After reviewing the results of your prediction, it's time to create one or more user segments of the users you want to engage. You can start by asking yourself a few questions:
Which users are most likely to help you achieve the outcomes you seek, and how can you encourage them to do so?
What about users who are perhaps less likely to convert, but still more likely than average? Do you want to do anything to motivate them?
Are there any surprises in your prediction that might impact existing campaigns or journeys?
In this article
Select users from your prediction
On the interactive chart on the Goal Details page, scale the range of the prediction to define a user segment that you can use in a contact (or segmentation) list. Use the values reported to the right to get a sense of how your selection impacts the results you might expect.
How you shape your user segment depends on how you intend to use it. You might consider how some of these metrics may influence the segment you create:
Audience size - Are you looking to reach a large audience with broad messaging, or a smaller, more exclusive audience with personalized messaging? As you size your user segment, pay attention to the number of selected contacts.
Conversion likelihood - Do you want to reach users with a higher than average likelihood of conversion, or users who need to be incentivized to convert? Depending on your answer, your segment should have either a higher or lower "more likely to convert" value.
Number of likely conversions - Will the success of your campaign be evaluated based on the number of conversions it produces? If so, you might be inclined to focus on a large user segment, equating a larger audience to more conversions. Instead, however, try creating more than one user segment from a single prediction so you can tailor messaging to users of varying degrees of conversion likelihood. For example, you might create a segment of people who are very likely to convert, and another for those who might need a little more encouragement.
There are, of course, lots of other things to consider when defining user segments. Chat with the people on your team who are familiar with your audiences and the types of data you should track to be sure you reach the right people.
To learn how a sample brand defined their user segments, see Use Case: Reward Loyalty with Memorable Experiences.
Create a user segment and contact list
Once you've settled on a user segment, click View selection in segmentation. This will bring you to the Segmentation page, where you'll see a range of probability values prepopulated. These are the top and bottom the values for the current selection from the y-axis in the Prediction chart. You'll also see details about the users in your selection under Results. To review the details, filter the fields shown in the list to see how the data stacks up.
To save the selected segment, click either Save list (to save as a dynamic or static list) or Export as (to export to a CSV file or Facebook custom audience).
View a user's conversion scores
A quick audit of individual users' percentile and probability scores can help you verify that the data points in a selected segment are within the range you specified in segmentation.
If you don't see these scores in the Segmentation view, go to
Audience > Contact Lookup and search for a user, or click a user's
address from the Segmentation page and look for
predictiveGoals under the
itblDS: Object field.
Under the user property field that matches the contact property name you
provided when you created the predictive goal (in our example,
HighlyEngagedPropensity), you’ll see a value for
percentile, which is where
in the selected range this user falls, and a value for
represents the likelihood that the user will convert on the goal.
Exclude users from segmentation
There might be times when a user segment includes people who've already satisfied the criteria you're evaluating for conversion. In most cases, you'd want to exclude those users from the selected segment so they don't receive messaging that encourages an action they've already taken.
For example, if a predictive goal searches for users who are likely to subscribe to a service in the next 30 days, you may want to exclude, from the resulting segment, users who've already subscribed.
From the Goal Details page, select the range of users you want to include in your segment.
Click View selection in segmentation.
On the Segmentation page, add an "And" statement that ensures that the segment includes None of the users who have a property that indicates they've already converted. In our example, we specify that the segment can't include users who have a user property that indicates they've subscribed already.
Want to learn more?
If you're looking for some ideas on other ways to use segmentation, check out Popular Segmentation Searches.
Please sign in to leave a comment.