which covers the lineage of data, velocity and how it is securely managed; “trust”, which includes search rankings and endorsements; and “value”, meaning how data is used to improve staff onboarding, time to insight and storage cost.
“We’ve got every digital media imaginable. It’s a data paradise, to go after deeper insight and action across all the different business domains to help them achieve their business goals”
Steve Pimblett, The Very Group
Prior to Very, Pimblett worked as CDO for Betsson.com, an online gambling company, and CDO for Wejo.com, which is focused on messaging from connected cars.
He is employing a “hub and spoke” model that he developed previously, with the data team at Very.
“I create centres of excellence. At Very, we’ve got a centre of excellence in data platforms and engineering, one in business intelligence, one in analytics, one in data science, and one in governance. And we partner with the business on whatever the spokes are.
“And then the way we distribute the capacities. So, the hub will either be allocating capacity to itself to build platform capability, or it will be allocating capacity to the spokes and the verticals across the business: retail, category planning, but also our financial services business.”
There are around 130 people in the data team that reports to him, and around 10 in the data governance centre. It is those people who mainly use the Alation data catalogue software.
Pimblett says the team also considered Collibra as a supplier for its data catalogue.
“We did an RFP [request for proposals] for five catalogues. We shortlisted two based on their feedback – Collibra and Alation. The preference was Alation, but the route was to run a POC [proof of concept] to make sure it did everything that they said it did in pre-sales. I’ve worked on the other side of the fence, selling software, so I know,” he says.
“We were worried about our complexity and scale. If we’ve got 250,000 tables in our Teradata estate, will Alation be able to scan it, and will the user interface be able to search on it?”
One area where it has found particular value is in bot scraping.
“Like many digital businesses, we struggle with bots. We’re a big brand, we’ve got thousands of products and millions of prices, so we tend to have a lot of bots scraping our catalogue to find out the prices, so they can use it in competitor analysis or whatever they’re doing. So, bots are a challenge,” says Pimblett.
“Historically, each area would manage bots differently in their data assets. We thought, ‘This is ridiculous, what we need is a central visitor table with bots removed, and also a community of people innovating around bot detection’. It has moved it from 10 silos reporting a different number to a single version of truth that’s trusted. And then everybody can innovate around improved bot detection on that single asset, building intelligence on top of it,” he adds.
“My job here really is to create value from the data assets, but to do that in a trusted way that makes sure we respect risk to, for example, personally identifiable information. Alation has been a big part of that, even though it can be seen as a governance tool. That example of bot management is a great one. It is a story of how cataloguing, indexes, collaboration and data stewardship can help move the dial on value creation, not just on risk, and simplification,” he says.