Use look-alike models to extend sold out inventory from first-party data

Last update: March 21, 2022

CREATED FOR:

  • Intermediate
  • User
  • Developer
  • Admin
  • Leader

In this tutorial, we will walk through the steps you should take to set up and use Look-Alike Models, so that you can create new look-alike audiences, and sell them as an extension to your conversion segment.

Use-case details

You are a content publisher. If you have already sold out inventory for converters on your site, you might think that your opportunity ends there. Enter AAM’s Look-Alike Models. By using this feature, you can extend sold out inventory further, and also sell audiences of people who maybe haven’t converted yet, but who look/act like people who have converted. This audience segment would typically sell for less than the actual converters, but nevertheless allows you to add to your bottom line by providing an additional audience option for Advertisers who want to place ads on your site. The additional benefit of this use case is that it doesn’t cost you anything to run this model on your first party data.

The steps in this tutorial will be as follows:

  1. Identify/Create an ideal user (conversion) trait or segment
  2. Create a model using this conversion trait/segment as the base item
  3. Choose First party data source(s) in the model and run the model
  4. Create an Algorithmic Trait from the model results and add the trait to a segment
  5. Offer the segment to interested Advertisers to extend the conversion segment sales

Identify or create an ideal user (conversion) trait or segment

What are you trying to get people to DO on your site? What is your conversion event? Of course, there are many different answers to this question, depending on your site type/vertical and your organizational goals. In any case, it is common in AAM to create a trait for visitors who have met those criteria.

In this use case, this is already assumed, because you have sold out the inventory for people who are converters. However, for the purpose of this tutorial, it is good to discuss it as reference for the rest of the use case.

Also, when using events to create traits, there is a major gotcha that you need to keep in mind, so that you don’t collect more users than you should into the trait. Watch the following video for the big reveal. :)

NOTE: In the video above, the example that I show assumes that you have Adobe Analytics. Obviously, this may not be the case. If you have Google Analytics (GA), we have a module that you can use to send data into AAM (see the documentation), and if your conversion activity on your site is sent to AAM by GA, then you can create your conversion trait from that. If you have a different analytics solution (or no analytics solution), you can still send in data to AAM via our DIL code and the submit function, etc. (see the documentation). Then, again, create the conversion trait based on the data sent when the conversion activity is performed on the site.