John Lewis is weaving operational intelligence into the fabric of the business – not just IT

As I've said before, Splunk is an interesting company to deal with – mostly because they are willing to put their top customers up for interview, which is both

surprisingly rare and always a good indication to me that a company is doing something right. For those of you that aren't familiar, Splunk operates in the area of machine data analysis and operational intelligence. To be honest when I first came across the topic I couldn't think of anything more dry, but I've come to realise that enterprise machine data is actually one of the few 'big data' topic areas that I've written about where you see real examples of how analytics can get you a clear return on investment. And that seems to be the case for John Lewis – one of the UK's largest retailers – which has expanded its use of the tool from a proof of concept to pushing it out across the business.

John Lewis began using Splunk for searching and investigating problems with a new £1bn e-commerce platform, which launched last year. Aleem Cummins, John Lewis' Splunk lead, told me that the tool allowed the retailer to drill down and investigate logs over longer periods of time than had previously been possible, something that he said was a “marvel”, and which also allowed the company to be proactive with technical problems, rather than reactive. Cummins said that Splunk has paid for itself thanks to the number of sales that went through the John Lewis website, which would have fallen off if the tool hadn't picked up on some issues. He said:

“We have had some tremendous gains that have given us the return on investment. When you are talking about a £1bn website, even the slightest incidents or issues can have a high value. We have been able to recoup everything we have spent on Splunk. 

“For example, we had a very slight issue with our customer checkout. There was a small section of people that when they used a certain build, or a certain browser, they were having difficulties. In fact they couldn't check out. Because it's a big website the value of that was something that we had to investigate. 

“Traditionally what you would do is meander through your logs, do some scripting. But we were able to use the tool to really figure out some of the behavioural intelligence. We were able to see that there was a dip in sessions and from there we were able to analyse what was in common. As we were going through that we could see that those people were using a certain browser, and they were using a certain build – it's not about finding a needle in a haystack, it's about finding a needle in a bunch of needles.”

Cummins said that by stopping this “leak of conversion”, John Lewis was not only able to keep sales going through, which is obviously good for the business, but it also improve the customer experience for users. Instead of experiencing problems, having to call the John Lewis helpline, or god forbid have to actually go into a store to purchase, the Splunk team was able to proactively pick up there was a problem occurring and get on the case.

You've got to impact the bottom line

Splunk got broader appeal, outside of IT, because of its ability to analysis the checkout funnel on the e-commerce platform. After picking up the initial problems on the new John Lewis website, Splunk is now consistently used for proactive forensic analysis of the online sales funnel and follows the permeations of the customer along the purchasing journey, digging down deep into any situations where orders aren't being taken, to figure out if there are any problems for the customer.  Cummins said that he initially struggled to get funding to invest in Splunk, but by using the tool for this, it allowed the business to quickly understand the importance of operational intelligence. Basically, it's a lot easier to go to the board and say, this tool saved us X number of sales conversions that otherwise would have dropped away – let's see what else it can do. Cummins said:

“We were then able to show some real value to people because of this. Sometimes I think you can show technical value, but when you can show that it impacted on the bottom line, then it has the broader appeal as well. That raised the profile of operational intelligence, which gave us the traction for us to bring it to the next level.

“Bringing people with you is key, so what we tried to do was perceptual positioning. We put ourselves in the shoes of the business, try and think what they want. If we do that we can act in certain ways to ensure that they get what they want. Also need to make sure that the quick wins are spread, that helps engage and bring people with you.”

This 'next-level' that Cummins talks about involves putting together a hard strategy for operational intelligence, which will include how the business can be better educated on the issue, how it can be adopted and how new teams can be on-boarded. He also said that although he couldn't go into detail, it is likely that Splunk will be used to improve John Lewis' security. For an idea of how this might be rolled out, you can read a case study here that I did with Barclays on how they are using the tool. 

Finally, Cummins said that Splunk and operational intelligence shouldn't be a capability that sits just within IT and serves the business from the inside out. It should be its own team and platform that sits in-between all functions. He said that if you just get Splunk out into the organisation and start getting some quick wins, the business will start to realise its importance.

“Once you get the hook and start to expose things you get people dreaming and envisioning, then it has its own life and it begins to fly around. Don't spend too long trying to strategise to the nth degree, to justify to the nth degree, because you won't be able to do it. Splunk is a bit like the fabric that sits over the organisation, it's not just for certain use cases. 

“What is needed in companies like John Lewis is a tectonic team, a team that floats above or floats below all the other teams and shapes that landscape. I think sometimes if you put an operational intelligence team sitting inside IT operations team, the scope can be a bit tricky. I think these things need to sit on their own platform, if you will.”