Behavioral optimization is the logical extension of behavioral targeting, which itself was an evolution of a practice initiated from direct mail. As behavioral targeting became more sophisticated online, it became possible to build audience models to understand the value of online behaviors for specific advertisers, but the application of these models was still the same – targeting. These links back to direct mail have restricted the progress of performance advertising, where optimization provides an opportunity for further, more sophisticated, applications of behavioral data.
Optimization is like playing poker; one needs to understand the value of all the cards and hands in the game in order to effectively react to what the other players are doing. Targeting on its own is like deciding the best cards in your hand, but then playing the game with your eyes shut. Optimization is the smart step to win the game.
Advertisers who run performance activity online need to ensure that their partners are applying audience data in the most sophisticated way possible and aren’t blindly showing their ads to users.
A look at traditional behavioral targeting
Behavioral targeting is a strategy most marketers are comfortable and familiar with. It is possible for campaigns to harness behavioral targeting, even without any data-driven attempt to attribute value to users. All that is necessary is that the client manually identifies the behaviorally defined audiences that might be particularly effective for them.
A more scientific approach is to use an audience model that identifies the relative value of certain behaviors and the reach that they provide. It is then possible to determine a satisfactory balance in terms of the volume of users the campaign wishes to reach and value these users should possess.
In either case, when the campaign runs, the behavioral targeting stipulations that the client defines will dictate that some (or all, depending on the reach that the behaviors provide) of the campaign’s impressions will be served to users with the selected behaviors. However, there are two major limitations here. First, assuming the campaign is being optimized traditionally, the relative value of the behaviors in a user’s profile is ignored in the decision process that determines which ads get served to which users. So a client’s ads are no more likely to be served to higher value users than any other user, as long as the users have behaviors in their profile that match the targeting settings – a particularly significant factor in performance advertising.
Second, when you exclude users with non-matching behaviors from your campaign, you are stating that these users have no value despite being aware of the possibility that any user has the potential to be valuable sometime in the future.
Behavioral targeting versus behavioral optimization
What distinguishes behavioral optimization is that the relative value of a user, identified by an audience model, is factored into the algorithm that determines which ad gets served.
Since the relative value of each behavior in a user’s profile is an input to a behaviorally-focused optimization algorithm, clients are more likely to be matched with high-value users. . The output of the algorithm is an indexed score for the user for each eligible client, with the client with the highest score winning the right to serve their ad to the user.
With behavioral optimization, every behavior for which it was possible to reliably determine its value will be included in the audience model. This is in contrast to behavioral targeting where (as described above) often the model is cut to dictate the minimum value of a behavior that should be included in the targeting. In short, behavioral optimization provides extended opportunities for the relative value of users to be considered. For instance, with behavioral optimization, it is possible that at a certain time, a user is adjudged to be twice as valuable as the average user, and may be the best user for a client, but this opportunity would have been missed had the campaign been targeted only to users who are perceived as being at least four times more valuable than the average user.
Performance display campaigns: a renewed look at prospecting and optimizing efficiently
So, how should these related, yet contrasting, applications of behavioral data be harnessed?
Where the goal of a digital campaign is to reach a certain audience, behavioral targeting has long served as the perfect solution, ensuring that your creative assets appear in front of the most appropriate audience.
But for performance advertising where the objective is more specific, perhaps a CPA goal or an ideal AOV (Average Order Value), then the prospecting activity needs to be optimized to this goal and on a behavioral basis. When prospecting, the optimization ensures that the client’s budget is spent as efficiently as possible. Therefore, each time an impression becomes available, the question we should ask is: How suited is this user for the client? And this can only be done by factoring a user’s value into the ad decisioning process.
Behavior optimization solution
Despite the theoretical advantages that behavioral optimization has over targeting, optimizing a campaign behaviorally is not as straightforward as one might hope – an effective approach here requires audience data with the necessary breadth and depth, an open and transparent audience modeling methodology, and the necessary processing power to harness these the audience models to optimization.
The ideal behavior optimization solution must be able to specifically reach your intended audience in order to deliver optimal results. By using transparent audience models, and applying them during the ad decisioning process, advertisers are empowered to reach audiences who are most likely to convert.