Lookalike Models: Determine which subscribers are likely to churn

Retaining customers is more important now than ever, and for businesses that have a subscription model, reaching out to users who are risk of churning is an effective way to improve retention and customer engagement.

Define Source and Target Audiences

The first step to building a Lookalike Model is selecting your source and target audiences. The target audience in this case includes subscribers who are at risk of churning. You can define this based on the users' last engagement time or some some other measure of engagement. The source audience will include current subscribers, which you can define using subscription field(s) from your imported data.

Build Lookalike Model

In the Lytics UI, head to the Laboratory section and click Create New Model at the top right. On the configuration page, you'll select the source and target audiences that you previously defined.

By default, the Auto Tune option is turned off. If you’re not getting enough performance out of your models, try Auto Tune, which will select the default parameters for you. Select Model Training Only if you're in an exploratory phase and not ready to target users based on this model's predictions. Learn more about the configuration options in our Model Builder documentation.

Depending on the sizes of your source and target audiences, your Lookalike Model may take a few minutes up to a few hours to build. On your Model Dashboard, you'll see a "Building" status until the model is complete.

Create Predictive Audience

Once your model is built, view the Model Summary to determine if it's ready for use in your targeting. You must Activate a Lookalike Model before creating a Predictive Audience, which will evaluate users and write the model's prediction scores to user profiles. This process can take up to three days for large audiences.

On the summary page, click Create Predictive Audience in the Model Usage section, which will open the audience builder pre-populated with your model's predictions. The default model decision threshold of 0.5 is used, but you can adjust the decision threshold as needed to reach more users or be more accurate to only target users who are most likely to churn.


Next Steps

After creating a Predictive Audience, you may want to target this audience with emails to help them re-engage with your brand. For more information, check out our playbook on retaining customers.