OMMA Behavioral Panel: Predictive Behavioral Targeting and Online Media Planning
From the agenda: The first stage of Behavioral Targeting was pretty simple: clickstream analysis, cookies and pixel-tracking. BT 2.0, as some call it, is much more complex. In this stage marketers are using their ISP and search-level data that tracks all consumer behavior, from clickstream, search to purchases to hypothesize what advertising might be relevant to the consumer. How can marketers stay ahead of the issue and how can they utilize these companies and their technologies to meet their client’s marketing objectives? Moderator: Cory Treffiletti, Catalyst: SF. Speakers: Eric Wheeler, 33Across; Bennett Zucker, aCerno; Christian Arens, Draft FCB; Frost Prioleau, Personifi; Derek Maxson, FrontPorch.
We talked about how the first version of BT is clickstream data. What is the second layer? What kind of datapoints are we looking at?
Frost: From our perspective two things: 1 is really looking at behavioral clickstream data in a detailed way. Instead of looking at a page and saying it’s about sports, understanding the data in a semantic way — it’s about tennis. The second thing is that it’s about hypertargeting and bundling in as many attributes as possible.
Eric: In the first stage of behavioral it was ‘not a page, a person.’ In the second stage, it’s that person and their network — that person’s social graph and their active engagement.
Bennett: At aCerno we look at the time honored direct response principle: what a person shops for and buys is the best indicator of what they’re likely to buy next. By collecting data from ecommerce sites we’re able to use clickstream data — cookies — to create more accurate segments.
Derek: One of the problems is that there are so many different kinds of data that people can bring to the table: demographic, purchase, social media, clickstreams, etc. I think that what BT 2.0 can be best summarized with is the concept that any kind of data that’s out there is being explored right now, because no one really knows what the payoff is going to be for one kind of data vs. another.
Christian: I would echo that. Depending on the advertiser you’re working with, one of these approaches is going to be the right one, but I want access to all of them.
Are all these definitions too big for people to work with yet?
Frost: A lot of media planners have a good idea of who their clients are and what they’re looking for.
Bennett: Compared with the early days of BT 1.0, which I experienced at Tacoda, there’s a great willingness to hear about new ways of doing this. The buying public is willing to give a good hearing to any effort designed specifically to help them target their audiences better.
Christian: You can go in there and say you have all the technology and we’re the best and you should use us, but where you add value is what you can provide to them which they can then take back to the client.
Eric: Going to the tactile nature of media planning — we’re going so much further in being able to show the media planner what we’re talking about. Here’s the user, here’s the social graph of this user, here’s how influence flows through the network. These are the early days — this is emergent data.
All the data we have — what do you really think the goal of this information is? Is it to resonate with the consumer in the longer term, or to drive a more immediate action they would have taken in the longer term?
Derek: It really begins with what the advertiser’s desire is, if their desire is to push a brand campaign. One of the things is the ability to purchase specific impressions. One of the things that’s incumbent on this industry is to provide an open model and that they can leverage the different values that are being offered by companies in our spaces.
Frost: One of the biggest uses right now is reach extensions (extending the reach of advertisers) — we’re trying to reach people who are interested in golf, maybe they’re not all on the golf pages — where else do you find them?
Bennett: We’ve got about 140 million US online shopper profiles over 90 days; using analytics, we’re able to know that 30 million of those are in market for something at any given moment. We know that today somebody may be in the market to buy a bigscreen TV but tomorrow they’re not. It’s a question of having the right data and then having the right tools to manipulate that data.
Eric: We’re moving from marketing to people to conversations with people, and that not only takes new tools to better understand that, but also answering questions like where does an app start and an ad end. The kind of things you can pull out of this environment are incredible — you can understand did they join a social group / did they start a social group / did they tag a page… are they actively engaged in that topic.
We work with a lot of social media companies, publishers, etc.; we’re building profiles across the graph, so we can actually see that people behave differently in different environments.
Frost: We encourage people to create their own database so that they can target those users on their own campaigns, but very few companies are doing that as yet.
Eric: Media agencies are just now getting smarter about empowering the clients to use that data to their advantage.
Christian: I’m struggling with this because if they’re pushing it off to the marketers and agencies to build this database then why are you sitting here? Sure, empower me with that, but then realize that you’ve just placed yourself out of what I’m going to do in the future. Sooner or later I’m going to reach these people who are int he mindset of I’m going to buy Pampers tomorrow. There is a lack of knowledge of what my campaign is doing and why, standard things like clickthrough rates, why one site did better than another, and it’s not just better content — is it better segmented or what? Communication goes beyond a media buy all the way to how I’m communicating in banners or anywhere.
Frost: Content does matter in many cases, depending on what the back-end metric is. Users are more likely to click or convert on ads that are on certain types of content than on others. We build the database and offer full and transparent access to it so our clients can see where their money is going.
Derek: Part of the difficulty is how many systems does an agency need to know how to run? That’s the biggest issue that encumbers us, and speaking as one of the many companies that aren’t names Google, Microsoft or Yahoo, it’s incumbent on us to create a model in which the agencies want to do business with us and we aren’t going to get squeezed out. It’s just so complicated today to be an agency.
Christian: One of the things we’re doing differently at Draft is that each person is responsible for all media, not just online. You’re exactly right, in some ways us agency folk are lazy, and we need the technology providers to come in and show us how to do things better. If I’m doing 20 things on a daily basis and you’re marking off one of them, then I’m going to be grateful for your help. I need to be able to play with all of you in the same realm but understand that what FrontPorch is telling me is the same as what Acerno is telling me. It needs to be standardized in some way. We’ve now taken the group that’s typically been in charge of target audiences and melded that with media planning.
Eric: What the budgets are, what kind of data they have, what’s interesting for them, and how they want to learn — if I’m showing somebody data they’ve never seen before, it’s people that are actually interested in getting past the 1.0 display/click rate conversation.
Frost: Our folks have a very flexible audience management system to respond to the need of the client, then they can say, “Here’s the segment you said you wanted; is it working?” Being able to slice and dice that audience.
Bennett: It’s important to spend time with the Head of Research — those are people that we have conversations with where we can get a better picture of what kinds of datapoints our clients are looking for and we can present something more meaningful.
Christian: Much like search, behavioral targeting should have an evergreen campaign, where we are continually optimizing and learning. Most of you have mentioned the RFP process, but Google has already gotten past the RFP process. How do you get to that point? How do you help us to get to a point where this isn’t a project-by-project campaign, but rather a thing that you do?
Eric: As you get to a larger network, where is the bulk of media sales happening? It’s in the RFP, so you need to be able to intercept that.
(KC: I’m really missing a lot of what these guys are saying. They’re sentence structures are like German, where they go for a while before you know what the verb is, and sometimes they mumble the verb so I miss the whole sentence!)
Bennett: We respond to RFPs all the time, and the first thing we do is call and make sure the RFP is applicable to us as a network. In the old days it was a similar situation where they weren’t selling BT and the advertisers weren’t buying, and it took a while before 5-10-15% of the buy was allocated to behavior. I don’t see behavioral targeting or any specific form of targeting being a line item the way search is.
