CDO interview: Barry Panayi, chief data and insight officer, John Lewis Partnership | Computer Weekly

Building data and leadership expertise

Panayi’s first job after graduation was with a marketing agency that was using data for direct mail, which he says is how he started to learn his trade. After that, he started moving through industries, companies and leadership roles. He spent a considerable time with EY, helping to establish the consultant’s first data insights practice.

“That was brilliant timing,” he says. “The phrase ‘big data’ had just taken off, so I could ride that wave and it was accelerating massively at that time in the mid-2000s.”

Panayi then became head of data and analytics for the Virgin Group, head of data science for Bupa and, prior to joining JLP, group chief data and analytics officer at Lloyds Banking Group. He recognises that this plethora of roles has helped sharpen his awareness.

“I’m either one of the most boring or interesting people you’ll ever meet, depending on what you’re into. I’m incredibly narrow – so there’s not much chat outside of data, analytics and insight”

Barry Panayi, John Lewis Partnership

“That’s the gamble I took,” he says. “I don’t see many data professionals bouncing around the banks or the retailers or healthcare, which is fine. But I really do enjoy seeing stuff that I learned in one place, picking the good bits and learning from all the things I do.”

Panayi says his role at EY helped to build confidence because it was a cross-industry position. “One day I’d be working on a pricing algorithm for maxi dresses, and the next day we were trying to reconcile energy trades. You can apply the same techniques, tools and learnings from one place to another,” he says.

“There is some industry knowledge, of course. But I’ve tended to find that whenever I go somewhere, there’s a tonne of people who know more about what the business does than me and I can learn it. So, am I a retail expert? Absolutely not. But there are 80,000 other people at the partnership who get it. I deliberately aim to be the voice from outside.”

Taking on a fresh challenge

Panayi says he was attracted by the immense cache of the JLP brands – John Lewis, Waitrose and John Lewis Financial Services – which he says offer a unique blend of non-food retail, grocery retail and financial services.

“It fills that gap where I like to apply my knowledge across different industries,” he says, before saying he was also attracted by the aims of the business. “I met the new chairman, Dame Sharon White, and I was absolutely sold on her vision – brand new chairman, brand new board, and trying to use data and put it at the heart of everything we do.”

Panayi says John Lewis is not hugely data- and technology-driven traditionally. He says the company is well known for its great customer service but now must think about how to carry on delighting its clients in a digital age. Luckily, he believes all these challenges can be potentially solved through his team’s canny exploitation of data.

On a day-to-day basis, Panayi says the work of his team is split into three main parts. First, data management, which covers governance and privacy, but also using and presenting information in a way that makes sense for customers. In simple terms, that work is about ensuring products are correctly tagged so that they can be found and bought.

The second key area is data science and business intelligence. Organisational challenges can range from setting pricing and promotions to designing staff rotas and onto finding the most efficient routes for lorry delivery drivers. He says data science can help John Lewis deal with these concerns.

Finding answers to important questions

The final element of Panayi’s remit covers research and insight, which includes quantitative and qualitative analysis. He says taking responsibility for this kind of research is pretty unusual for a data chief. The good news, however, is this aspect of the role really appealed to him.

“It gave me the chance to work with my director of research and insight again, who I worked with at Virgin. He’s come over now to work with me on this programme. We take what the customers say, which is the qualitative data, and then we compare that to what they have actually done and we look for patterns,” he says.

Panayi says the research normally covers specific business-led topics. One example is the launch of a new product or service, such as the recently launched Anyday range, which aims to provide affordable products at a high quality in a sustainable manner.

“So, we were asking, ‘What sort of products should be in there, what sort of prices do people want and do people associate our brands with them?’ Any new proposition would go through there. But it’s also about understanding how people are feeling about key issues, such as affordability, ethics and sustainability – how important are those things?” he says.

“Half the work comes from a particular place in the organisation, which has a question that they want to understand and get close to customers. Sometimes we bring customers into board meetings to talk to members of the executive committee. We invite them to have tea and cake. But sometimes, there are just themes that we spot that are coming and that we want to ask customers about.”

Creating capabilities and platforms

Panayi is just over 18 months into the role and says one of the things he’s most proud of is developing the capability internally to serve the business’s long-term data aims.

“The team didn’t exist as it does now when I joined,” he says. “The first thing was to create a team that would be a one-stop shop for all data insight and analytics requirements. That was hard because we had some leadership roles to fill.”

With capability honed, Panayi turned to systems and services. One of the things he noticed was that there were elements of the technology landscape that weren’t serving John Lewis effectively. Panayi says he is “very interested” in the tools the company uses.

“As soon as I joined, I had a very healthy feedback culture where I got my team to tell me exactly what was wrong with every platform. I used that feedback as my to-do list and prioritised their requirements. There was some software we were using that we weren’t getting the most out of, so I killed some platforms and software and brought in some other technology,” he says.

“I like our stack now. I didn’t like it a year and a half ago, but what we have now has enabled our data scientists to do amazing things and to manage and govern our data much more effectively. One of the key things we did was drop one database technology and bring Snowflake in.”

Snowflake now sits at the heart of a tightly controlled data ecosystem. As well as Google Cloud Platform, the retailer runs Tableau business intelligence on top of Snowflake. Panayi’s team also uses specialist data tools, such as dbt and Collibra, while most coding takes place using Python. His team is also beginning to explore machine learning tools.

“These changes made a big difference for my team. They all play nicely together. And there’s loads of features in the Snowflake tool that we’re not using yet, but that I’d like to in the future.”

Focusing on the right priorities

Panayi says his team’s priority projects for the next year cover two key areas. The first includes what he refers to as “the more traditional customer marketing-type projects”, such as segmenting shoppers and targeting them with personalised offers and working with suppliers to give them insights to get the right products in the right ranges.

The second priority area covers behind-the-scenes operational applications. Panayi is excited by developments here and believes tactical use of data could have a big impact. He says examples include thinking about where the partnership’s vans should be driving, how many people need to be working at certain times, and how much Waitrose should reduce the price of strawberries at the end of the day to make sure waste is reduced.

“All of those big optimisation problems are really operational issues. That’s a core challenge. And we couldn’t do any of that if the data wasn’t in the right place. The temptation a year and a half ago was to go after a lot of those. And although we did do a few, it was pretty manual and pretty painful,” he says.

“We’ve had to be strict on getting everything set up correctly, but now we’re in a position where we can start doing that really cool stuff and we’re seeing it work.”