Adobe Target Q & A Session

Trying to use Adobe Target for Personalization or Optimization? Unsure about your strategy or how to get started? Establishing an effective program can be daunting and confusing until you understand the basics of Adobe Target. Listen to Andrew Johns & Brett Crawford discuss some of the best practices and help organizations demystify what is needed to get started and create an effective program.

Brett:

Hello, everyone. We here at Softcrylic hope everyone is staying safe and well during these times. My name is Brett Crawford and I’m an analyst in the Data Science and Analytics practice at Softcrylic. I’m here with Andrew Johns, a senior consultant in our Data Activation practice.

Andrew:

Hey everyone. Yes, so thanks for the intro. I’ve been with Softcrylic for a couple of years now. And yeah, I focus a lot on personalization and optimization, and I lead that practice working with a lot of clients, helping them, you know, start the program from the ground up and enhancing their, you know, they’re currently running programs.

Brett:

That’s very interesting. Could you maybe help us demystify some of the lingo and tell us the difference between personalization and optimization?

Andrew:

Yeah, yeah, absolutely. So, yeah, those two terms get thrown around pretty much interchangeably, even though they’re related but different. So you can think of optimization kind of as the baseline that is trying to make the user experience better across, you know, every audience. You know, for every visitor. So that those are things like, you know, improving your overall site navigation or your mobile app design or the page layout, stuff like that, you know, something that would apply to pretty much everyone.

Personalization kind of stacks on top of that. It’s kind of like optimization for individual audiences all the way down to the individual level, you know, trying to make things as relevant as possible for each user that touches your site or app. And, you know, a lot of people don’t realize, or a lot of IT folks don’t realize is that your digital properties, they’re all marketing. Absolutely everything is marketing. And all the stuff we do with personalization and optimization just kind of builds upon that marketing strategy.

Brett:

That’s very interesting. What sort of benefits can personalization or optimization programs provide?

Andrew:

Well, I mean, basically the goal is to make more money for every single client, regardless of how they make money. The goal is to make money and also improve, you know, customer satisfaction, which in turn is to make money. So it’s really just a revenue generating thing. So it depends on the test and the client. So, for example, for an electronics company, you might want to personalize a phone trade-in sort of program like detecting that type of device that you’re on, exactly what the value of the trade-in of that device is.

And target and upsell to a new phone with the exact price that you’re going to get when you trade in. So that would be a way to get more people into the trade-in flow as opposed to just doing like a generic trading your phone sort of thing. So, you know, the benefits are all over the map and it completely varies by industry and by client. But, you know, ultimately, you just got to be creative and think of ways, a hypothesis around what you think is going to improve the user experience, to lead to more sales, more revenue, more whatever your goal metric is.

And, you know, if you’re on the optimization side, then test it out, see if that works or if it’s more on the personalization side that you might already know what works and you’re trying to get more of it, trying to be more focused. So, yeah, I mean, there are some clients that see, you know, orders of magnitude, you know, lift. You know, you could have two to 300 percent lift just yet.

So you can kind of see where this is going. And depending on the types of things that you dedicate resources to actually testing, it could be a big money maker and it could pay for the platform and the program itself.

Brett:

So I think you think you’ve sold me. Say I’m a client, though, and I want to start one of these programs. How do I start this and how do I get buy in from everyone?

Andrew:

Well, I mean, always start with low hanging fruit. I mean, if you’re just starting let’s say you just purchased a tool to do it or you’re trying to evaluate a tool and you know, you’re able to convince one of the sales guys to let you demo it as like, you know, for a short period of time, then always pick low hanging fruit, something that’s relatively easy to implement from a technical standpoint that you have a feeling might make a big difference and something that you can execute pretty quickly, and that also has a lot of traffic. So one thing, when you’re getting into this, you’re dealing with, you know, statistical significance and you want to make sure that whatever you’re testing has a significant amount of traffic to be able to reach significance relatively quickly if you’re trying to prove the value of it. So a really good example might be a like a header navigation change. Like, if you’re trying to get someone to open an account, you might want to try to change, you know, how the open account functionality is presented.

So, for example, you know, if it’s normally in like a submenu or just in plain text over, you know, over on the right side or something you might want to try a different treatment for that as like a button or something. That would be a good, very simple, potentially high impact first test. In fact, I mean, I did a test with a bank that did that and that was their first test, and it was the thing that sold the program to get resources to support it and ramp it up.

Brett:

Wow, what sort of resources would I need to start this program?

Andrew:

Well, to start it, you don’t really need much. It really just depends on the level of technical expertise you have, because regardless of what all the vendor sales folks will tell you, you know, whether you remember how wizi-wig it is and how many fun things you can do, by changing colors and what-not you’re going need development. You’re going to need at least a basic HTML, CSS and JavaScript understanding if you’re dealing with the Web in order to really do anything useful.

But that being said, you don’t really need it to start. I don’t think you need like a full stack developer or anything. I mean, you need, you know, I’ve been on teams where we started off our junior developers as folks that worked on testing and optimization working on some of it. The UX, you know, implementation inside of the tools. So you can get by without any technical resources as long as you have, you know, some familiarity with, you know, HTML and CSS to get by. But ultimately, you’re either going to want to lean on your existing front-end developers. You know, beg, borrow or steal some of their time to help you out until, you know, maybe you can get a dedicated resource on your team to be able to help out with, you know, this type of this type of work. And, you know, it really just kind of depends on your organizational structure and your model. You know, you don’t want to try to fit a square peg into a round hole. So if you’re more of a matrix or organization and you know, your developers are multifaceted, then, you know, maybe trying to fit the optimization or personalization program into that model would be a natural fit because you could use a portion of their time to help out with this. Whereas if you had a more traditional sort of structure, you might need a dedicated developer on your team to work side by side with the analysts and, you know, like a program manager to help, you know, keep the train on the tracks, managing what’s live all at once, keeping track of the metrics, report outs, stuff like that. So start small. Figure out what you need. Pick something that fits into your organizational structure and build it slowly. It will pay for itself over time, as long as you’re, you know, picking things to test that are valuable and you are adhering to good statistical principles.

Brett:

So how would I select the first test to run?

Andrew:

Well, I mean, I’d pick something easy, you know, like a header navigation test is always the first thing that I look toward. But another good thing to do is, you know, a lot of times executives and upper management, you know, they’ll have ideas of their own, figure out those what those ideas are implement it. That’s a great way to prove the value in this sort of program. So ask around, see what people’s ideas are that are floating around and execute them.

Brett:

That’s very sage advice. Now, say I have this momentum. How would I choose an optimization program?

Andrew:

You mean like a platform to use?

Brett:

Exactly.

Andrew:

So I would, I mean, there’s a dozen of them out there. Pick one that fits in your overall marketing stack. So, for example, if you use Adobe, which a lot of clients use Adobe Analytics, then a natural fit would be to use Adobe Target, which is their, you know, A-B testing, optimization, and personalization platform, recommendations, engine does a lot of things, but then that would be the best fit.

Whereas if you are already a Google shop, you know, they have their they have Google Optimize, that might be a good starting point. It’s not quite as robust as some of the others out there, but it might fit very well with, like, your Google Analytics implementation. So first place I would look is, you know, what are the tools you’re currently using in that stack and build off it from there and see if you know that vendor’s offering, ticks all the boxes that that you’re looking for.

Brett:

You mentioned Adobe Target. What are the relative advantages of Adobe Target versus other platforms?

Andrew:

Adobes definitely one of the most popular that we deal with. Target is one of those tools that, you know, quite frankly, a lot of people have a license to use, but they don’t use it. So it’s a highly underutilized tool. And some of the advantages of it; I mean, honestly, if you’re an Adobe Analytics customer, there are huge advantages because you can use analytics as the reporting source. That way you can slice your data however you want and however you’re used to looking at your Adobe analytics site or app data.

A lot of other solutions don’t really integrate that well with an existing analytics solution. And if they do, then it’s kind of like a custom sort of stitching thing. And it’s you know, it’s either kludgy, it doesn’t work very well, it’s not very accurate or is it difficult to maintain. I think it Adobe’s integration works really well with some of their other tools. They have some pretty cool machine learning stuff for, you know, auto-targeting, the best experience to a particular user based on things we know about them.

But it’s a pretty solid tool. Their visual editor is pretty good. It allows a lot of flexibility for injecting custom code to do pretty much whatever you want. A lot of people look at it and think that it’s limited when they just see the wizzy-wig, you know, oh, I can change colors. I can rearrange things on the page. That’s great. But how does it work in like a single page app? I mean, quite frankly, you can flex that tool to work in just about any scenario.

And, you know, it’s all API based, too.

So if you have server side use cases like you want to personalize a kiosk, you know, or like a Coke freestyle machine or whatever, you know, those are the types of use cases you can you can do. I think in the last summit, since it was all online, I think they had a Panera use case where they personalized the kiosk there with Target. So the sky’s the limit. It’s very flexible. You just got to make sure you are going back to the resource question.

You got to make sure you have the right people. Both are there managing the tool from like a product management standpoint to make sure you’re keeping it up to date and that the implementation is sound and that the integrations are working. And also, from a day to day technical standpoint that you have the right people to actually implement various tests and, you know, personalization tactics and tool.

Brett:

It sounds very powerful. The integration aspect of it sounds very interesting. Would you mind elaborating on how it integrates with other products in the Adobe Experience Cloud?

Andrew:

Yeah, sure. A lot of it’s kind of, in terms of Adobe, you know, we do a lot with Adobe. So with Adobe, a lot of it is just controlled by the Adobe visitor I.D. service or the ECID service. That’s kind of the hub that does all the stitching, you know, to make sure that the analytics and the target calls are stitched together, that the visitor I.D. between all solutions is consistent. There are other ways in the Adobe stack to get data in.

You know, you can use something called customer attributes where you can upload like a flat CSV file of customer I.D. and, you know, non-personally identifiable information about the user that you can use for targeting. So, for example, a really good use case for that that I don’t see enough people doing is doing your own off-line modeling like propensity scoring around the next best offer, next best action sort of thing and uploading that via customer attributes and using that for targeting, you know, on the Web.

I mean, that feature is free. Everyone can use it. Even the most basic license of Target gives you the ability to use at least a couple of customer attributes fields that you can use for, you know, something like propensity scoring. So that is a big one. I mean, it takes a little bit of time, you know, but if you have your own in-house data scientists or you work with, you know, a company that does data science sort of work like we do, then, yeah, that’s a great use case. And that’s it. A good simple integration with offline data that is really powerful.

Brett:

That sounds incredibly powerful. Based on your experience, what are some pitfalls to avoid in establishing a personalization or optimization program?

Andrew:

Yeah. Everyone runs into these, you know, from the ground up. Number one is staffing the wrong people, you know, and getting off on the wrong foot. That’s, you know, hiring the wrong kinds of resources for the work that you need to do. You want to avoid that. I mean be open and honest about the type of people you need to make this effective. If you hire people that are not technically savvy enough to execute, you’re going to be very limited in what you can do.

Another thing to look out for is you’ve got to have a statistical mindset. You know, everything we do here is based on statistical significance. Is it actually a lift? Was it just noise? You know, were there some other factors? You have to have those types of people involved when you’re analyzing it. Alot of the tools kind of tell you that, you know, upfront. OK, this has X amount of lift in this amount and X significance, but you can’t just ignore that stuff, you know, and you can’t let that gut, you know, let your gut instinct get the best of you.

You know, when something is losing, you have to just kind of quit. You know, every single test you do is a learning experience and you have to look at it that way. You can’t look at like the first test where, like, you do the control versus the challenger that loses. Well, not every idea that you have is going to be better than what’s currently out there. So you can’t get discouraged. So keeping a positive attitude around that the whole the whole program’s about learning and optimizing, but not everything you do is going to be a winner.

Having that mindset, having the right people at the helm and having a good foundation, you know, having a good technical foundation for it to be set up properly. So you don’t have to do a bunch of hacks and workarounds just to, you know, just to make a test work. So those are probably the biggest ones. And then I guess probably more not from like an execution and technical standpoint, but more, you know, the organizational standpoint.

You know, you’ve got to have buy-in on this. You can’t really kind of push it from the bottom up. You need some senior leadership support and being able to, you know, make it successful. Ultimately, you’re going to need to prove the value of the program, both from a licensing cost for the tool to help you execute as well as any resources you need or anything like that. So, you know, having support from leadership is almost a necessity.

So, yeah. But if you have those things, then you can kind of evolve it over time and it will naturally become successful if you have all the right pieces. It just might take a little bit time to figure out exactly how they fit together. For example, the organizational structure thing is usually a big one. You know, how do we form this team? Where do they sit? Who do they work with? You know? What’s the reporting structure?

You know, the last thing you want to do is form a team that feels like they’re coming in and competing with other resources or anything like that. So finding a good way to merge that in with the rest of your overall workflow. Once you do that you’re kind of off to the races.

Brett:

Well, this has been incredibly helpful. Maybe in the future we can speak again and go a little deeper on the specific capabilities of Adobe Target.

Andrew:

Yeah, sure. Yeah that is one of the most popular ones. So a deep dive into that will be fun.

Brett:

I would very much look forward to it. I can’t thank you enough for your time Andrew. And we here at Softcrylic hope everyone continues to do well during these times. Thank you.