View transcript
My name is Matthias Elsesser, I'm working for PwC, there I'm heading the data-driven marketing practice in Europe. I'm since 22 years now in consulting, originally studied physics, later then an MBA, and then I don't know how this happened but I just went over into CRM IT consulting and from there more in the marketing area and since 15 years I would say I'm purely in the business consulting for data-driven marketing for huge advertisers all over the world. It's crucial to use data I think for marketeers because in the old days it was these sneakers, jeans, we sometimes say it's from madman to mathman, and that's what we see here, so it was the mad times. We just by gut feeling decided this is the right way. If you now look what today happens, we have millions of data points, our customers have 20, 30 million data points per day for each individual which visit their webpage or whatever, and this gives us now the possibility to really do database decisions driven by data and we see they are much more effective than what people do straight from the gut here. If you look at the projects we are running, we see up to a factor of 10 in conversion rates if you rely on data, better conversion rates. For sure the costs are a bit higher, there's a lot of infrastructure you have behind, but still the cost per action for depending on the action are up to a factor of three better than by gut decision we have done it in the past. So I would say this is one of the reasons why relying on data gives you transparency through the full blown marketing supply chain. From the original idea of an advertiser, yeah that's what I'd like to do, and then you get your creative concepts. And if you then start bringing them into a business, you're going to have a lot of data. You're going to have a structure to measure everything from the creativity to the planning to the budget to the audience and all the other things. And then later bring it through that you finally see this in the UTM tracking parameter in a click or whatever. If it's really consistent and close to it, then it gives you also transparency for your money runs on it. And that's the two things CMOs are quite happy about. What we often see, and that's also depending on this big trend, big players going in like an SAP at OV. It doesn't matter if they're all of the big and millions of small ones now adding or coming to the market. It's quite easy to sign a contract with them. That's not the problem. But then you're not data driven. And I think to overcome this issue, we have to use these platforms in the right way is one of the main things. And that's where we help them. And that's a bit of an iterative approach of bringing things together in a personalized way for your company. Because if you look what Porter said here, a competitive advantage is if you build up a system where a competition is not able to copy. So if I just sign a contract with a huge marketing cloud and send out emails, that's for sure every competitor can do this within a few days. If you start doing data driven marketing on all channels, then it doesn't matter whether it's a programmatic display, video, you have your bot, you have your .com page, you have your apps, you have your point of sale, you have your field force organization, all of them act in the same way. In a closed loop by planning consistently the message to the audience you want to tackle. And then automate and push in the right way via all channels. And then do some nice analytics on it and use artificial intelligence to close this loop and getting back in the next planning period. And then once this is done, starting getting faster and faster and going away from this waterfall to a more agile way of learning with a fast iteration and then seeing what the data tells you and going to the next iteration. I think this is the main thing. Big enterprises have to manage. And there's also a huge difference between an e-commerce startup where you have just one room where people are sitting and doing the marketing and a huge enterprise which does whatever 300 million marketing spend over 80 countries with multiple brands. And I think there the real issues appear. How do I transform this kind of enterprises to a data driven world? A few classical KPIs or metrics. Let's call it measures. It just gives you the full-blown picture and that's a wonderful question because we referred even you and Omar on this. And we are deeply convinced it's still the case. So as I said I studied physics a long time ago and the things I've done there was a huge laboratory with lasers and whatever and at the end I had three measure points. It was the temperature, the power of the laser beam and the time. There wasn't more in there. So we start thinking also with PWA. Do you see, hey, what is data driven marketing? What is the essence? And what does a ZMO really need? And what we found out, it's not more than 10 smart measures. That's all. And that's really the classical one. You need your universe, your audience size, reach one, two, three plus. You need a bit of costs. You need the impressions you have sent out, the actions and the sales amount. And then some KPIs which gives you a bit of a happiness of your audience, which is mostly in the promoter score or some brand awareness. And that's it. So we are talking about a really tiny set of 10 measures. And it gives you the full-blown picture because everything else is calculated out of it. And that's the interesting part. And as said, the ZMO is often totally overwhelmed by the sheer amount because you're just logging into one of these nice marketing backends from Mr. Zuckerberg or whatever. Then you get 5,000 KPIs. But if you're honest, they're all calculated out of 10. Because you can't measure more in these systems. We're throwing balls on people. And we count how often we hit someone and how loud he shouted out. And that's it. So these are the things. So I'm deeply convinced there's only a few measures. You could calculate a massive amount of KPIs on it. And on the opposite side, if you shrink down now the measures to this tiny set of 10, the dimensions we measure this getting super broad. And that's as you ask for the challenge. That's exactly one of these challenges we see at our customers. Even if it's only 10 measures, but you want to measure this for every creative, in every campaign, for every brand, for every audience segment you hit, and all these things. Then suddenly you get data cubes with just an impressive example of 1.6 trillion cells of data. Even if you only have 10 measures. And that for sure gives you a nice picture of the reality. That's, I would say, an answer on this question. Video, in comparison to all these static things where you just push out a message, the interesting part is the engagement within the video. So where's the bounce? Where does this come from? These things. At least it's, again, the same. The only thing you could really measure is, has he started a video and how long has he looked at this video? And where is the bounce? And what's the point he bounced out? So from, again, the scientist's perspective of this, there is nothing more. The interesting part is then if you start thinking about why is he bouncing out of this video? What is the reason why he got out? And there, just an example from a customer, they start tagging their videos really completely. And that was an interesting part because they tag their videos with hundreds of different tags per second. So they knew in this scene we see a sponsor. A partner. He has sunglasses off. There's a green meadow. It's raining and whatever. This was the way they tagged their videos. And on the opposite side, they also used the DMP with heavy usage of third-party data to analyze their audience. And then machine learning was quite nice because even if you only have to measure how long has he watched a video and where's the point he bounced out, you could use these massive data pods of this is the video and we know exactly how the video looks like and this is the audience we come from. And then the machine started to cluster this together and found the overlapping cluster and said, okay, due to the fact that it rains here, that this is an island, and that this is the guy XYZ, the ladies with a household income of XYZ and two children is bouncing out. And that's, again, it's not we need 5,000 measures. It's how we combine data and get them together. And that's where I see video metrics. So just focus on the core. Yeah. And then use. I think it's a multiple dimension of machine learning to get this together and really generate insights, which makes it then data-driven instead of using the next 5,000 KPIs for someone around the corner and then tells you that's what you have to look for. Yes, so what works better in the world of videos and the channels, we often get asked this question, are you able to help us exactly on this? Where should we play now our 30-second video or is 60 seconds better or even five seconds here and then teaser it and move them around to our video portal to then show the two-minute thing, whatever? To be honest, I don't know it. If I would know it, I would count my money somewhere on the island here because nobody in the world knows this here. We are all humans. We behave here. So the only way of doing this is testing. So the only way of doing this is massive A-B testing. That's part of our consulting approach and framework that we implement this. And as I said earlier, if you have this, ongoing loops of planning something, doing something, measuring, and we start accelerating on this, we could also get broader, not only doing one loop, we could at the same time doing 10, 20 loops by massive A-B testing. And we test the 10-second, 20-second, whatever, 30-second video, the blue one, the pink one here against different audiences on different challenges. And again, we have the same what I told you with this video. We start clustering and overlapping and then we ask the data. We say, give me what is most successful. And that works quite perfect. There we see these conversion rates, factor 10 or whatever, if you work like this. There's an easy thing, and shit in, shit out. So creativity is still, as a physicist, I would say the light speed in marketing. Everything is measured against this. So first of all, I would say it still has to be engaging, creative. People are curious. They want to see it. To really get viral, that's some. Some herbs you have to put on top of your video. Otherwise, it will not work. That's for sure. But once this is done, and let's assume this is our fundamental basement where we start, and we have these nice, creative videos here where we have this possibility to get viral, then it's again this pushing them out in the right way. And not just uploading them in your YouTube channel and hoping something happens magic. There's really no teasering them. Maybe going first via social. Thinking about digital. Social attributions. And first doing some nice clips or whatever to get them later into your longer YouTube channel and things like that. So it's more this experience building around this. And bringing people into the game. Instead of just this is the one and only Christmas XYZ ad. And for sure this works from time to time. But I would say this is more a lucky shot. There are others who are doing this frequently. And if you look how they work, they're doing it. They have multiple channels. Even the audio or whatever. Instead, have you heard about hmm hmm hmm? And then people getting curious and maybe searched. And you have to be in the search channel that you're ranked quite high. And then from search they jump to YouTube or wherever or their own video portal. And this is the way video works. General trends in the data driven or even in the full blown marketing area. So I think the first big trend is we're going away from this digital drunkenness. That's what we've seen on the Mexico last year. That's what we've seen here on Omar. We're coming back to playing the full blown possibilities in marketing. So out of home suddenly, whatever TV getting back and people are really starting thinking what is the right concept for me. They're advertisers, big advertisers for sports apparel or whatever. Who's just announced we're going off of TV or whatever. The others, if you look at Brooklyn Gambler or whatever, suddenly announces other things that they want to do. Things like this and this. So this is what I see as a trend. Good marketeers getting aware about this is the way I'm able to play this piano. And starting using everything. And the channel is getting everyday broader and broader. So you're no longer able to say, okay, that's like in the classical days that I have my TV, my radio, out of home cinema and that's it. You have to be on the edge of the market. You have to visit these conferences and whatever to ongoing see. Here the Snapchat arises, whatever. Now Instagram is the right one. And suddenly the channel is getting broader. Now we have voice. Now suddenly Alexa and all these things coming up. So this is a huge trend. And then the other trend is what we have talked now in the last 30 minutes is being able to play this piano. The one is to understand this. The other one is playing the piano. And knowing how to get really nice music out of it. And this is building up this consistent closed loop marketeering. And that's what we see at our clients. What we are doing on our clients' floor, we mostly help them to get structure in their marketing department. To go away from this sneaker jeans, Excel area to a more enterprise software driven, industrialized production of data driven marketing. We even developed a known framework for it. We call it Matomics. That's AI. That's 80 tiny atoms. Each covering a dedicated topic. And you can take them and bring them together for a dedicated problem we see on a customer's floor. And then we start in a quite agile way every two weeks. Develop tiny pieces. Develop them in a more data driven way.
