Julian Pimm-Smith has only been the head business intelligence (BI) at Pret a Manger for less than a year, but the first thing he did was “hive off” business intelligence from the rest of the IT environment.
Speaking at SAP’s Innovation Forum in London, Pimm-Smith explained his strategy, saying: “As soon as I got to Pret we purchased our own set of servers, and it’s all physical - we got everything off virtual, which is incredibly important to us in terms of the power we needed.”
This runs counter to many IT vendors’ party line that business critical processes should be moving to the cloud. Pimm-Smith offers an alternative strategy: “So anyone out there who’s not looking at cloud, I would recommend if you’re ever looking at refreshing your BI set up, to hive it off and operate it outside of your IT infrastructure.”
Pret don’t use SAP for enterprise resource planning (ERP) but Pimm-Smith’s team uses SAP’s BusinessObjects enterprise [version 4.1] and runs a SQL server underneath as a data warehouse.
Pimm-Smith says that running on-premises allows for the speed he requires but also, with the BusinessObjects mobile app, this information is available across the business in real time. “So if any of your IT department are telling you that you are not allowed real time data then you can happily show them this kind of example,” he says.
Boring reporting
Despite calling it “boring reporting” Pimm-Smith recognises that reporting is the foundation of what his team does. He says: “It is what differentiates a successful business from a not so successful business. Making sure that the staff who do the work, understand how they are performing on a daily basis and how that rolls up.”
Pimm-Smith has moved to a self-service model, where Pret shop managers can monitor their key performance indicators (KPIs) at the end of every working day. He explains: “As soon as they close their day they’re getting a full breakdown of their financial performance.” So instead of a prescriptive model where actions are dictated, shop managers have the information to make decisions relating to stock and labour for the next day immediately.
To serve this, Pret uses SAP’s BusinessObjects to deliver mobile reports. Pimm-Smith wanted to avoid serving complex dashboards to staff, preferring to deliver “just the information they need, when they need it.”
This information is primarily live sales performance for each store, including historic and mapped information. “It’s an incredibly simple report,” says Pimm-Smith, “but it was so popular that we had to go back to SAP and buy more licences for BusinessObjects because every single operations manager is on this all the time, from 5am through to close of their shop, through to the weekends.”
Delivering data
The business intelligence team at Pret is split between technical staff who run the servers, applications and the data warehouse, and the reporting staff.
The reporting team doesn’t act as a BI service provider though: “We’re not waiting for someone to fill out a report request. They know what they want in terms of business performance. We as BI professionals know how to build that report, know how to do performance management, we know how to visualise it, so it’s a dialogue rather than a blind service.”
Analytics
Pret has 350 shops in four regions (UK, Europe, USA and China) with a pipeline that will see more than 400 shops in the portfolio by the end of this year. Each store would traditionally have an existing manager come over to help with the launch. However, with an increasingly international footprint Pimm-Smith sees the need for analytics to help share knowledge remotely. “This is relevant for our analytics in terms of how to manage the quality around those ever expanding Prets around the world,” says Pimm-Smith.
By serving staff historic information on how factors like weather and special events affect buying patterns, managers can make smarter decisions when it comes to stock and labour, decisions that have a direct impact on profits. For example, the average Pret store makes around 60 percent of its daily sandwich and salad output in the morning and then makes stock decisions on the fly. Good, accurate analytics are critical to this way of working.
Predicting successful stores
Pret doesn’t do all of in BI in-house though. Pimm-Smith looked to bring a consultancy in for a data science project that would help the company decide on new shop locations. As Pimm-Smith put it: “There is no low-hanging fruit left for Pret locations in London. In America we are eating that up. It is harder to find a Pret that is just going to be successful.”
By using a consultancy Pimm-Smith is able to blend Pret’s internal data with more augmented data which is harder to come by, such as store visibility, competition data and cannibalisation impact, census, and mobile phone footfall maps. “We look at 100 data points, they’re looking at over 10,000,” says Pimm-Smith. The metric he wanted was a weekly average sales revenue for new locations so that Pret can make a buying decision that would be more accurate than his team could have provided.
Pret a Manger is no longer a small business, and with a growing international footprint the way it delivers real time data to its store managers is key to keeping not just Londoners, but busy people all over the world well fed.