First of all, welcome to our third Open MIC and today about the topic Digital Analytics and its role in the future e-commerce. Today I have another great expert sitting here with me A real authority on digital analytics He has been involved in many international and large projects for many years and today we can draw on some of his expert knowledge, if I may call it that. For your information, you can always ask questions via the chat and we will answer them as soon as possible in our dialogue here for you. If you have any questions or if something is unclear, I always try to question technical terms even if I don’t take them for granted as general knowledge in our industry. Should anything be unclear, please let us know. Um, yes, my guest today: Matthias Bettag As I said, a proven expert in digital analytics. And for those of you who don’t know him yet, please introduce yourself Matthias. What are you doing and who are you? Yes, I am now with odoscope as Senior Data Strategist. I have been in the consulting business for the last seven years. Before that, I spent seven years working as an analyst. I also helped set up the Digital Analytics Association (DAA) in Germany. From 2009 until last year I worked as Country Manager for the DAA. I’m hosting a conference currently running in the U.S. designed for Senior Digital Analysts as a technical exchange. Without presentations, only for discussion. Ah, okay. That’s it. That’s a little about my background. Over the last few years I have been very active in concepts and consulting for setting up even slightly larger analytics departments in complex business models in various industries. Yeah. I’m sure you’ve got some exciting stories to tell today. I’m trying to tickle as much as I can out of him. As I said, if you have any questions, feel free to write them to us. Don’t be shy. This format is unique. But Matthias, first tell us, we’ll start with the basics, what do you understand or what do you generally mean by digital analytics? Well, the definition is actually quite simple, if you take the definition of the DA. There are certainly different ones, but the principle is to collect, measure, analyze, visualize and interpret data, essentially through the use of digital platforms, websites, apps, all sorts of other interactions between… Internet of Things still comes with pure and complete data traffic. But in the end, it’s all about user behavior to optimize these platforms for their respective objectives. So as a general definition. I see what you mean. You just broached it a little bit already. There are many channels nowadays, 2019, unbelievably many channels. So, there’s apps, there’s chats, there’s equipment, like you said. Maybe in the future refrigerators with which one can interact can collect data points. What would you say, what has changed from your point of view, with your experience, the significance of Digital Analytics in companies, especially in e-commerce over the years or what was the evolution like? Oh, that’s, I think, that would take a whole hour to talk about. But that digital analytics, at that time web analysis, were actually a waste product of weblogs, so the web servers in the 90s with the advent of websites have also written protocols about which accesses have taken place. Somebody did, there are still some really funny reports from the very first time, where people printed them out on the pin strip printer and then made certain markings with the yellow marker. That’s great! From the time comes also the one, that’s right at the beginning, who remembers, there were still counters with hits. Who had the most hits? Yeah, I’ve got the most hits! The abbreviation in the analytics scene H I T S is How Idiots Track Success. Because a hit doesn’t say anything. You need context. You need a visit and a visitor, too. Well, from the, how shall I say, from the IT coming, the first curious recording of such log data which in turn say much about visits, visit intentions, visit situations and of course then also courses, a huge market has developed, which tries in the broadest sense to quickly capture and optimize and to recognize what is there. And over the years something like this has basically worked its way through that it is now no niche discipline at all but has really arrived a central control even on executive floors for several years and also has a corresponding evaluation, which is also explained by the fact that many analytics departments for good reasons are no longer hung up in technology, but directly in the decisive business factors. Sometimes even in finance but very often in marketing because they deliver much faster and better. This in turn leads to a hybrid situation between the understanding of business, of course digital technology and a certain part of technology/IT. I see what you mean. Would you say that in recent years it has also changed in a similar way to IT itself, a bit from a view of a cost center, something that produces costs but you have to have, like a human resources department, to a profit center, something that you see as something that really contributes to revenue? Would you say that is so? I’d say that’s the way it should be. Often, unfortunately, there is when, let me say it like this, when Analysis or Analytics Department, so every call to ask: I need the report or that and the evaluation and so on. If all this is a cost factor for the requester, there will be an inhibition threshold. That’s clear, huh? Okay. And this does not necessarily lead to professionalization and better integration. I do indeed see analytics as a giant option to be built in as a profit center, and that per se has a completely different acceptance. IT comes classically from the cost center corner and there they are also difficult to remove, just as a fleet also costs money but of course has a benefit, that is no question at all. But it is a completely different form of involvement for the processes that take place there. So, I’m actually pleading for analytics to be defined as a profit center. And, to come back directly to the costs, what do you think is a ROI, a Return on Invest, something with which you can make a profit, if you invest something, what is the role of a professional, a paid, therefore vs. a free analytics software? Yes, so the advantage of the free solution is obvious, you can use it immediately and it costs nothing. The disadvantage is, you pay with another currency or who doesn’t care from a technical point of view, you don’t have all the functions and power, keyword raw data export, keyword black box understanding for certain configurations or even incomprehension. More difficult to no integration possibilities outside the ecosystem of the respective provider. But it can have a big advantage, because it is just a lower entry threshold, because you can try it out well and you basically have to see it under the aspect of how complex is the requirement? And if I run a big webshop with a lot of traffic, then I can’t really work with a free solution anymore, at least not alone. For others this is quite sufficient. Is there maybe a threshold? So there is a guideline where you would say “Okay, from 100,000 visits, from 50,000, from 500,000, from 5 million visits you should or it is highest… So, a known problem with Google Analytics standard solution may be, does not have to be, as I said, actually the sampling that eventually takes effect is from 500,000 data points per report. I’m not quite sure now. The sampling is the… Yes, sampling is the extrapolation in a broader sense. So, you don’t take all the data available and calculate something from it, but you may estimate it sufficiently well. But it is ultimately a projection and not a pitch over all the data. That contradicts a little bit the Web analysis, because one has all data, right? I remind you of the logs where you enter every line. They’re here! Why is it that you suddenly don’t expect everything for speed or functional reasons? But you have to see how that affects who. The other point is data integration. If I can’t export the raw data to another system, we’ll get right to it, I think when we talk about trends and development, then I’m limited here and that can be a drawback that eventually forces me to take a solution that’s paid for and other things. The third point is data governments, because there are simply regulations, regardless of the hard laws, that companies say, “We want a system where we have absolute control and a dedicated specific contract and so on.” But then there are regulations per company again. Sure, keyword DSGVO, I think every one of us was affected in some way, if only by press coverage. Do you think it makes sense to use other analytics systems besides an analytics system, what should be the standard, that you have something like that? Well, I’ve seen someone tell me before: “Yes, I have Google Analytics but somehow I don’t trust the system. Now I have Matomo installed here again and then I also have other testing systems and AB tests and everything. And somehow I track everything a little bit and make my picture as I like it.” Do you think that makes any sense? The way you put it, no. There are often parallel implementations for, how shall I say, reasons of state. So, for example the IT uses Google Analytics only to look: What is the traffic load? It’s not interested in a specific evaluation, it’s a smooth tool, counts fast and so on. And at the same time, marketing may be working with another tool. That’s possible. God, that’s, that’s still possible. But if I have different systems between the SEA campaign managers and the affiliates and the retargeting managers and the web analysts, which are not consolidated, then everyone looks into his own truth and it becomes very difficult to make comparisons or to substantiate findings and an endless discussion starts. That’s not recommended. I have deliberately formulated provocatively because I experienced such discussions in the agency time. Unfortunately, this is very common and there are sometimes two instances of the same tool that are used and even that can be that the one department Oh great And then it’s about simply building a roll-up that wasn’t planned and you just have possibly duplicated counts with two different ones Those are the classics. You should avoid it. Nevertheless, different tools that have different goal orientations also have their absolute justification. And some tools measure on the side. So, if you have a targeting tool that supports all page views, you have almost indirectly captured the entire web analysis and given to the retargeter. One would not go, if one measures after Page Views, of course into the Retargeting Tool, which does not prepare these data in such a way also but the land there and in this respect there is already double measurement. Important is: which key figures in which system are the validated standard on which is decided. This means that your recommendation would be to establish a source of truth, a Single Source of Truth, which all departments then refer to, so that one speaks with one voice. That’s it. Single Source of Truth is not necessarily a tool! It could be a bigger platform. It can also be a fully integrated platform with many data sources in it. But if, for example, you work in a BI Warehouse system with different data sources and the evaluation takes place there, then you cannot simultaneously make an evaluation only within one data stream… I see what you mean. …which in turn is interpolated differently, weighted differently, perhaps cannot take certain things into account. This is important. So, a Single Source of Truth is sometimes misunderstood as: What is the one sacred tool that fulfills everything? You can look there for a long time. Yeah. Well, I see it’s a very complex subject. Still very exciting. You just mentioned the page views. What would you say, what are the most important key figures for the modern product manager, product owner, category manager in e-commerce. What would you say, which key figures should you look at on a daily, weekly or even monthly basis? What’s the guide? I would also make the frequency dependent on the course of business. Okay. Okay. It’s something completely different when you sell insurance or shoelaces or whatever, lunchtime meals delivery. The Turn-Around is of course incredibly fast compared to the product search for a new car. Absolutely. Well, I’ll leave that out because there you can answer anything depending on the business. But what is important in this area, of course, is essentially the comparison between costs and revenues. So, some call it ROI or ROAS, Return of Advertising Spend, or some go on mage, some go on CSR, cost sales ratio. These are relatively similar values, which in the end always make a comparison with: “What do I spend, what do I take? What’s left? Is it worth it?” From my point of view, it is enormously important that you don’t just measure these pure numbers as a whole, because you might have average values over different ranges that can be better viewed individually. So, you have to segment. You must segment by: Who does what? Even if I keep going to the same website over and over again, I might still go in there one time with a completely different interest. When I order pizza for the family in the evening, it’s a different meal because the whole family is involved than when I order a salad for lunch or something because I want a snack at work. Well, these are different situations, these are different conditions. It may also be that I research something during the day, which I finish later. And that too is another situation. You can’t measure me by my purchase, you can measure me by my commitment. Segmentation “who does what” and the insight from it explains or at least triggers the question of “why”. Why do I have so much traffic today that doesn’t close but is interested? Or why do I have exactly the opposite? And that’s… and that’s the interaction between marketing, what brings in traffic, analytics, what evaluates and the business case that has certain segments, business areas, goals and objectives that are achieved and measured. Yes. You’ve just cut it a little bit already or you’ve dodged this weekly, daily & you’ve referred to the industries. Of course, every business segment is different. Every industry is different, even in e-commerce, right? That’s what people like to think of it as a piece of cake. E-commerce, yeah, it’s kind of all retail or… I don’t see it that way. What would you say, what is the difference, or what are important differences between the industries, between individual business models/business areas? In terms of analytics, of course. That’s right. Very rough: B2C versus B2B. That’s a big difference in principle. There is also B2B2C. So, 2-page marketplaces? A pharmaceutical supplier provides information to the specialist groups and the specialist groups in turn have the customers/patients. But this is even possible in both directions . because the patients can also welcome the specialist groups with requirements. This is a… Complex, yes. is a completely different kind of thing than Fastmoving Consumer Goods, yes? The Customer Journey time, per, how shall I say, conclusion, if we are now in e-commerce, there’s in some way … a purchase… Yeah yeah. In the B2B area, this is usually a form that is sent. I’m gonna ask something or start another trial. B2B lives from forms, e-commerce, too, because even the purchase is ultimately via forms but there’s just straight through the checkout. And it is really important to look with confidence: “How long are the decision-making phases between initial contact and conclusion? Is it perhaps during a long decision phase, are there also short decision phases? What would be such a classic example of, say, a long phase, I need a new bed? That’s it. Furniture. It’s such a classic example. I look at 12 different beds, loosely, I would guess now … You spend your life researching. That’s it. Go to the furniture store, look at it. What would you say are online, maybe even offline, ways how to measure different short-term micro conversions decisions I would call it now? That’s it. So, if you have such a long phase, let’s take a bed or a complicated wardrobe or something and in between you buy, I don’t know, a side table or a laundry basket or the devil knows. Of course, these are quick buys. In the end, one can recognize by a segmentation, quite simply by the product interest, that the visit to the laundry basket, I take now, has no relation to the bed. Okay You can measure yourself. . But if you simply say “Last conversion completes everything that was before” without thinking about it, then you would not be able to see the long period of time well, so to speak. This is actually a challenge, how to approach it methodically. How important this is, but basically I think it is very important to recognize industries, that is, within different industries, but also for his own business: How many touchpoints do I have? And how long does the average or longest to shortest decision phase for a deal take? Okay. Right, and depending on how it turns out, there are other things that have a greater relevance or you’re in fast-moving-good and it’s simply about having qualified traffic on the website and converting that quickly because they just don’t think four days over or whatever. Yes, it is. This is a current trend in the food sector, so brands start to sell directly to the consumer and circumvent the trade. What would you say, what is particularly important in this food industry in terms of key figures, what should you look at? Basically, similar to e-commerce, if the … … But then something else depends, of course, whether I sell the food now, which is more durable and, whatever, then I get cat food for a month in advance or anything, Indian, Vietnamese, any special ingredients that one likes to have in the household, if one cooks accordingly. Yes. The one you buy regularly but the faster the exploits or delivery. This is of course something completely different, because I am much fresher on the way. I can’t just sum that up. But depending on what the use case is, you just have another importance, to look at the periods, to interact quickly, to be personal so that the conclusion goes immediately. So as a rule of thumb, personalization may become more and more important at the point where a quick and easy conclusion from the user’s point of view takes place. Of course, this can vary greatly. So, the closer you are to the product, the shorter the sales cycles, the more relevant is the personalization, the individualization of the experience at the end. Yes. Logically. Because you know what you want. You want to be through with it in two minutes and now you’re not in a rummaging mood… Yes. …when the stomach growls. Unless it’s Sunday afternoon, you might even spend an hour there … I might! …picking out a great new luxury food. That would be, that would be a different Use-Case than “I order myself a pizza… That’s right. …and the kids want noodles or usually the other way around.” Exactly, but in that area you’re already in comparison with a longer decision phase, I’d guess. My first recommendation would actually be to take a look at my own data first, because the customer knows it best. He knows his business or at least suspects something. And this can then be documented accordingly in the data. That’s it. Good. To use this data, which you have then collected… you have a super analytics system in the ideal case and what would you say are the biggest mistakes you can make in dealing with, implementing analytics systems and also setting up such departments? Because I see it quite often, you read in the trade press, it is very difficult to find possible experts, they are scarce, and you might have to build them up internally. What would you say, are challenges that companies there then have to cope with? I always see such a triangle between people, platforms/tools/technology, and processes. Some people say that is actually a rhombus, because culture still belongs in it. Totally fine! This is a field of tension, when you bring in new technologies, then you need people to operate them. I see. And who have the know-how. That can even be externalized, but you kind of need a central. That is, it needs governance that is there. And it’s an interplay. What you have to keep in mind, I think the biggest risk is if you jump too short. If you imagine the figurative, not only jump over the stream, I am missing 10 cm, I have come quite far, but I get soaking wet. And that’s as stupid as jumping halfway or not jumping at all. Well, not jumping at all might be even better. Interesting analogy. But the expectation is sometimes: we either have two fancy people who can handle it. And the technology is underestimated which is necessary for this, which may also require investments and which will certainly pay off. Or exactly the opposite: you have a technology there, but not one that can’t be used to full capacity and that’s of course a shame That, that’s right. If you then invest in an expensive technology and you realize that you only have half the people that you actually need to use it sensibly and efficiently. . That’s too bad, of course. What would you say, what homework do you have to do as a team, as a department, as a company before you can really act data-driven in the digital age? What would you say is the basis, must be given, before I can think about things like personalization, individualization and such things? As banal as it sounds, the question “who is your target group? And what is your goal? ” sometimes leads to interesting discussions. Especially when the question is: “How do I recognize the target group that I would like to have?” If you are now in pure market research personas, you can well imagine that, but the question is: “How do I identify target group A to B with the digital data I have, if people are not completely transparent with a full log-in?” But one must make an assumption or derive an insight from the behavior, from the data points that are there. But if that is clarified, then it is just about a clean basic data collection. I think that’s the next most important step. It must be as complete and natural as possible, valid and understood. I see what you mean. After that, one actually first gets into the area of data preparation and thus reporting. And at this point it already starts, that in many larger companies at least, or with different Stake-Holder employees there are no, in the American one says Data democratization. So, there are no standard reports that are based on this Single Source of Truth. These are therefore based on generally accepted and understood rules and evaluations. And only when the reporting is such that every report recipient gets the data he or she needs and can understand it so well or gets a recommendation or explanation that an action is derived from it, only then do I actually get into the optimization cycle. There are several steps you have to take. From a basic definition over the data acquisition, over the data preparation, over the reportings up to one can actually only pretend in areas, which bring then clearly more complex and really the real benefit. These are basically preliminary stages that take place and it has indeed become easier compared to earlier because the tools have become simple, yes, “comoder”. And because it no longer takes four years to lay such a foundation. That’s all right, yeah. But there is a basis that basically controls everything else. Okay. You just said it used to take four years to get a foundation. No, I was just saying that. Okay. Depending on how big the team is There are companies that have actually needed a long time for this, but also a web analysis implementation that takes a few months until everything fits and no longer wobbles, sounds terribly long, but is unfortunately often the practice. And there you are well advised not to jump too short and maybe to think about what I need and how. But of course, so my rule of thumb really would be to have data integration as a very important property in mind so that raw data can be exported and further processed, even if I don’t yet know exactly what I will need. That means, if I have understood that correctly, I have to consider relatively carefully when designing an analytics system, which data sources, let’s say the classics newsletter, perhaps certain social media data, which I collect in other monitoring tools, whether they can all really run into this system, whether this is possible interface-wise? They wouldn’t run into web analysis, but what you have called, we have data sources, like web analysis, web analysis is perhaps a larger one in terms of volume, but then they run into some warehouse system, where perhaps the things are calculated, if the use cases are given for it. If the company is such that email marketing within it works good, the website aswell. That’s okay. This gains a higher importance, if more knowledge, keyword user centricity, if the data sources become very many, if the company perhaps has very many areas and countries, then such systems certainly come into question. Okay, I get it. You just mentioned it, user-centricity is a buzzword I believe at almost all conferences on e-commerce and digital marketing. What would you say, where are the limits of certain analytics systems when people think about it: “how can I become even more customer-centric? How can I pick up my customer even better where he is? In what situation, with what intention?” That’s it. There are two classical boundaries, I say, which are harder to solve or simply not so easily solvable, sometimes not at all solvable. One is online/offline. And the other is Cross-Device. Okay. If someone owns, how shall I say, a digital universe, a gated community, dominates or owns Amazon, Google, Facebook… The classics. those, where virtually every visit is logged in by itself via different devices, then those with cross-device have no big problems. But all the others have that, as far as cross-device is concerned, and here we are again on the subject of customer journeys, I’m in the fast moving area, where I’m only interested in the visit that I want to serve quickly cross-device is not so important. For an insurer or automobile or a retailer regarding longer chains, cross-device is certainly quite significant There’s a way to tie that together, of course. Email marketing is one of them. And with online offline there are more and more digital touchpoints that allow links. And you have to think about, that from an analytical point of view, people like to be fired at right away and not think about how intrusively this can be perceived. Yes. From the user’s point of view, I think, or from the user’s perspective, one has to think: “What added value does the user, the customer, have, which he himself likes to perceive and which can also be presented transparently?” Because this is the only way to create a confidential customer interaction and to communicate such services as, not as spying, but as added value and commodity. Do you think that this non-intrusive behavior can also be a differentiating feature in e-commerce in the future? Well, data protection in general as a quality feature I said years ago that I believe that this will be one such as a trusted shops seal has, ok? That’s what I think, that’s what I don’t think anymore, an atypical opinion, the ……. Yes, sorry, lost thread. What was the beginning of the question? That differentiation feature is not intrusive behavior or data protection for e-commerce… That’s right. So, I, as I said, data protection as a quality feature is one thing and the other is really, what added value does the customer really have from it? Yes. If I’m with my airline, well not the one I own, but the one I fly with a lot, or with the Deutsche Bahn, they should know a lot about me, because that helps me to get around quickly. When I’m in a shop once, I don’t necessarily want to have a personal profile, which sparks forever and thunders me with things that I don’t want to know at all and, on top of that, are distributed with data brokers. So that’s, that doesn’t have much to do with e-commerce anymore. But that’s exactly the interplay where you have to think: Where does the service stop and where does the intrusiveness begin and what is just a little different, even target group specific and also business-specific? Yes. Apart from many other trends, personalization has been a major trend topic for years. It’s said over and over again, “Personalization is indispensable” What would you say, apart from personalization or more customer focus, what is a trend, a driver in e-commerce of the future? With the K5 just around the corner, it’s certainly things to think about. So, I see three things. One is so-called user centricity, the targeted 360 degree perspective, although I think 320 degrees might be quite enough, we are on the subject of just that. But the topic is: How do I capture the customer? How do I reduce duplicates? Cross-device, for example. How can you personalize a customer if it may not yet be super-personal but personal enough? Group personalization can be much more than none at all. Predictions, prognoses, forecasts, yes? Is certainly an issue, what you can build on it so also regarding costs and income, but perhaps also simply in logistics chains or personnel provision or “yes there are 50 people more at the Sixt counter next week, there can then also open a second counter.” That’s very simple, right? But of course, there is more behind it, because behind it the cars have to be provided, maybe the mechanic is responsible for the cars. So there’s a lot going on. Then the topic of data integration, business intelligence, data warehousing, customer data platform is a huge topic. And automation. What the giant Buzzword Artificial Intelligence or Machine Learning has to do with, , what it is usually more common to do, but automated processes including chat bots, Internet of Things, voice control, augmented reality These are topics that are also currently on the agenda in e-commerce, where further experimentation and more mature solutions are happening. Yes, absolutely. To sum up, I think it can be said that Digital Analytics is and remains a very important backbone, so to speak, of many, actually of all digital business models. And it certainly won’t be boring. Unfortunately, our time is already over, we are already talking about half an hour. It just went by. I hope you’ve taken a lot too. In conclusion, do you have any questions? Are there any more uncertainties? You were very quiet today. Perhaps it was also because today we have made a great ride from the history of digital analytics to the future through the present. Are there any more questions? Are there any special requests about topics that we should briefly touch on again? What would you say, dear audience? Is there anything else that burns you under the nails that burns on the keyboard? If there still is a need, please do not hesitate to contact us. We like to be open and plan to continue talking about things in the digital universe that move your business in the future. We hope you took a lot with you and of course we hope to see you next week at the K5. There’s gonna be odoscope on the way, there’s gonna be Matthias. Yeah, I’m there too. And if you have any questions until then, feel free to collect them, drop by the odoscope booth. You can find the exact information on the website. And yes, we’ll be happy when you’re back at the Open MIC next time. You will surely find the dates in our newsletter and on our website. And yes, we thank you for your ear! See you next time! See you next time! Bye. Bye.