Process Mining at IKEA Retail (Ingka Group): Insights and Opportunities with Tim Hills, CoE Lead for Process Mining at Ingka Group

Welcome, welcome, welcome back to the Mining Your Business Podcast to show all about process mining, data science and advanced analytics, Tim Hills, Ikea, Meatballs, Billy Bookcase and Process Mining. Y'all group, please assemble this podcast into yourself. Um, sorry, I don't have a manual. Ah, well, quick disclaimer before we get into it, the numbers in this episode are based on understanding at the time, but as it's still exploratory work at Inca, they're illustrative. All right, with that out of the way, let's get into the episode. Hey, Patrick, you always asked me in the intro about how am I doing? And you know, let me return the favor and ask you how are you doing today? Oh, this is weird. This is a new one. I'm doing well. Thank you, Yaku. I also have another question for you, and when was the last time you had something ordered from Ikea? Oh, two and a half years ago, a big closet. And how was your experience? It was actually, it was lovely actually, it was great, arrived on time, all the parts couldn't complain. Well, I have a feeling that our today's guest will have something to do with it, but before I get into the introduction of our guests, I also say that I had experience with the delivery just today, because I moved recently and I was thinking how to improve the sound acoustics for recording the podcast. So I ordered the carpet, headed delivered everything on time perfect, but then there was this little customer experience feature at the end when the driver was like, who has a small carpet like that ordered to his house for this type of money? And I was like, okay, interesting insights. I'm still happy to have this carpet here, but it is what it is. And today, we actually bring a very, very interesting guest, Tim Hill's, Tim Hill's welcome to our podcast. It's pleasure to have you. Many thanks. Thank you for having me. Starting with this type of question, do you think a customer experience can be impacted by a well-executed process? Absolutely. And yeah, well, I really hope so. That's one of the areas that we're really working with, with deploying process mining within Inka, because yeah, if we actually look at customer placing their order and receiving it in us meeting that customer expectations of being on time in full, absolutely, it's and we also do see that. One of the things that we'll discuss a bit later on is that the customer experience and the feedback that we get when we do things as we should do, or indeed the impact where things don't quite happen the way we wish them to be. Absolutely. Now, Tim, you mentioned a name, Inka. I also said something about Ikea. We had this little discussion before the recording. I would be very happy to hear how do these companies actually, how does it actually unfold? What does it mean? What is Inka with Ikea? Sure. So, Ikea, when you come to our stores, the big blue boxes with the yellow writing on the side, following the Swedish national colors, that's the concept, that's what the customers then see. We then have the owners of the franchise and the people that are responsible for setting the products and the concepts of our stores and so on, which is Ikea's systems. I'm then representing the primary franchisee of that group, Inka, and then within that I, because Inka also has three main sections, and I then work in the retail domain, so working then with our end customers and in-store experience, orderful films and so on. Now, you actually work as a process and data inside development manager, and you also say that it's on the customer side. What does it mean and what I assume you are overseeing processes and what kind of processes are you actually overseeing there? Certainly, because with my background in Ikea, the previous role that I had before, the process and data insights development manager, which is a ridiculously long job title that's the first phase of optimization I need to go through. I was the stream leader in our order to cache process, looking at an SAP project and so on that we were implementing, so when our SAP project where that was going and as that was ongoing, we noticed that we had a particular issue in our sales order flow, so if you come to our store to buy the rug and take it home yourself, those processes all work as expected. We have more of a challenge in the examples that the two of you gave at the beginning year of ordering online for delivery and those extra steps in gold. It's still something that we are building into our concept and we are getting more familiar with the customer expectation. So what I am then responsible for, so this processing data insights, we have then created our processing data insights of our PDL, which is then what I am the COE lead of and at the moment, the first process in which we are deploying is in our sales order flow, so we will be growing beyond that and that's something we will also talk about a little bit later. Right now, we initiated a transformative initiative within in called sales order movement and I am then the head of process insights and analytics within that project. Really then looking at the end to end order flow from the point of which the customer places their order through to fulfillment and then potentially any after sales process. And here you have been in the field of process improvement and subject matter expertise for quite a while now, when did you insert process mining into the equation and how does it fit the overall picture of your process management topic? Certainly, because I have been in process development within in for nearly 10 years now. Just over two years ago, actually on my 37th birthday, I started doing a lot of online training through Coursera, LinkedIn, learning, things like that and one of the courses that came up was from Will van der Alst, process mining data science in action and I love that course, fantastic content and nice and geeky and technical, but the thing is then it's going beyond that to actually then deploying that into reality and actually not just being interested in the numbers and the data, but actually the outcome of that data and deploying those insights into something that improves operation. And then so as we were going through with this sales order movement, it was apparent that we had lots of ways of saying we think we've got this kind of problem or we think we've got an issue here, but the ability to articulate that and to actually get some data as a result of that or some data to give you a result. We were really limited in our ability to do so and we've been talking with certain providers of process mining capabilities, one of which was solonous and yeah, it was just a case of we we'd had conversations about it in the past but the maturity of the business to actually be able to incorporate that into way of thinking and our rationality took a little bit of time. But as that then sort of reached a tipping point and then we started looking at we did a proof of concept and then just over a year ago I was asked to then lead this basically what we then delivering is process insights as a service. Sales orders will be our first process, our first paradigm and then we look to deploy it throughout the rest of the business, which we're already starting to do. I'm always curious as to see when big companies or enterprises decide to deploy such a thing as process mining. I'm most curious how what the timeline is on this because I know that some are self-starters and do 20 processes in the first year and things like that and some are you know a little bit more served and try to do one and get the most out of that as they can and then see maybe in the next year or after that to see where next they could deploy this technology. So where does where does it go fall into it? I think actually I was I was having exactly this conversation with my account executive from Salonis Manu. I think the first question I would like to ask there is what is a process because does everybody have the same definition of what a process is because I also reflected on Timo's presentation at Celestir when he was sort of saying right look at doing two or three use cases within a process and then start looking into other processes. Now the thing is the process that we're taking is the full end to end of the sales order process and when I've sort of spoken with some other customers where they say look at the picking process or the delivery process, I consider that a sub-process and so it's calibrating what we mean by these things because ideally the way we look to deploy this is looking at say order to cash, source to pay, record to report, higher to retire as a process in which case there are only four five six kind of processes and so when people say we look at 20 processes I'm thinking what 20 processes and so from that calibration that's been a really interesting series of conversations and that kind of thought process was also triggered when we saw process sphere and business minor at Celestir and so on. Where it's like you can take these bits of the process and connect them together into object oriented processes and I remember the facial expressions between me and my colleagues who were there as well that's like that's what we're doing that's what we've started so it's a really interesting conversation with and exactly as you say so going back to your question which I've taken a bit of a security to get to understanding what we've gone after first is where we felt we've had the biggest pain and also where we had the biggest passion and momentum so there has been sort of that emotional aspect for it and also as well as then easier to get the buy in because someone's then saying we've got a problem we want to go after if you can come in and say we think we have a way to help do that it's much easier journey for us to go through and yes that will be very very different throughout different corporations and that's really something that we've learned when talking with different groups because if you have a very top-down structure things can look very different from if you then take sort of the Swedish structure and Swedish mentality where it really is management by consensus a lot of collaboration a lot of consensus involved we're also as a business we've always been very front-led so it's closer to the customer guides the principles but our customers are not the same across 360 stores in 30 plus countries so you've got quite a bit of fluidity and flexibility there so there will not be one answer to that question but I think that if you've without that some senior stakeholder buy in and some support to really drive this through the organization it's going to flounder a little and then from that perspective we've got a lot of support from our CFO from our global business relations manager and then that's now starting to evolve throughout the rest of the organization through our digital organizations and so so yeah start where you've got the passion and grow it from there yeah we also heard this before about the top-down support and how important this is to really get the necessary resources and drive the initiative forward we even did a solo episode on this with Max Roglinger so they're interesting for you guys if you want to listen to it just go back and do so but moving forward you in IKEA work on something called perfect order and you kind of define it here already sort of when you said what is actually the process and I find this also very interesting because it's never as simple as taking one one identifier would be a sales order or the final invoice and just put it in there without the context so how do you currently look at your perfect order process and what can we and the audience picture and imagine when we say okay this is IKEA's perfect order. Certainly one of the things that we're really trying to do is use some external benchmarks to also make sure that we've got some comparability with the rest of the market so the base definition that we're using is from APQC and there is a perfect order definition there about the perfect capturing allocating the inventory delivering on time in full with the correct tax document both as sort of the basic principles and we're then sticking to that as much as possible because then that is going to give us the best ability to set a baseline that we can then compare internally but also then when we compare to other retailers CPG other companies of our size because APQC has the ability to do those analyses we can then really see where we are and we can we can benchmark ourselves so yeah when when we then take that as a base that's when we look to mirror that in what we can see in process explorer and so on with insulin so that we can then say the business events that we've then captured from our systems and many of our systems are in house built in a few off-the-shelf solutions that interact generally pretty successfully but then what we need to do is understand okay capturing of the order that's this event here allocating of the inventory that's here so on and so forth and then when we talk about the on-time delivery that's the one that we're just finalizing at the moment that's then understanding what was our promise to the customer in the first place then when did we actually deliver the goods so your rug hopefully you wanted it to be delivered today and we did deliver it today that would be the dream and and and so that's then the perfect order and of course we want to maximize that as much as possible but the counter to that is well what are the things that cause an order to be imperfect because that's ultimately then what are the levers that we can pull to either increase the rate of perfect order or decrease the rate of imperfect and then even when we do things perfectly that doesn't necessarily mean that we do them optimally because we can do the perfect sequence of events but are they all done in the right way so that's that's also that next level of analysis that we can do but right now the the current focus that we've got is then analyzing once again going back to the process explorer we've got a series of events that must happen for an order to be perfect but then there's another series of events that must not happen because if they do happen that's what the identifier is being imperfect and then it's that breakdown of the analysis of those events I think we got 31 of those at the moment where we can then say right when this event occurs what is the impact of that and then what are the root causes of that impact and how do each of those events interact with each other because just analyzing what happens when this event occurs actually with insolowness you can do that with the conformance check but what we're now starting to do is look beyond that to then really say right and how do all of those interact with each other and what are the root causes of that and that's what we'll discuss when we get into the data fun in a while I kind of wanted to ask you you said that you take your perfect order definition from my well established sources I'm assuming and but I was also going to do does that translate to the perfect customer experience like do do those things like intersect like in the middle or is it more of a even though it's a perfect order we might not still have the perfect customer experience that's there's definitely an intersection between the two but there's also definitely the things that the customer would expect before the consideration of a perfect order that we might not want and vice versa for example customers may want a lot of freedom in changing updating and doing things to the that may cause what we would consider to be imperfect and that calibration of those definitions that's also something that the what we're able to investigate now can actually give that input into so that we can get either consensus between a company perfect order and a customer perfect order or a clear articulation of potentially what the differences are now ideally and Inca is a very customer focused company now our ethos our tagline is a better everyday life for the many people so it really is then about that being customer centric and that's definitely a perspective that we do need to have but it's then balancing that because it's taking the right thing at the right time to make sure that we are pulling the right levers at the right moment to then get that result that we're looking for. Now I kind of want to create this bridge into our second part of discussion which is the data science part and I'm wondering if you are a company that does process mining for a year to such as yourself so I assume that you have built a few reports you have some insights you have you you mentioned over 30 activities over 30 events that are occurring there how do you make this next step into a proper and adequate analysis so so what does it even conclude or or captures in your opinion sure because there can be so many different paths that you can take into analyzing the data absolutely and and the thing is at the moment we are exploring in some way shape or form as many of those as possible because we do have people in the company with a huge amount of experience I mean I I've been within now for just under 18 and a half years in in my previous team that made me the third least experienced person one of my team members back in November hit 40 years and so if you then listen to the people who have the experience they've got a sixth sense and a gut feeling around we've got pain here pain here pain here now we're not necessarily going to make decisions purely on that we then want to explore the data to then say right is what you're sensing genuinely the challenge is there another input to this is there another input to another perspective so we definitely take that we look at the the feedback from the countries because once again they're the ones meeting the customer it's the co-workers in the customer support center who are answering the phone calls the co-workers in the store helping the customers plan their kitchens they know a little because they experience what is then really hitting and hurting but what process mining then gives us the ability to do is accelerate and optimize that analysis and we had one example in there was a team doing a time in motion study in the kitchens department they spent 10 or 12 hours in that department they analyzed six kitchens they found a certain pattern with four clicks of the mouse we could show them it had happened over 6000 times so you can really get that quantification the acceleration the industrialization of the quantification is a real potential and then also as well if I take my team in my center of excellence and we've been talking with multiple countries on multiple topics for some time now we have an idea you know we can see where the smoke is coming up and with you know where the smoke there's fire and and we are getting better and better at saying look there's a potential issue here there's something to look at there but the reason that we're looking at moving into this data science area and really going into the depths of getting the data to tell us stuff is that I can do the analysis or our team can do the analysis based on what we can see and we can say this is a small medium or large problem but when I get my senior stakeholders come to me and say what are our three biggest issues I have no idea and no ability to say whether or not the things that I've found are the biggest issues right so they are the three biggest issues you've found so far so far they are three sufficiently big issues that we should definitely pay attention but whether or not they are the biggest things that are going to move the needle in the biggest way that's a real that's a completely different question and our ability to actually analyse the data that we have available to come to an answer on that is totally different so if I take just a little brief example if we were to say because we do kitchens we do bedrooms we do sofers we do you know the whole range of your home now kitchens is a big area when you're ordering a kitchen and there are appliances and there are the cabinets there are the custom worktops there are the you know but whatever else we then go into your potentially then some electrical work plumbing work gas work lots of things that are involved and they are high value orders complex processes and so on and when it comes to bedrooms I mean like a a big cupboard if you go to get like the big packs ward yeah yeah I knew exactly which one yeah it's always that because you've also then got all of the stuff that goes in the wardrobe right you've got the beam that you hang things from and the shoe rack and all of those kind of now if we were to just say right we've got a certain perfect order rate within the company which based on our current analysis is running at about 80% the question is is that the same across our entire range is that the same in kitchens bedrooms or other let's just take those three options now let's also consider did the customer come into our store go online either through the app or on on a laptop or something or did they then ring our customer support centre for then our remote selling team when we then said right this is what we're going to provide to the customer did it come from one of our stores or did it come from a central place when you were maybe ordering a kitchen did you have a planning service before you then or did you have a measuring service so that we had to really confirmed measurement in your kitchen did you do the installation yourself what did we do the installation now if we take these variables here so kitchens bedrooms or other three options store online remember three options store fulfilled central fulfilled is two option measuring and planning installation or neither there's three options now just with those four variables if I were to do the analysis of which combination has the biggest problem I have to do three times three times two times three analyses that's 54 analyses to then say which thing is the one that I should go off and if I then added one more variable whatever that might be and let's just say that they're wrong yes yeah it was just then thrown his arm up to the sky because also because also as well I massively simplified this why just said kitchen bedroom or rather yeah because if you look at our furniture range we've got around about 20 different ranges depending on whether it's the solid furniture or like like cook shop where we have our cups and plates and so on all the rugs that's a different area and we call those home furnishing businesses we have about 20 of them is that the important layer or is it actually the product area so is it leather sofas or cloth sofas or is it arm chairs or is you know it depends on what kind of granularity we want to then be able to analyze a rough a very very very rough estimation there are somewhere around at a minimum level there are 36 variables at a maximum level there are 585 oh god yeah no no no no I mean if I start with the 30s let's just say those 36 variables were just binary they were the yes or no yeah two to the power of 36 is a rather large number I mean two to the power of 30 is a billion two to the 36 is what 64 billion options now two to the power of 585 is a ridiculous one yeah it's approximately 10 to the power of 195 that's a lot if if every atom in the universe contained a thousand billion billion billion universe the total number of atoms if all of that would be the number of permutations that we're talking about right now it's just I remember back to when I first was looking around universities and there was a lecturer giving a little lecture called how big is big so I love 10 to the power of but yes it's just fantastically big numbers you can't do that with a brute force methodology yeah now this this is where something like linear regression analysis things like applying proper ml concepts neural net and then we get the final four propagation and back propagation and all of those combination of components there because if we were to then take each of the 31 events that we mentioned previously the cause imperfection and let's say that we have these 36 variables that may contribute to those if we get that understanding of how much each of those variables contribute to the likelihood of an imperfect order recurring we then have a map that we can then put a heat layer on top of to then say that's what we've got to go off this attribute here contributes significantly to 15 of our 31 imperfectors we must focus on and maybe it is kitchens maybe it is installation service maybe to send maybe it is when the customer comes to the store to order things I don't know I don't want to go in with preconceived ideas because the one thing I don't want is to then fulfill the confirmation bias we're trying to move from being data aware that we know there is data to being data fueled where data takes us where we want to go to being data driven where we go where the data tells us and that's the phase that I want to evolve to when you say like 35 to 500 variables and then just without inputting your own biases isn't it then just kind of seeing what comes out and doing your interpreting it as best as you can and if it's like three variables that are most relevant to contributing to an imperfect order and wouldn't then the next step to be to interpret it back into the context of the business and see then does that even make any sense can this even possibly relate to each other that's that's absolutely and that's one of the challenges that we often get and a lot of areas that have been deploying AI and ML is that afterwards you get a result and then someone asks how did you come to that result and what is it me and that's absolutely a yeah there is the potential that's why we'll start with a smaller number of variables let's not start with 585 that's a bit because yes and and you're absolutely correct that that is a potential result of this but and then because no matter what analysis we do which analysis we run that's always the next step because we do not want the the next step of the work that we do to or the final step of the work we do to be oh that's interesting that that should trigger the engagement and the excitement it's like oh really so because one of the things that we have seen so far and we were reflecting on this today as a team is that actually so far we haven't necessarily discovered anything that's like brand new and that no one's ever thought was going to happen but what we have done in in almost every case is then truly articulate what the impact is so oh that's what that means because so for example I mentioned earlier we've got this perfect order rate of 80% one of the things that came up when we were just eyeballing some figures is the the bigger the order the lower the likelihood of perfect order by bigger the order you mean more pieces or exactly the number of items on the order and it's a case of well yeah okay I don't think anyone's particularly surprised at but we can then do the analysis to say when there's one item on an order the likelihood of a perfect order is 86 or 87% for every item you add to an order that drops by half a percent oh that's that's that's what we're seeing at the moment so when you get a kitchen where you've got hinges and doors and cabinets and worktops and appliances they can be up into the hundreds of items yeah it doesn't go negative by the way we know so it does it it does tie it off a little bit and they order just the rack it's exactly yeah yeah for a rocket you absolutely nailed your selection and yes that actual aspect of you know it's actually a scalability challenge that we face you know as the order scales we get these challenges but yes so so that realization but the thing is and the reason why it triggered me into action around deploying these deeper data analytics techniques is because if you only analyze one variable at a time if I were to give you the because we've got a graph with a line that goes down if I were to draw a conclusion based on that one graph I would then say we should force every order to only have one item on it because that's the one that's going to pass the one that's most likely to be delivered perfectly and and and that's the thing it's the case of we've got such a complex reality and everybody does all all of the listens for all the companies or anyone that we're working with the complexity is such that looking for the silver bullet and looking for that one thing that fixes everything yeah is somewhere between optimistic and naive and and it's then finding that balancing act because obviously then you know 585 variables is the other extreme completely aware but it's then making sure that we can really go in and not do not give a single variable answer to a multivariable situation because that doesn't make sense it's not right and and yeah it's just finding that balancing act because simplicity is exceptionally important when we're broadcasting to either 10 decision-makers in my steering group a thousand decision-makers in the organization who then need to actually push these operational changes out or the 172,000 color workers we have in the business who may need to operate differently the kind of simplicity that you then need to have inherent in the analysis is very different yeah and simplicity is a lovely result is definitely something you have to push for but we have a phrase internally within Inka which is complex for the few simple for the many and it's then and I like to subtly enrich that to say it's enabling the few to embrace the complexity to extract the simplicity for the many so a little bit more action involved in the steps through that process and that's the bit that we're coming to now because the things that we're talking about here and some of the numbers I've just given you that's based on the analysis of the last two weeks has been such an acceleration through this at the moment it's so we're we start small and scale infinitely as solanists like to say we're definitely going through that scaling moment now can I ask what would you rather have would you rather discover that one of those preconceived notions of what is contributes to an imperfect order like ordering more than one thing right lessons the the chance of a perfect order right that kind of makes sense to everyone you could have guessed that but now it's just confirmed by data would you rather have an intriguing insight like this or would you rather find something that we're seemingly three variables that have nothing in common contribute to a majority of the imperfect orders well what do you think would would you rather have I'd rather make my business better and how I get there I don't mind yeah because because the thing that I'm really aiming for here and is it's the the objectives that we have of our process and data insights about is to make every decision that we have in Inca better now would I on a personal level like to see you know this this causal link that no one would ever have thought to have detected in the past and then snap it out of thinner of course I would that's from 60 stuff but but ultimately if if we then turn around and say some of the ideas that we have had in the business for some time around where our pain is being manifest is true is correct and actually we then properly go after these things to make changes in our operation to make our business better that's the ultimate win would I like to do this huge amount of analysis and say well actually it turns out left handed customers wearing blue socks who shop in April that's what causes the issues although I will say just in case anyone's curious I don't actually have that information because we know GPR no no no GPR compliance and so on so I don't know your handiness I don't know what color the socks are so but yeah that you know that that would be really interesting I think that the thing is then if we discover that in the data and it is something that's that disconnected the journey to explain that to the rest of the business can get the buy-ins and make changes based on that will be a whole other kind of journey so yeah yeah that that will be a really yeah that's an interesting that will be a lovely problem to have to experience and to figure out a way through now what would you find lacking about the usual approach on analyzing your process and I'll say I'm guilty of this problem in your eyes as well when you look at your process you found those process steps that you don't like or you know that probably shouldn't occur in your opinion and you know you filter on that you do this typical drill down you know you sort it by vendors who have this worst performance and then maybe drill it some more find some materials and so on so forth in your view what is lacking this approach compared to this to this very holistic view and analysis of data and different variables in the process I think that there's absolutely a place for it and that's still work that we have ongoing and it's very important to look in that direction I think that the I think I was back to to one of the points I mentioned earlier that we don't have any guarantee that that actually is the biggest thing and the right thing and the important thing to go off I think then now I would also say that we've already experienced the evolution of our ability to see the problem that's right in front of us and actually chase it back to the genuine root cause and all the way back to its conclusion and there of course is a place for it because the the speed of identifying those the ability to bring people on board to then chase it down and to look into it is a lot easier but then yeah it goes back to if my CFO comes to me and says right Tim what are the three biggest issues we've got in the business at the moment as as was mentioned earlier I can give him a list of the three biggest things that we're going after but that's not the question he asked and yeah to then be specific enough to then say right we've got our biggest pain point is when a customer does not have a measuring service before they do their planning to order a kitchen we see that if there is no measuring service there the perfect order rate drops 25% for example something like that I can probably discover just by doing some analysis by doing one by one by one but in order to have that confidence that especially in this big end-to-end holistic process that we're in I've been struggling to find another way to say these are definitely the three biggest things if we were doing something a little bit smaller and we've got a couple of these kinds of initiatives where we focus on if you're ordering some goods and they're going to come from one of our stores and then be delivered to we have this picking service in the store that is a small area of focus it's only 10 or 12 events there aren't as many variables involved then doing this sort of bottom-up analysis absolutely works absolutely when I'm analysing a full end-to-end process we actually have in total 158 business events that we've we're capturing now actually that's gone up it's 173 events that are in the end-to-end flow um going to someone in a store and saying what do you think might go wrong when a customer orders a kitchen gives a perspective a very valuable perspective but is that going to be a holistic trip and so yeah it's just taking it from that other direction right now we are still in this I would say framing part of the of the whole exercise of the process improvement once you identify and let's say that you are at least sufficiently happy with the identification of this potential root cause might not beat up three but might be high enough on your chain of problems that you actually decide to go sorted out now Inca and Ikea are it is a huge organization with so many people involved and process mining is essentially just a part of this whole process improvement exercise so what would be the next step then if you actually say okay when I order kitchen if people don't measure it before the order if we don't send there someone to measure it before we have a huge return rate how do you then first of all probably persuade the stakeholders to improve something in the process and then second roll out this change which can be massive in the organization of your size absolutely and and they're too hugely different those those last two questions that you had are too huge to be different answers because the getting the the decision makers involved and the folks involved to then frame the problem frame the opportunity because we we have in the in the work that we've done so far we actually go in we talk to individual countries once again it's getting that enthusiasm and energy and momentum and so we go into countries we understand what are in their business plans what are they going after we find something specific in the data we bring the business experts together to really understand how the operation actually works to confirm that what our data is showing is relevant to reality and then especially with the work that we're doing with the countries at the moment we've got the buy we then get the buy-in from the senior stakeholders in the country so whether that's then the fulfillment manager the selling manager the commercial manager whoever wherever the area of the business is that we're going to then need to make the change one of the other things that we do because we are processing sites as a service and we can make suggestions to say when you're doing this picking process for customers if you're doing click and collect so the customer orders online comes in order to collect we've seen that this works better than here but we've seen in the data that here is something that needs to be gone on we also then anchor that with the global responsible for that operation because we do then have a matrix leadership set up so there are fulfillment leaders and commercial leaders and someone at a global level who then have their matrices in the organization so we've then got the support of the people who define the concept the support of the people who manage the operation the support of the people who perform the operation and then when it comes to actually rolling this out so your second question um we ourselves are not the ones who are going to be driving the change what we're then doing is facilitating the line making the change to themselves and we're then supporting them with the input the analyses and also the follow-up to then say right you have this rate that's 5% you have a target of 2.5% we've made certain changes we can see we're trending to 3.7 we're halfway there and then that repetitive loop to then say right is this the right thing to do have we done it in the right way did we miss something right have we made the thing better we wanted to make better but then we didn't realize we've made something else worse these kind of things can happen yeah but but at the moment if we look at our first use cases where we are being very country um centric from the origin we you know we go to the country where we find the problem we then make the change there we evidence that that change was the right one to make and then it's the case of when we're doing that analysis we also then look at all the other countries because you know we may be working with the UK and Ireland on something but actually that topic we're working on can also be beneficial for Spain and Finland and then when we've packaged that we can say right here's the work that we did here's the change that we made here's the result that we saw and and actually we're doing some work on this next week where we're packaging together this is like a little video that we can then put on our internet and so but then also those global teams who are responsible for click and collect for example can then also start rolling this functionality out to the other markets because our our markets are as I mentioned earlier we're a very front-led organization stores have a lot of autonomy countries have a lot of autonomy so in most cases the things that we find are beneficial quite universally the change journey we need to take people through to realize that benefit is quite different starting point in Finland and in Spain and in the UK won't be the same final result what we want to get to will be at the same level but getting everybody there quite an interesting and set up and quite an interesting journey for us to go for we're learning how to do it well so if you consider implementing a change an operational change and then you measure the outcome of said change how much is it a problem to attribute the rise and fall of the KPIs that you're measuring to specifically that change that you made or is it best guess or is there some sort of best practice that you use we're we've got a loosely defined way to do this work actually from two till four o'clock tomorrow afternoon this is the workshop that we've got to actually then finalize these things so with each of the use cases that we then look at when we do a value framing target setting value realization phase we're then looking at the different contributing factors to value which could then be a reduction in customer returns or an increase in sales an increase in available inventory and or it could then be customer satisfaction because we've also then factored in the the customer satisfaction surveys and so so we can then see when there's a perfect order the customers of this happen when there's an imperfect order they are differently happy and in most cases less so that we can then when we look at right we make a change on this particular date in this particular flow in this particular country so far the use cases that we've had so far fortunately the area that we've been looking to change has had a relatively stable base that will not be university tree because as with most companies and especially you know during the pandemic and coming out of that we're in pretty much a state of permatransformation things are constantly moving all of the time and so then yes the being able to say this change by this initiative kills this benefit and nothing else had any other impact is a bit optimistic but we do then you know normalize for seasonality we normalize for you know exceptions as much as possible and and so far we've got a pretty good core concept agreed with our business navigation and business steering functions and and that's what we're then going to build upon and formalize so at least then even if we can't be absolutely guaranteed that the number is correct we can at least be consistent and comparable between each of the analyses that we do and that's you know we we we and as we grow and evolve over time we'll get better but yeah it's it's not an easy thing you're absolutely right then with that question so now Tim I think there are two types of listeners now after listening to this episode they will be like I'm super excited about process mining I want to get in I want to just do root-cost analysis and learn some statistics do a course by a wolf on that house on process mining and then the second group which is this is so much work screw it I'm just staying with the old way a question maybe also for especially those excited process owners or practitioners of process mining would be maybe how does your team look like because it seems that there is a lot of work a lot of initiatives that are ongoing how big is it and what roles do you actually fill there in currently my team is five people including me okay so yeah and and actually this is a series of conversations for the week after mix actually based on some of the things that Tim spoke about when he was with you a few episodes ago and and also his presentation in cellosphere is that right now my my core COE is myself as the CEO we read I've then got to what I call process improvement leads which would then be the equivalent to say value architects they're the ones who are really at the call face with the countries with the group functions and understanding the understanding the problem framing the issue and so I've then got a data analyst who's then doing the work of putting together the analyses and the the views and the analyses in in EMS and so on and then I've got a data engineer who's then getting the system the data out of our source systems providing the data model and making sure that all of that structure is so that's my Inca COE at the moment I'm also supplementing with team members from silvers so I've then got my AE and CDM and CDA and EMCs at the moment it's the equivalent of one and a half FTEs I've not got anybody full time but it's then just sort of bringing that head count together I will say that last week sort of the the big initiative this sales order movement initiative I mentioned at the beginning about conversation here we had a steering group meeting there and the material that we've presented at the successes that we've had so far the analyses that we've had so far triggered a lot of interest and excitement I also did a presentation to our extended management team in our digital organisation just before Christmas that also triggered a lot of excitement so two to three times a day I'm now getting approached with what about what about what about you know new paradigms in which to deploy this so what we're then looking at from a from a COE setup is then COE leadership someone who specifically looks after the communications so the packaging of the results and you know the onboarding and the engagement and making sure that we bring in the new people communicate to the people who are reworking alongside us someone from a business navigation business steering when we look at the evaluation and the you know the the facts figures the pounds and pants heroes and cents and so on that process improvement lead data analyst data engineer setup potentially over time the process improvement lead and the data analyst roles will potentially merge that a little bit or maybe emerge the other way and the engineering and the analysis will actually become closer together we'll see how that one evolves over time and then also under the analysis area I'm going to look to have a full-time data scientist because yeah as much as I love the well vandalist course and I've done the Google data analytics specialization on course Sarah there's still a site distance between what I'm capable of and so he's got a PhD in this stuff so you know let's let's bring the true might of that intelligence to bear in this and that competence and capability to bear on this but that's then in my course CEO but the way that I'm looking to then set up is as like a hub and spoke setup so in each of the country is in each of the group functions actually having the experts there and the people who know their business and know their operation know their move because otherwise I'm going to end up with a course CEO of 150 people yeah and ideally we want then this to be an ever presence in each of the whether we're talking about you know logistics and selling and marketing and everybody that we mentioned early finance as well so we're where I've most recently spent time in my company but also then in each of our 31 countries because that's where that's where the rubber hits the road that's where we're meeting the customers you know it's it's like I started in IKEA in the countries in the stores but the last time I worked there full time was 16 and a half years ago so I don't really have a lot of relevant experience anymore I know the emotion of it but I don't know the reality so you've got to get close to where it really happens yeah otherwise what we've got is an ivory tower with a wonderful piece of software with no business applicability yeah and that's what we've got to avoid and that's why I talk about it and going back to the earlier question of what would you prefer finding something that everyone already knew or finding something brand new it's like as long as we're making a difference to the business and as long as we're actually making a difference to the lives of the customers and the co-workers I don't care where it comes from so yeah that's that's that's the idea of the setup that we would then have and then second to last question actually you mentioned that you are getting excited the the management is getting excited the people are getting excited and are approaching you now a personal question to you what is exciting for you in this whole initiative what gets you excited I love I love huge piles of data a bit of a data hoarder myself currently working from home this 65 terabytes of mass just sitting over there so so I got playing with big data sets that's that's that's lovely stuff and the the the transparency that it gives and the undeniable ability of the result is so so valuable here and because the one of the phrases that I used a cellosphere and I also because the week after cellosphere we had Salonist a Sweden where Michael Zink is the CEO of Obsolonist in the Nordics he started quoting me in one of the phrases I've started using of you don't have to wait for everything and before you think you can get anything and you know because and I heard this come up at Salonist a lot a cellosphere a lot and what do you do if you don't have perfect data how do we get perfect data how do we get the perfect input whatever data you can get your hands on you can come to conclusions and one of the biggest conclusions you can come to is how to get better yeah how to make your data cleaner how to make your processes better how to make your validations in the first place to make your processes smoother and and the the thing that really comes up is just that aspect of improvement and getting better and because because there's a couple of things that sort of happened over the last couple of years as I mentioned on my 37th birthday when I started doing you know the studying and so it was actually something my mom wrote in my birthday card because when my mom was 37 she went back to university to become a primary school teacher and so she sort of said look at 37 you can do anything you want you they're still the whole of your life ahead of you and it's like actually the one thing out because I've got a pretty good career I'm positioned pretty well I'm financially rather stable that's all not a problem but what I want to become is better and actually then spreading that message and and also then and it took me it took me longer than it should have done to realise the resonance of this but it goes back to that statement I made about IKEA a better everyday life for the many people that's the word that's the phrase and then the the other side of that is then doing a lot of reading that I've done with Simon Sinek's work with start with why and the infinite game and these are fantastic materials there's such an interesting perspective and a great TED talk as well with this and start with my TED talk it's like you can never win retail you're never going to get to the end of it you're never going to get to so that constant desire to be better and once again going back to because in 1976 Invar Camproud who's the founder of IKEA's the I and the K like here published a book that we hold very dear to our hearts the furniture dealer's testament and point nine of the nine points that he raised is many things remain to be done a glorious future and it's that phrase in particular of we're not done we're not finished we can always strive to be that little bit better that little bit or I'm not even just that little bit but in some cases a big bit and so so yeah that the the greatest ability that I have seen in my time with the company to truly articulate what we can do to be better than we are today oh very well said I really resonate with these words and final question thing before we let you go where can people find you and eventually raise a question or catch up with you linked in is probably the easiest way it's the good thing about having a nice short name on my side to find I'm Tim Hill's the the well we're not recording video here but the only difference there is that I don't have the facial hair this is very much a this is very much a covid situation so yeah Tim Hill's an easy place to find me on LinkedIn so yeah if anyone wants to get in touch anyone has any questions anyone has any thoughts absolutely please feel free yeah we will definitely tech them in the show notes so be sure to read them either on process and block or just go on a mining your business podcast.com Tim nothing more to say than thank you very much for this enlightening discussion I really really enjoyed it and I hope and I know actually our audience did as well so thank you for coming to our podcast you're more than welcome thank you for your time everybody dear listeners thank you as well for tuning in on another show of full of process mining and discussion on how to get more out of your process data we are very happy to have you if you like the show rate us you can leave a few good stars and Spotify or Apple podcast and if you have any questions just reach out either through email at mining your business podcast at gmail.com or just reach out to LinkedIn as well we are they're quite active ourselves so patriotic then thank you very much for today's session and thank you all