David DeVisser - Stop Analyzing Devices—Start Understanding the People

November 23, 2016

Slide 1:

I’m really excited to be here and am really excited about this format for this seminar. Today, the subject of people based marketing is really what has me truly excited.

Slide 2:

This is a challenge in the marketing space, be it analytics or other disciplines because devices don’t buy products, people do. Adobe has been working on a technology called the “Marketing Cloud Device Co-op” and what it does, is it brings people-centric data into the marketing cloud solutions. So let’s take a step back and understand why this is such a challenge and how we can help solve this problem together.

Slide 3:

To do that, we need to take a step back and look at the consumer. The consumer that is interacting with our brands. This is of course who we are trying to satisfy with our marketing and our messages. But the consumer is not just an individual. This is an individual who has many states in his or her day. With these different states, there are different activities, different devices, and really just different context that might change how she’s interacting with our brand. That absolutely is about mobile versus desktop, various browsers within desktop. With that context, is the propensity to authenticate with a set brand. Whether that is a stored authentication or not. In the last image there, she’s in the airport. She might be short on time so she might just skip over authentication. This creates challenges for the marketer.

Slide 4:

The expectations though for the consumer are strong these days. Consumers are using multiple devices throughout their day and week. And their expectations are very high for material and communication from brands to be tailored to them and synchronized. The shopping experiences should also be tailored in addition to the advertising.

Slide 5:

When we look at these four different devices for the said consumer, when we look at the back-end systems, all of these devices–and these are just hardware devices we’re looking at—all are really silos of data on the back end. So they have their own reporting and personalization and segmentation and advertising for each of these said devices. This looks like, in this simple case right here, this looks like four different people, when in reality it’s just one.

Slide 6:

So let’s double-click on that a little bit further. The challenges imposed by basically current devices-centric systems span marketing discipline across reporting, targeting, advertising, and segmentation. When you’re looking at reporting data that is device based, or unique visitor based, you’re seeing fragmented conversion paths. You might have a successful conversion on one device, and a failure on another. Which is the right metric to track upon? So all of the reporting is really not understanding the person. Targeting or personalization is really around the inconsistency across those devices. If you learn about some activity on one device, and they continue on another, that’s where those expectations come in. The consumer would expect that personalization to continue on that second device rather than having to start over and teach the system of their desired action with the brand.

In the case of advertising, this is really getting into dollars and cents side of things. Where we need to understand the attribution path and really the value and accuracy of our campaigns. And then lastly, segmentation. This is when you’re very data-centric on your marketing discipline. Using DMPs, data management platforms, etc. If you are gathering traits around individual devices and not across multiple devices, you have a limited view. Therefore segment qualification or disqualification is less accurate than if you’re looking at a more holistic approach.

Slide 7:

So now let’s look at how Adobe aims to solve this problem. We are using a cooperative approach that uses multiple brands to have a better understanding of how people interact with said brands. That’s why it’s called “device co-op” or “device cooperative.” If we take an example of just two brands: a retailer and a travel site. When a consumer authenticates with the travel site, you’re telling the travel site who that person is. When they then authenticate on another device, the travel site can understand the linkage between those two devices. Therefore, treat these two devices and understand that they are from the same person. They might have another device—like in that use case, when she was at the airport—that they did not authenticate. There’s all this activity on the travel site, but they understand these three devices as really, in this case, two visitors because two are authenticated and one is not.

The idea with the co-op is to take that knowledge that the travel site has, and that authentication across multiple devices, and share it, in a secure way that protects privacy, with other brands. Now when this retailer sees one login, but not a login within this other device, because the co-op understands those linkages, that link can be shared between the co-op and the retailer. So now the retailer can see all three of these devices belong to the same person. They don’t know that it is Kara, for example, they just know that it is the same person, which creates huge value for the various marketing discipline challenges that we described.

Slide 8:

Now let’s zoom out again and look at what the overall graph looks like. The brand participation in a cooperative model—of course there’s brands of all different sizes–so that participation is all combined together and the Adobe device co-op combines all that information and pulls out all of the connected devices. Then each brand gets their relative share of the connections for those devices, i.e. people that are interacting with their brand.

Slide 9:

To be clear, data that is not collected by the device co-op is anything that is personally identifiable. Here’s a list: name, address, behavioral data, any information that identifies people is not gathered. We are only gathering device linkage information. Adobe sees a hashed ID for this information, but only the links are then shared within the co-op members because we are very concerned about privacy at Adobe.

Slide 10:

We are putting the Adobe brand in front of this and actually created a privacy center and built this from the ground up, even before we wrote a line of code, we built this around privacy-centric means because this is about consumers interacting with our brand. The other piece is, we’re protecting the member brands. Our customers, Adobe customers are protected from data going into the system. We are not sharing information that’s going to create a competitive imbalance and so forth. We are only focused on sharing those device links. We are focused on the US and Canada markets for the first phase of deployment and we will expand into other regions, likely next year.

Now let’s take it back to the consumer side of things. First of all, we want to be transparent with this technology. We want to tell consumers what we are doing, and be very clear on the impact and value of cross-device technology. So we will actually disclose the brands as well as the devices that we understand that belong to you in this tool we’ve created. And these are requirements, this just a nice-to-have thing that Adobe is doing, these are requirements for co-op membership. What this looks like for the consumer is an Adobe-hosted site. This is the site that co-op members link to.

Slide 11:

It’s an education site and the first panel is link devices. That actually will show a list of devices that are associated with this said individual

Slide 12:

Again, it’s not showing their name or what you even call the said device, it is just that generic information that we understand about these devices. If you click on one of these devices, you can choose to exclude them from the linkage. For example, if it’s a shared tablet that’s at home, we can actually exclude that from linking so it is not associated with any said person at the home. The other piece is the disclosure of members.

Slide 13:

We have been working very hard with our premium brands at Adobe for early membership. And by the way, we’re currently in production, but we’re in a beta state for the device co-op. So I’m very excited to share this new technology that we’re about to go into release as well as offer an early access program for companies. But as you can see already, we’ve really primed the pump with some big brands that are helping us already understand people interacting with our brands.

Slide 14:

In partnership with the actual device graph technology is, of course, the solutions that are going to leverage the technology. We’ve done that with four solutions at Adobe: analytics, target, audience manager, and media optimizer. So each of these solutions have chosen to leverage the data source for whatever makes sense for their solution.

Slide 15:

Let’s jump right in and look at analytics. This is of course the number one use case. In fact, the request from customers came from analytics customers where they wanted this value. Let’s double-click on that.

Slide 16:

We showed an example with a consumer with four devices. But the reality of it goes beyond just the physical devices. You can see, for example, the Chrome on the laptop, Safari on the laptop—that is the same laptop, but to two analytics or other solutions, that would look like two separate users. Just because there’s two separate browsers on the same device. That creates further challenges. When you think about the average devices per person, you might have seen numbers exceeding seven per consumer in the US market. That’s because it’s the instances of applications on these devices that are really expanding, not necessarily just physical devices. In any case, this is a big challenge and it’s a growing challenge.

Slide 17:

When you look at analytics today, you’re seeing metrics based around unique visitors. Which is a valuable metric. This was a big value when unique visitors was introduced; to help us collapse repeat visitors and so forth, to your brand. We want to move from this. Really, unique visitors is actually a device view. Probably the correct term for unique visitors is a device user or device period.

Slide 18:

We want to move from device-centric to a person-centric view so we can have a much better and more accurate understanding of the customer journey through our brands.

Slide 19:

That looks like true audience size, for example. Revenue per person instead of per device. And your true customer retention rate.

Slide 20:

So we take a step back and look at what this looks like for campaigns. If you look at two separate campaigns, one is a three dollar campaign that actually hit three devices. And another is a two dollar campaign that hit one device. One looks like a better value campaign because we got one dollar per device versus the two dollar campaign which is twice as expensive. But the reality is, if these three devices are actually for a single person, of course the tables have turned here and campaign one turns out to be the more expensive campaign versus campaign two as the better value. That’s one simple example just by reorganizing data and what the results there are.

Slide 21:

If we move to looking at people-based data—and I’ll just highlight this at the bottom here—unique visitors is at the top and people-based data is along the bottom. What Adobe Analytics has done, is taken the device co-op data and added a new metric called the “people metric.” You can apply this to any report and see we move from a 1.9 million uniques to 1.7 million uniques. That’s a more accurate audience size value. Number of orders per person, revenue per person, and visits per person. You can look at time spent metrics, things like this.

Slide 22:

Then it gets really interesting when the next phase of development in analytics is getting further resolution on the customer journey. What are the effects on various marketing channels? What content is consumed where and on which device for a said person? We look at the journey here, they view an ad on the phone, they continue searching for the same—travel in this case–on their laptop, and they end up viewing a video and buying on their desktop. So conversation occurs on that third device. Where of course, the desktop is getting the credit. Just the one device is getting all the value and attribution. But if we expand that with the device co-op, we know that this journey has spanned three devices. So we can now spread the attribution across the three devices so we can actually know that there was value in that ad on mobile, in that search spent on the desktop web.

Slide 23:

So what we want to do is really collapse that understand for any number of devices and here’s a real simple example of just those three. You remember how much more complicated this is when we look at many more than just three devices a user would have and multiple browsers and so forth.

Slide 24:

The team is also looking at visualizations for this as well. We have a report builder that lets us use this new technology for flow pathing. So we can actually see the number of users going through the various channels. With the device co-op data, collapsing this for a said person, we can actually see that in one step. Here’s just a simple report. Again, this is phase two they’re building in analytics. We’re very excited about this and what they’re able to do with this new data source and they continue to develop on this.

Slide 25:

Next up is personalization. I want to quickly review the other products that are using the technology. This is pretty straight-forward for everyone to understand.

Slide 26:

When you’re interacting with a brand, or on a said device, you continue that interaction on other devices. You really want that conversation to continue rather than start over. The device co-op enables that.

Slide 27:

And for the marketer, all they need do is, when they’re creating that a personalization activity, or editing one, they simply now are able to turn on, to use the device co-op data. That switches this activity, if we have that information about these devices, will be able to continue that personalization across those said devices.

Slide 28:

Further, on the reporting side, we can actually see that instead of device-based data, we’re actually seeing people-based data. We’re going to see that conversion rate that is much more accurate. Whether that conversion occurs on any of the devices that the co-op understands, we’ll actually see those results.

Slide 29:

And Media Optimizer is our advertising product. This is similar to the customer journey illustration for analytics.

Slide 30:

The attribution challenge here is when you’re seeing that mobile display impression, the search click, and finally, the conversion on that final machine. When you don’t have attribution across devices, all the credit and conversion data comes from a single device. With Media Optimizer and the device co-op, they’re able to see that attribution, and understand that attribution across the devices so you can properly adjusted your campaign as needed.

Slide 31:

Further, retargeting. This is around accurate spend on campaigns. If you’re interacting on one device, and in this case we’re on Adobe.com and our creative cloud pages, then you go off-site on a different device. We’re able to actually—if we have that device to link information—we’re actually able to target that user on the other device using the co-op data.

Slide 32:

Lastly, in Audience Manager, this is our DMP product. This is really around the data-centric marketer. I just want to quickly illustrate how this works with the device co-op data.

Slide 33:

If we look at how the DMPs work, if you’re not familiar, you’re basically gathering traits around individuals. Those traits will create segment qualification. So the combination of those traits are going to qualify for or disqualify for various segments based on all the various combinations of that data. Here we have an example of four different devices and we have two different segment qualifications. One selection of devices, there’s actually three—maybe because they’ve authenticated—and we have a fourth device that they have not authenticated, but there are traits on that device as well, so they’ve qualified for two very different segments.

Slide 34:

The combination of our DMP and device co-op allows us to actually combine the knowledge of those devices thereby combining all of those traits therefore seeing a completely different segment qualification.

Slide 35:

I hope that gives you a good overview of people-based marketing for the Adobe marketing cloud. We are very excited about what is capable in all of our marketing cloud solutions that use this technology and we’re happy to share more information.

Slide 36:

To get more information, please navigate to adobe.com/go/co-op or talk to your Adobe account manager. Thanks again for attending today, and now we’re going to jump to the concluding keynote. We’re excited to hear that content as well.

Previous Video
Peter O'Neill - Google Analytics Crash Course: 30 of the Best Tricks and Hacks
Peter O'Neill - Google Analytics Crash Course: 30 of the Best Tricks and Hacks

Learn simple tricks to improve data quality and extract actionable insights from Google Analytics.

Next Video
Lori McNeill - No Dev Resources? No Problem! 6 Technical Hacks for Analytics Practitioners
Lori McNeill - No Dev Resources? No Problem! 6 Technical Hacks for Analytics Practitioners

Lori will be walking through some "hacks" for analysts to increase self-sufficiency in getting the data the...