Process Mining Camp 2014 — Fireside Chat with John Müller
Process Mining Camp Tickets are sold out this year but why don’t you sign up for the Process Mining Camp email list? You will receive the links to the slides from the speakers and to the video recordings of the talks as soon as they become available. And you will be the first to know when the registration opens for next year.
As a warm-up for camp, we have asked some of the speakers for an up-front interview. Previously, we have already spoken with Frank van Geffen from the Rabobank, with Johan Lammers from the Centraal Bureau voor de Statistiek, with Shaun Moran from Customer Dimension Analytics, Antonio Valle from G2, and Nicholas Hartman from CKM Advisors.
Today, you can read the interview with John Müller.
John is a data scientist at ING bank and will give a practice talk about how he helped to improve the customer experience at ING Australia with process mining.
Interview with John
Anne: Hi John, you came across a use case for process mining that will be very interesting for many people at camp: The analysis of the customer journey path on a website before the customer calls the help desk. Can you tell us a bit about that moment when you realized that process mining was a solution for the problem you were facing and why?
John: First of all thank you for the invite. I’m honored to be invited. I came across the idea to use process mining already some months into my analysis. I was struggling with giving back the right kind of visualization of my analysis to the business user, I had tried the more traditional ways of plotting some things in R or some graphs in Excel but none of those options came close to my dream of giving something back to the business user where I could empower him to find his own answers to his questions.
We discussed how best to approach this when it hit me that this entire analysis could be seen as a customer journey or process if you will. Just because a website has no specific order in which people have to click didn’t mean it wasn’t fit to use process mining.
There was a clear start; the login, a clear middle, the switch to the call center and a clear end; hopefully a satisfied customer hanging up. The Disco tool by Fluxicon gave me the chance to give the preprocessed logs back to the business user and let him explore his own process and find his own answers, thus using his domain knowledge to the fullest.
Anne: That is so good to hear, because this is exactly what we are trying to do: Making process mining accessible for the people who have the domain knowledge about the process! Do you have an example of where they could find something in their process that you had completely missed when you looked at the same data yourself before?
John: I would actually put it the other way around, being able to visualize the data in such a powerful way opened up my eyes to how many things I had not seen when exploring the data by myself.
I had noticed a few recurring patterns of parts of the website that seemed to be causes of calls, but there would have been simply no way of finding all the relevant parts without a significant time investment in getting to know the domain.
To put it differently, the use of such a powerful tool for exploration allowed me and the business user to explore the data without having to think of specific questions to ask up front. No longer was the question something like “How many customers called about account opening that also visited the how to open an account page”, but it had turned into “It looks like a large percentage of clients had visited the interest page before calling, let’s do a deep dive there”.
Anne: Right! So it changed the way questions were asked? Because they did not need to know all the questions upfront?
John: Exactly! We could now go in with a blank slate, letting the data show us where to look next. We could start off by just looking at all the different ways the website was used before any sort of call and then base our next step off of the interesting-looking patterns.
Obviously not everything that we could see at first glance was shocking and new. For example, quite unsurprisingly a lot of customers logged in and clicked on the contact us page right before calling, but by combining the domain knowledge of the business user with process mining we were able to quickly spot unexpected patterns as well as confirm his suspicions about failing parts of the website. The latter part might not sound very exciting but this was the first time that these kind of suspicions were backed up by numbers at short notice, providing them with solid facts to convince other parts of the bank to act.
Anne: Fantastic, it will be great to hear more about this in your talk. Thanks a lot for making the time for this chat and see you at camp on Wednesday!
Process Mining Camp takes place on 18 June in Eindhoven! Tickets are sold out for this year’s camp but why don’t you sign up for the camp email list? You will receive the links to the video recordings of the talks, and you will be the first to know when the registration opens for next year…