[premium]{class="badge informative"}

Use Artificial Intelligence and Automation to Meet the Challenges of Personalization

This video covers the importance of personalization, the challenges of doing personalization well, the solution to these challenges (the four AI/Automation offerings in Adobe Target), and the key building blocks for delivering personalized experiences. It is an edited version of Jamie Brighton’s 2018 Adobe Summit presentation.

Personalization in itself is no longer really optional. What we’ve seen various pieces of research indicate that organizations that aren’t personalizing are simply leaving money on the table. Customers are increasingly willing to give information to an organization that that organization can use to drive personalization to deliver a personalized experience. And therefore, if they’re willing to give that over they expect that they will get a personalized experience in return. We’ve also found that personalization itself has a massive impact and a massive benefit.
There is still a bit of a disconnect between what the consumer expects and what brands are actually doing with personalization but generally speaking, there’s an expectation that the consumer has, and in many cases that’s not being met by the technology that’s available or or rather by the companies that are out there. And the technology exists though to, you know to make that happen. Personalization, using automation is the answer in many ways to these things, to these challenges.
And the first is really you need to speak to your customers how they want to be spoken to start simple and do things like automate your AB testing. So many of you will be kind of familiar with the idea of AB and multivariate testing but you can start to automate some of those processes in order to receive results more quickly and to get return on that kind of activity in a, you know in a much more efficient manner.
We’re also going to touch on how automation can actually scale the personalization that you’re running the personalization initiatives and Campaign.
So what are the challenges? You know, why aren’t we all doing this straightaway, or you know, already doing this? Simply knowing where to start can be a challenge, tying up the right content to the right visitor at the right time. There’s lots of different factors involved in those processes. So how do you actually make that happen? It can be hard to know where to go with the personalization that you’re doing. So we often see organizations come up with a kind of a one-off personalization activity very sort of tactical approach to personalization. And that doesn’t really enable them to build on the success that they see with those personalization activities. And you really need a program in place and a roadmap that you can share with the organization channels are not fully connected for action. So we’ve been talking for a long time at Adobe about having data and content together in the same platform. We’ll show you some examples of where having those two things together and an understanding of the customer mean that you can really capitalize on the opportunities here. And personalization is and should be about, you know helping you to improve your primary key performance indicators, but ultimately about increasing the relevancy for the customer so that they get an enhanced or or a better customer experience overall. So key building blocks of personalization. You need data, you need to know your customers and some of these things are easy to get. Some of them are, you know have historically been more of a challenge to source in the first place, but then combine together. So what we’re increasingly seeing with things like the the marketing cloud ID and the ID of the unified profile is making it much easier for you to bring these data sets together. So the behavioral data that we are all pretty comfortable and should be fairly comfortable with capturing from digital properties like app and website give us the way to to understand kind of customer intent to some degree. And marrying that with things like CRM data, which you know, for many years has been sort of locked up in enterprise systems and not available or accessible to the types of systems like Adobe Target which can capitalize on that close to real time. And through data management platforms the likes of audience manager we’re increasingly seeing marketers look at the the value that can be delivered by bringing in third party and also second party data. And increasingly there’s the opportunity where you have a direct business relationship with an organization and there’s some overlap between the customer base for you to obviously within the, you know, the grounds of privacy and under certain terms and conditions share the data that is valuable about the overlap between those customers so that you can use that. And that’s really how we think of second party data. And you need content. So you need to understand what are the offers that you’re going to be able to display what are the things that you can manually prescribe what are the business rules that you should be using? And the offers that are relevant for those rules. You know, it’s fairly obvious if you want to try and message to a gold card holder about a platinum upgrade. Those are fairly obvious things that you, you know the offers are fairly obvious in themselves but how do you then start to automate the content of those offers and the way in which they’re actually built up the layout and the user experience. And then the last piece here certainly from, you know from this particular view of what the key building blocks are, is strategy. And this is about understanding, you know where to use what kind of capability and what feature. And you know, as it says on here, when the rules aren’t clear, then automate. And I think that this is really one of the key messages that we want to try and get across is that you don’t have to think about this as a sort of a direct, you know path where you have to start with AB testing before you can then start to think about automation some great new features that are coming to Adobe Target that will make it easier for you to actually understand what the algorithms and machine learning are actually doing in the platform. And use that to then, you know, build a better business case internally for the deployment of some of these kind of automation features. So what I want to very briefly talk you through is the sort of the four flavors of AI driven automation in Adobe Target that are available today. So auto allocate this is really for us is about starting simple and maybe automating some of the AB testing that you’re doing. So using automation to get results faster to run things more efficiently, but also to to maximize the ROI that you’re getting And that’s really going to help you to actually build the business case. If you can take the test results that you are that you’re getting and put that back to the business then that’s going to help you to get buy-in for some of the more sophisticated things that we are talking about on that automation spectrum.
But I mean, putting it simply auto allocate, this is really about running an AB test, but having algorithms that are running in the background to actually identify which is the best performing, I guess experience from the AB test that you’re running. And then have the system automatically expose that particular creative, that particular experience to the general population while still continuing to test all of the other options so that you get the results in terms of, you know, which of your, you know ABC, N options were the best overall but you are maximizing the display of the the best performing creative for any time or best performing experience at any time while that test is running in order to maximize the the return on investment, the clickthroughs, the engagement whatever the results or you know the goals that you’re trying to get from that particular activity.
I think, you know, one great example from a generic point of view, if you’re a retailer and you have a holiday period or seasonality or you simply want to Launch a new product you can put that kind of, you know put new items into the mix what you’re doing within this particular AB test and actually have the population or have the platform work out based on the way in which the population is engaging how soon they want to engage with that particular product. So if you imagine that you’re running some AB tests on the homepage, it’s coming up to Christmas you add some Christmas themed creative, as people start to respond to that more sort of proactively, the system will pick that up and start to serve that particular creative more often, Auto-target another fairly new capability within the platform. Again, this is about running AB tests, but here the efficiency is we’re still getting the results from the AB tests that we’re running, but we are using machine learning to actually understand customer behavior and target automatically the best performing creatives for an individual based on their behavior or a segment of, you know, of customers on the site. So this is really using combination of algorithms a combination of capabilities, again more efficiency here running AB tests, but rather than just serving up the best performing creative to everyone on the site, it’s working out what is the most appropriate version of that particular AB test and targeting that particular one to those users. And then recommendations. So hopefully everyone’s kind of familiar with the idea of recommended content, video, offers, you know within a knowledge base recommended articles based on the article that the person is looking at. Recommendations is again about using the algorithms to actually predict what is the best piece to serve up and then serving up that combination of items that this is not just the sort of wisdom of the crowd model of recommendations but we have the ability to do personalized recommendations within Adobe Target. So rather than simply, you know, understanding what everyone on the site is buying or engaging with we can actually make a set of recommendations based on the individual visitor.
And then the the final example is automated personalization. So here what we’re actually doing this is kind of the true sort of, you know model building or machine learning that we’ve had within the platform for a while. This is about building models to predict customer intent customer behavior, and then using that model to actually serve up the right content to the right person hopefully at the right time. While then build having the feedback loop to understand do they engage with the content or the experience what impact is that having and you, you know have the ability to actually tie this to specific goals that you might have. So not just engagement, but all the way down to the revenue. So any of the metrics that you are capturing in analytics or if you’re just using Target, any of the goals that you want to capture directly in Target. And we are rebuilding those models on a regular basis. This is really kind of balancing what we think of as balancing exploration with exploitation. So it’s serving up the offers in a random way to a control group and it’s then also targeting individuals where we have information, you know the right kind of information to target them and we’re able to see the impact that we are having with that sort of activity. And that’s reported directly back into the platform. And this is one way of, because we have that kind of balance and we have multiple different ways in which we’re modeling we are able to take into account changes in the general population behavior as well as changes at the individual level that might have an impact on how someone is likely to respond. So that makes it a very sort of flexible system. So that’s the kind of the four those are the four main flavors of automation and personalization that are available within the Target platform.