Adobe Summit 2019 Super Session - High Tech

Why should HP be re-platforming? At that time you already had a DMP and we already had an analytics asset, all of that great stuff. But the question was, what is the benefit of it? Surely from a platform’s perspective, the DMP back then was actually agency managed. So we bought a DMP, we handed them full control. You guys do what you got to do.

The other big issue was the disconnect between customer journey and data signals. I’m quite sure many of you have your own customer journey. We had a fantastic framework around customer journey, but the problem with the customer journey was, how do you tie it to the data signals that are available? So how do you make sense of the intent, interest, lifestyle, demography, and behavioral data? How do you tie it to the customer journey? So that was another big problem that we had. So, how did we drive change? First one was around platform synergy. We had access to all of this great media data. We had access to customer data, We had behavioral data from our .com or hp.com. How do you condense, consolidate all of this? Especially if you’re going to be re-platforming, if you’re going to be going to a platform away from your current DMP to your new DMP, how do you make sense with all of this data? We actually created an abstraction layer or all of those analytics implementations and created one common taxonomy that gets used for a majority of the media work and also for a lot of the site-site personalization.

The next piece was around tagging, right? This is really core, we had seven deployment teams. So this was kind of, if you look at this from a people process platform perspective, this was start off of a process-people issue. So, what we did was we took over all of the media tagging, Floodlights, your Facebooks, et cetera, into our team and we were, we are currently running at about three business days in terms of a turn-around compared to, the four weeks that we were in.

And last but not least was of course, you know we moved our agency managed DMP in-house and we shifted everything over to Adobe Audience Manager. What was the primary reason that drove us to this? In fact, when we ran tests between our agency manage DMP and Adobe, we run a four month pilot, both of these assets performed equivalently well. But there were a couple of key areas where we noticed Adobe performing better. One was around commercial audiences, so that was a big one for us. Since we are spending a lot more money and energy on commercial campaigns, we needed to make sure that our DMP was going to support commercial and the second big one, which was actually the key reason why he said, " Yes, we’re going to move to Adobe." Was the fact that our partners like Intel, Microsoft, Best Buy, they’re all either moving to Adobe, or already had moved to Adobe. So that was a big win for us because now we could tap into the whole universe of second party data, which we historically had.

The next piece was around platform synergy, around home data needs to be consumed, and tying it back to the customer journey. We had access to tons and tons of data coming from Google’s, from the Bing’s, and of course we were buying data from companies like Bombora, which was the whole intent data. What we realized is using this data, we could ask questions like, what are people looking for? What location do they live in? What time of day are they coming in from? And are they coming from a mobile or desktop? So we could ask a lot of these questions. Of course, when you kind of embody this, this is what is intent data, especially what are you looking for? That ultimately translates into intent. And what we realized to series of test is if we had to pivot our campaigns towards a raising awareness campaigns, the more important element in this conversation is the intent data.

The next piece was around, working with companies like Axiom, Experian, Comscore. What we realized is these companies had now access to data which focused more on the likes and dislikes as opposed to just the intent. Now there was brand affinity data in there. There was lifestyle choices in there and more importantly we knew who was looking for what. I think that’s an element that we were able to kind of grab out of this conversation. So in a nutshell what we’re able to do, is stream all of that data back into the DMP, of course, DMP allows, Adobe Audience Manager already allows you to piece select a lot of these through third party relationships so to speak. We were now able to focus our campaigns around driving concentration. And last but not least, we combined the data from hp.com made sure again, remember this was the whole abstraction layer that I was talking to you guys about. Used the abstraction layer, plumbed in a lot of the hp.com data into the DMP. In parallel, we took our CDP, which is our customer data where we had information about who are our customers? What products they own? How many times have they called into support? When is their warranty expiring? We had a lot of the data consolidated that, and again pushed that into the DMP. Now, we were essentially able to ask questions like, how much of time did they spend on our site? Or what actions did they take? What company do they belong to? Et cetera. And more importantly we were able to get to a certainty when it came to why are they doing this, right? Combining that, we were now able to very clearly get to driving conversion conversations or driving expedience conversations. So that was a big shift in our thinking of taking the data, tying it to all of these different types of signals and then being able to map it to a customer journey stage.

So, also from a data-sciences perspective, as we were expanding the scope of the work that we do, we had to ask the question, are there other segments or other audiences that we have ignored that perform far better? So that’s part of the process that we started building from a machine learning perspective. And again, focusing on objectives like purchase, in this example or engagement, whichever ultimately mattered to the campaign manager. The next piece was around attribute selection. So you’d give this from an insights effort. So what we did was we started building models, train those models to essentially be able to optimize on attribute selection. In this example if you guys notice, the algorithm came back and said, in fact you shouldn’t be going after 25 to 50 year olds. Your prime audience should be 25 to 34 and they are video game enthusiasts and they drink craft beer, et cetera. So those were things that the machine was able to tell us. So there’ll be today, of course, big shift. Think about this, given our 2% about two years back, we’re now responsible for 26% of all impressions. Of course, our goal is to kind of get to the 50 to 60% mark in the next one year, but currently 26% of all media impressions are selected by the data sciences team. We’re concrete running hundred and four campaigns worldwide across 40 countries and responsible for roughly 9.7 billion impressions. That’s a huge shift for us from where we were. Of course, re-platforming was a key driver in order for us to be able to achieve this. Think about customer journey, being able to put right message, right time, right place, right person, right? That’s a very important argument that if you want to try and get to. So that’s really what the crux of our, where the crux of our transformation is going. But what the DMP did was it became central and a catalyst of sorts for us in additional marketing transformation. Thank you guys, appreciate it. -