Plans & Pricing
Does My Company Subscribe?
David JakubowskiDirector, Product Marketing for AdtechFacebook
Erik Johnson(not pictured)Head of AtlasFacebook
Atlas is an ad-serving service that Facebook acquired from Microsoft and then rebuilt from the ground up. Atlas lets advertisers target ads to people based on data that Facebook gathers, including your Facebook activity and your web-browsing activity. Facebook associates all of this activity with a unique Facebook ID. Advertisers can track ad performance across devices and publishers without having to rely on cookies. eMarketer spoke with Facebook executives David Jakubowski and Erik Johnson about how it works and what might be in store for the future.
eMarketer: How has the rollout of Atlas been going?
David Jakubowski: At Adweek, we announced that we are pushing the market into the next paradigm of “people-based marketing.” We’ve gone out to all six agency holding companies, and we are in the process of cutting those strategic deals. The one that’s public is Omnicom, and we have been onboarding some of their clients.
eMarketer: How does Atlas work, and what makes it different?
Jakubowski: There are two primary functions. One is, where do I house my creative so that when the ad calls happen, the ads flow in the right sequence and they get targeted to the right groups?
The second is ad measurement. That’s where the people-based marketing really begins to shine. You get to see how many people you actually reached vs. how many cookies did I reach.
eMarketer: What’s the benefit of taking a people-based marketing approach?
Erik Johnson: People are using multiple devices, and they’re beginning transactions or tasks on one device and finishing on something else. The technology that we use to measure on a mobile device is a cookie, and a cookie is fundamentally flawed. Atlas solves cross-device because Facebook is being used by 1.3 billion people, and if people are using Facebook on one device, they’re very likely using it on all of their devices.
We’ve got the demographic information to allow people to build targeting segments that are relevant, allow for better targeting demographically and allow for better tracking and conversions across devices to make sure you’re not serving ads to people who have already seen them.
Increasingly, merchants are allowing you to leave an email address or a phone number and have your receipt sent to you rather than giving you a paper copy. If that email address is the same email address you use when you’re working with Facebook, we can tie those things together. That will solve the offline-to-online problem.
We think we’re addressing—I won’t say solving since that sounds a bit arrogant—but we’re addressing what we see as the three largest problems in the mobile marketing world.
eMarketer: Can you describe in more detail how you are working on the online-offline problem?
Johnson: An advertiser can use any CRM system they want. If they’re collecting information on their customers that includes a form of identification that we can match using the same technology we use in Custom Audiences on Facebook, then you would be able to upload that [information] into Atlas.
We’re still working on some aspects of that platform, and not every merchant collects that information. I don’t want to overstate that we’ve solved offline to online in every single instance, but it does seem to me that, based on my own personal experience and the people I’ve talked to, people are sharing more of that information.
eMarketer: You are running Atlas as a separate company within Facebook. Why?
Johnson: We believe that we need to be able to talk to our advertisers and agency partners about a separation between Facebook the publisher and Atlas the service. The Facebook sales team and the Atlas sales team are independent from each other, and data that we show you in Atlas will always be accurate data; we won’t do anything to make Facebook look better or worse.
What we use in Atlas should be independent and as close as the third party as you can get. Our agreement also says that any data that sits in Atlas cannot flow back to Facebook. The data that Facebook has on its 1.3 billion users is data that we can use in Atlas, but any data that we collect or use does not flow back in the other direction.
eMarketer: What are some of the biggest things you have learned so far in the rollout?
Jakubowski: The No. 1 thing that surprised me was that when you shift to people-based marketing, you see that there are people that you have served ads to who have bought something, either in your store or on your website, that you weren’t giving yourself credit for.
The reason we don’t report it like that today is that the existing infrastructure that we’ve been built on has been focused around cookies, and it doesn’t really account for mobile at all. The people-based measurement solution solves that.
eMarketer: How do the CPMs for people-based marketing compare with other CPMs?
Jakubowski: People-based CPMs are generally better—in some cases higher, others lower, but always reflective of the true value—for a whole host of obvious reasons. They perform better; the reach and frequency can be managed better. The cookie-based solutions don’t have that same confidence because of the inherent errors that everybody has come to live with. There is a little bit of guesswork in there.
eMarketer: With regard to the ad measurement function of Atlas, what are its attribution measurement capabilities?
Johnson: We have two attribution levels. The first is last click—everyone has a last-click model. The second is a multitouch attribution model, which is an even-credit model that allows advertisers to get a better sense of how all their ads are actually performing.
It takes into account any place an ad is served. So we could be looking at cross-device or cross-publisher interactions. Any instance of a served ad can be tied back to a person in our even-credit model.
Again, we want to shed light on how ads are actually performing. Cookies are the standard on desktop, and we know they’re fairly inaccurate for things like reach and frequency, so the message we’re trying to land with Atlas is that, “Hey, you can actually measure all of this in a much higher fidelity.”
Moving forward, we hope to expand the capabilities of the multitouch attribution model, but we want to make sure people understand the one we have first before we introduce more complicated models. Building additional multitouch attribution models is high on our list of priority features.
eMarketer: How will people-based marketing evolve in the future?
Jakubowski: When you start to put the pieces together, you see the shape of the next trend, which is personalization. When you look at what’s happening in mobile and what’s happening in video and what Facebook has had a tremendous amount of success with, it’s personalization at scale.
When we talk to advertisers and publishers, we hear a lot of requests for personalization. For example, I am a Food & Wine person—that’s a magazine that I still subscribe to. But when I get into the app, there are certain things about food and wine that I gravitate toward. I expect them to start updating that content so that it’s actually different than the print magazine and it’s more tailored toward things I’m interested in.
Learn more about eMarketer data and insights »
You've never experienced research like this.
Nearly all Fortune 500 companies rely on us.
Inquire about corporate subscriptions today.