From data to decisions: management's reporting needs

Frode Stenstrøm: Good reporting is essential for good decision support, and I have been wondering a bit more about what this actually means. In today's episode, I discuss this with Per Kveim and Eivind Lofthus, who are two experienced managers and analysts. My name is Frode Stenstrøm, and this is ProsessPodden. And before I get started, I would just like to say welcome to you, Per. Who are you? Per Kveim: Thank you, my name is Per Kveim and I am the general manager of ProsessPilotene, and I am a mature man who has had the task of managing various types of IT organizations over a long career. Frode Stenstrøm: Yes, welcome. Hi, Eivind. Eivind Lofthus: I have worked pretty much my whole life on creating good decision support, or what I believe is good decision support for companies across many sectors. Frode Stenstrøm: So you are the supplier that Per is supposed to be satisfied with? Eivind Lofthus: Actually, in many ways, yes. Frode Stenstrøm: Data-driven management is something everyone talks about. Every company would like to become data-driven. But what does it actually mean, Per? What do you mean by data-driven management and business development? Per Kveim: Yes, for my part, it is about having a factual basis on which to make decisions, and not just gut feeling and experience, as we call it when we get old. Decisions do feel much safer to make when you have a basis for decision-making and something concrete to relate to. And then it is about both understanding and analyzing the past to learn something about how your business model and your processes work, but first and foremost, you do it to try to find out something about the future. Frode Stenstrøm: Make a forecast, in a way? Per Kveim: For example. Those of us whose job it is to manage, we need to convince ourselves, but we also need to convince those we work with. I have often experienced being challenged on why do you want us to do it that way? And then it is always useful to be able to explain the reasoning behind it. That is one thing; the other is that we do not decide our future and our accounting figures and these factors that we have to comply with, but we decide on the input factors. Frode Stenstrøm: In other words, the underlying figures for the report. Eivind Lofthus: I would also like to connect this to that somewhat worn-out concept of digitalization. Because digitalization 1.0 is, in a way, about adopting and digitizing processes, adopting digital tools, while digitalization 2.0 is really about making use of the data that all these digital processes enter, and using it for something sensible. I think that is a good way to explain what it means to be data-driven. Per Kveim: Today, we spend a lot of money on electrifying the processes, as we call it, that is... Integrating our work processes with the information systems, and then having a potential ocean of information that we can use. And back to what I said, we do not decide the result, but we decide the input factors, and learning something about how the input factors affect the final result is very useful for a manager to know. Because those are what we can decide to change. Eivind Lofthus: I mean, there is something about gut feeling; it can be good, but it takes 10-20 years to establish. Per Kveim: Now we live in times that have changed dramatically over the past two years, and perhaps we can say that we can include the coronavirus period as well. Clearly, when you end up in situations that are not perceived as normal, much of the old knowledge will fall short. That is when having these decision support systems and analyses helps. And being able to repeat them and adapt more frequently is a challenge for managers. And that type of methodology is particularly useful in rapidly changing times. Eivind Lofthus: Something about changes in the framework conditions for what you do. The faster that change happens, the more you need concrete figures, and the less use you have for that gut feeling. Per Kveim: If we are to conclude, Eivind, then it is probably... In short, data-driven management is about being able to look into the crystal ball in a more credible and accurate way. That is what we managers want the opportunity to do. Frode Stenstrøm: But they have mentioned a term, perhaps not a foreign one, but they talk a little about these input factors. Could you give a couple of concrete examples of what you mean by that term? Per Kveim: Yes, take a sales department. One input factor there is the extent to which, how much marketing we use to find sales opportunities, how many salespeople we have. How efficiently they work, the way they work. Those are input factors that we can control, right? And then it is useful to know which of these changes or measures produce which effect. That is what we are looking for. In a modern company, you experiment and measure, right? Then we gradually get a little better. Eivind Lofthus: I have no evidence to support this, but I would probably claim that every company in Norway is engaged in activities that are not profitable, without knowing it. They could have known if they had measured, or known to a greater extent if they had measured Frode Stenstrøm: the underlying reasons for it, plain and simple. I think that in most companies, there are a great many people working with the best of intentions; that is, every employee wants to perform, but it is not certain that you might end up with suboptimization instead of global optimization, and I do not think you can figure that out without measuring, plain and simple. Per Kveim: No. You are blind. Frode Stenstrøm: Yes, but Eivind, I am thinking about the most important prerequisites for getting started with data-driven management. Could you pick out a few points that you have seen from your experience that are relevant? Eivind Lofthus: Well, the most important point is management buy-in. Without that, you will not get particularly far, I would say. When it comes to data-driven management, it presupposes, in a way, that you have systems that enter data that you can capture and store and use and look at the history. Those two things are important. And then there is one thing about analytics and BI that may be somewhat undercommunicated, and that is that it is very long-term work. You reap the benefits of the things you started doing a couple of years ago; you reap them today, so to speak. So you depend on having management that thinks somewhat strategically about it. Per Kveim: You need to have electrified the processes, as you said. And that is perhaps the main content of these digitalization strategies that we see now, at least where the digitalization strategy is about improving existing processes and not changing them radically, which is also on the agenda. There, it is about merging the information systems with the work processes so that we actually partly control the processes, but can also measure them, right? Frode Stenstrøm: I agree about the need for a long-term perspective, because there are never any quick wins in these things, and even if you found an answer, you have to measure and see the effect before you can arrive at a new answer, and that necessarily takes some calendar time? Per Kveim: It does, but on the other hand, there are other benefits you reap by automating processes, as they become more efficient. They make it possible to include the customer in the process and allow them to use self-service to a greater extent. Frode Stenstrøm: But do you believe in maturity theory? I mean, the idea that people understand why, because it is easy to say that I am going to report on a sales figure. But that people perhaps understand that the management team also needs to build expertise in this, that it is not just technology? Per Kveim: It is completely obvious that it is not. This is first and foremost about business, and how you manage your input factors, which we have talked about many times. Eivind Lofthus: In a way, the technology is becoming less and less interesting, really, because it is becoming easier and easier to master. So there has been a kind of democratization, and not least a sharp drop in the price of what it costs to create a good, shall we say, BI and reporting analytics system. Per Kveim: If we look at those who were pioneers, both men and women, in this field, they may have started in the 1990s with analytical tools to manage their businesses in banks and telecom companies, insurance companies, and so on. Eivind Lofthus: I will not mention names, but we purchased an analytical tool with a license costing over one million kroner in the 1990s to create churn models that we can now obtain as public domain tools that do it just as well. Per Kveim: And these companies I mentioned earlier, they had staffs with budgets of over a billion for IT a year, right? Frode Stenstrøm: But in that regard, Per, have you seen any industries that you believe have come further along the path toward Nirvana, a data-driven Per Kveim: We mentioned some just now. These are companies that have been doing this for at least 30 years. Built data warehouses, retained and built up data. Like banks, telecom, I assume also these large B2C companies in grocery retail, for example, have been doing this for a very long time. And use advanced analytics for a range of purposes that... have enhanced the competitiveness of each of them. Eivind Lofthus: To put it this way, when you enter a Norwegian grocery store nothing is left to chance; it has all been thoroughly analyzed. Per Kveim: I worked a lot with banking in the 90s, and clearly banks have managed to, I don't know, they have at least increased their transaction volume tenfold perhaps more, while reducing staffing to a third, and have streamlined their operations considerably and, of course, gained analytical capabilities in the process. Frode Stenstrøm: But are we being realistic, then? I'm thinking... It is somewhat appealing to say that you should figure things out by looking at history. Per Kveim: Yes, we were just saying that managers are now much more focused on this, but clearly there are sectors that do not have such a long track record with this. That is, businesses that have considerably less money to spend on information technology, which are looking at this today. Some of them may have experience from industries that have been doing it for a long time, but others do not. I think many believe in it, but I think many are a little unsure about how to proceed. They are unsure about... how expensive it is. The reality is that it takes longer than one would like, but it does not cost as much as one would like. Fear, I mean. It is cheaper than it used to be. It sounds demanding and difficult, but I think this is part of the business's digitalization strategy. You start by wanting to get your data in order, and then the processes, and measure and manage them, and then you get these analytical capabilities relatively cheaply in the bargain once the data is in place. Eivind Lofthus: In any case, it is a very bad idea to ignore the whole thing, as some perhaps do. Because there is no doubt that if you master this, you gain a competitive advantage. Per Kveim: Yes Frode Stenstrøm: The opportunity cost of not doing it may be much greater than the costs you have to invest in it? Per Kveim: You have to assess how good your competitors are. Frode Stenstrøm: right Per Kveim: You have to stay a horse's head ahead, and then you'll be fine. Frode Stenstrøm: But Eivind, when you have worked on these projects. I know you have been doing this for quite a few years now. Do you get requests that are completely off base? That is, are there completely unrealistic expectations of what they can get out of these types of projects? Or are expectations too high? Eivind Lofthus: Yes, I have experienced that. I perhaps won't mention examples, but I have experienced things so far-fetched that it is obvious that someone has been deceived. Frode Stenstrøm: Are there harmful analyses, or analyses of how it works against its intended purpose? Eivind Lofthus: Yes, there probably are. Everything exists. But to a large extent, that is linked to simply poor work, I would say. But a recurring issue in large companies is that you have somewhat different teams producing somewhat different figures with somewhat different definitions. Sources, and then some poor manager has to look at it, and the figures diverge. That is a very bad idea. It is much better to centralize figure production in large companies than to let it be decentralized. Per Kveim: Another good piece of advice, if it seems a little overwhelming, is to build such an analytical culture, and start small. Most companies have... KPIs, that is, key performance indicators, that they measure themselves against. Perhaps they do not know the term, but they know they measure revenue, they measure costs, and they measure profit. Finding such KPIs in the company, that is, starting to measure them, for example for inputs and results, simple things. That will build Per Kveim: such a culture, and then share them throughout the company, and make it something people rally around, and develop a Per Kveim: gut feeling for how these KPIs move when you do different things, and so on. That is a start, you might say, and a good place to begin if you have not done it. The most important thing is to share the figures with the entire organization. Frode Stenstrøm: Eivind, I would like to ask you about that because you have been involved in this for quite a few years. Many times now, it has been said that measuring inputs is important. Perhaps more important than measuring the result, because in principle it is a function of the inputs. In the projects you have been involved in, do you find that the customer is aware of what the KPIs should be, or do they just pick some figures? Higher revenue, for example. Eivind Lofthus: It is very difficult to say anything general about defining a good KPI, because it depends so much on the company, the value chain, the sales process, and everything else. There are of course quite a few commonalities, but it varies somewhat how well thought-out KPIs can be. Frode Stenstrøm: Yes, because in the projects I have been involved in myself, I have often seen that customers want to measure results, I find that you talk a great deal about the opposite, measuring Frode Stenstrøm: the effort. In the projects I have been involved in, I have not really seen that there has been much of a culture for that. There have been KPIs defined by the board, and revenue, top-line growth, that kind of thing. Per Kveim: But sales have always been measured, for example, on things like the number of leads, number of meetings, and so on. Those are inputs. A service organization is measured on... how quickly you can respond to an inquiry, for example in relation to a service level that has been decided on, right? That is, those are inputs. And then you can look at results and customer satisfaction in other measurements that are results. So it can be done, and I think companies do both today, but now we are talking about the very simple KPIs; it is when we dig a little deeper that we gain new knowledge that may be unfamiliar to many. Eivind Lofthus: I think one thing that managers may have a tendency to do is get too involved in the details of what the KPI should look like, instead of leaving it to professionals to put the figures together. It Eivind Lofthus: is a bit like going to an architect and saying I want a window here and here and here and here, instead of saying I want a house that lets in a lot of light. That may be a better idea, I think. I don't know if you completely agree, but that is how I see it. It is always Per Kveim: wise, when introducing a KPI, to have a clear purpose for it, right? It is much more important to measure the things you want to improve than those you are not so concerned about. And having many such measurements is not wise. It is much better to have a few Eivind Lofthus: completely agree Per Kveim: To rally the entire organization behind them. Frode Stenstrøm: While we are on the subject. With KPIs, I am wondering how do you create a good KPI, Per? Per Kveim: Well, a KPI has to be measured in a consistent way over a long period, and not all KPIs are such that the figure shown by the KPI tells you whether it is good or bad. What may be most important is whether it is going up or down. Eivind Lofthus: Yes, it is the trend that matters. You will never find the perfect KPI that defines exactly what you are looking for. But the trend is always interesting Per Kveim: Right, and then you have to try to measure it in an objective and consistent way over a long period. Frequently. Frode Stenstrøm: Frequently, it sounds like, yes. If we think about the fact that we are going to measure, and report on either tabular data or KPIs, which areas do you think are the most underanalyzed or underreported, Per? Per Kveim: Well, we have spent a lot of time on financial analysis, we have used a lot of results, we have spent a lot of time on logistics, the flow of goods, and the like. Now, most of us work on things other than these. We have spent little time on sales, service, different aspects of delivery, that is, we have spent little time on all customer-oriented processes, and we are also seeing that now in these digitalization processes or strategies being developed in various places, it is often the customer-oriented processes that are in focus, next in line in a way, and industrial production is so thoroughly analyzed and optimized Per Kveim: and saved on the fact that it has been moved out of the country to a large extent, unfortunately Frode Stenstrøm: When we are going to create these reports, we need data, and Eivind, what challenges do you see in obtaining the basis for the reports? Or the analysis? Eivind Lofthus: If you have thought a little long-term and made sure to store the data, then you have done a great deal, but you must have done it a few years ago to be able to use it today The actual collection and storage, Eivind Lofthus: that has only become easier and cheaper over the years. For example, Microsoft with Fabric has actually made it quite simple, and quite inexpensive. It is not really a major challenge; it is more of a challenge to find the interesting information in that data. It is the same as before. Frode Stenstrøm: In many places I have been, the data quality is really poor, isn't it? Can you get through that somehow? Eivind Lofthus: Yes, it is probably often worse than people realize. That is a general observation about data quality. But what you should do with poor data, or unusable data. You should get it out of the operational systems and store it on a dedicated analytics platform. And then you can identify and process the bad data. And a good, what shall we say, not statistician, but a good BI guy knows how to do that. Then you can still get value from that data. Frode Stenstrøm: You do not have to have twelve people sitting there Frode Stenstrøm: cleaning the customer database before you can get started? Eivind Lofthus: No, you do not, but Per Kveim: for one-off analyses, you can do what you just described. If you are going to have ongoing analyses, which are the most valuable, at least initially, then you have to clean up your data. Eivind Lofthus: The best thing is to change it when the data is entered, that is, when it is created; that is of course the best approach Per Kveim: And that is also a prerequisite for achieving process automation and everything else we are working on now, so getting the data to be consistent across all systems, integrating it, making sure it is clean and tidy everywhere and uniform, that Eivind Lofthus: That is a somewhat neglected point when you create a CRM system or another operational system; it is a somewhat neglected point that this is a system that should not only help streamline the process, it should leave behind sensible data. Not everyone always designs for that. Per Kveim: It is important to plan for that when building the next generation of business systems. Frode Stenstrøm: That brings me to another question, which is what data is relevant to collect? Eivind Lofthus: There are two schools of thought. One is that you start with a question or a KPI that you want to measure, and then drill down and find the data that fits it. The other option is that you simply store everything. You do not know what is in there, but you know there is something useful, perhaps a few years down the line. The latter option has become more and more popular as storage gets cheaper and cheaper. I actually believe in both methods, yes. Per Kveim: It is impossible for us to say anything about what you will need in terms of analytics in the future. Once you start doing this, you will want more and more every year. And then it is good to have old data that you may be able to use. Frode Stenstrøm: Are there any processes you could recommend to customers to reach that point, because the challenge is often that if you want to find a good report, you think you want A, and then you retrieve the data to get A, and then you create report A, and then you see that it was not what you needed, and it quickly becomes a cycle. Are there any better ways to figure out what kind of data you should collect over time? Or does it become an iterative development that simply evolves over time? Per Kveim: I think storing everything you have the opportunity to store, as Eivind mentioned, is a popular school of thought now, but I think it must be done by people with an analytics background, so that this data... Is preserved in a way that allows people in the future who are unfamiliar with the data to use it. I would like to put in a good word for the iterative approach; I learned from my very Eivind Lofthus: first boss that a good analysis should provide many answers, but then it should point the way to a question for a new analysis Frode Stenstrøm: Good point. Eivind Lofthus: that is kind of how it is, but yes, both, I like that a lot. I wonder Frode Stenstrøm: if you could say a little about profitability analyses and lifetime values, because we have talked a little about KPIs and background parameters, but I think most companies would like to know a little about their profitability. Per Kveim: Yes, so to know something about your profitability, you need revenue forecasts. That is largely what concerns all of us. Most, 9 out of 10 companies that go bankrupt, go bankrupt because their revenue Per Kveim: becomes too low. And not because they were unable to keep their accounts. And... Being able to create forecasts for future revenue, which in turn can be used to manage your delivery capacity over time, take corrective action if you perhaps identify a problem six months in advance, and then perhaps manage to avoid the problem instead of finding out three weeks beforehand. That can of course be the difference between life and death for businesses, and at the very least a source of heart attacks for many managers. So being able to see revenue figures into the future is probably what all of us want most of all. Frode Stenstrøm: If there were one piece of advice you would give to listeners who have a bit of a bad Frode Stenstrøm: feeling in their gut about how their company is doing and perhaps want to know, to take some steps to gain a little more control and direction, what would your advice be? Per Kveim: Be open and honest about the situation, share challenges with all the employees, and be pleasantly surprised by how much positive response you can get from your employees if you are open about all the information, Frode Stenstrøm: and share the performance analyses with the organization? Yes, Per Kveim: challenges, figures make people, when they see figures, naturally think: how can I use this, how can I contribute? Eivind Lofthus: Yes, I think you are touching on something very important. Do not bury your head in the sand and do not engage in wishful thinking. Eivind Lofthus: Is it what I think it is? Per Kveim: As realistic and open as possible. Frode Stenstrøm: And with those words, I would like to thank both of you for your contributions to today's episode.