In this recipe, we’ll explore how Twilio Segment’s new Predictive Traits feature (Currently in Beta) empowers FinTech companies to anticipate their customers' readiness to buy. With Predictive Traits, you can unlock insights into customer behavior, fine-tune your marketing efforts, and maximize revenue opportunities.
Whether you're a FinTech professional seeking to optimize sales processes or a data-driven marketer aiming to improve targeting strategies, this recipe will provide you with practical knowledge and step-by-step instructions to harness the potential of Twilio Segment. Are you ready? Let's dive in.
Prerequisites
To follow this recipe step by step, you will need the following:
A Twilio Segment account (sign up here for free). Segment is Twilio’s customer data platform.
Access to Twilio Engage, which is a data-first multichannel marketing solution offered as a product add-on for Segment. It is available for Business Plan users. Contact your account representative to enable access to Engage.
Access to Predictive Traits, which are a type of Computed Trait used for segmenting audiences in Engage. Predictive Traits is currently in beta. Request an invite by going to Engage -> Audiences -> Computed Traits. Select Predictive Traits and click the Request Demo button.
Data Setup
For this recipe demo, we’re working with a sample dataset of customers, products, and customer orders. For each customer, we have generated a set of events of the type “Purchased <product name>”. Each event represents a time when a customer purchased one of our products.
To load the customer data in Segment, we created an HTTP Tracking API source. Then, we made Identify calls to Segment in order to capture basic customer information. Segment can use this information to associate any future actions with customers based on customer ID or email. Then, we made Track calls to associate “Purchased <product name>” events (one per order) with our customers.
For this recipe, we want Segment to predict which customers are most likely to purchase “Special Product D”. To do this, we’ll build leverage AI/ML through a Predictive Trait.
Build a Predictive Trait
If this is your first time working with Twilio Engage, you’ll first need to set up Unify. You can find detailed setup instructions for Unify here. Once you’re set up, click on Engage in the navbar on the left.
Navigate to the Audiences page. Then, click Computed Traits and then Create Computed Trait. Finally, we select Predictive Traits.
For this recipe, we are using the “Likelihood to purchase” option.
From there, you can select destinations where you want to sync the predictive trait. For now, we don’t need to select any, so we click Next.
In our final step, we name our Predictive Trait and provide a description. Then, we click Create Trait.
From there, Segment takes you to the page for your newly created Predictive Trait. The Status will display In Progress. Segment needs some time to complete the initial calculations for this trait. For larger datasets, this could be up to 48 hours. For smaller datasets, you can expect to see results in less than an hour.
Data Requirements for Successful Predictions
As with most AI/ML predictive modeling, the effectiveness of predictions depends heavily on the size and quality of the data. Along these lines, Segment states the following in its documentation:
Segment recommends that you make predictions for at least 50,000 users and choose a target event that at least 5,000 users have performed in the last 30 days.
In fact, because the dataset we used for demonstration purposes in this recipe was somewhat small, we encountered the following warning when setting up our Predictive Trait:
If you encounter this warning, or if the results of trait computation fail or are not as you would expect, we would recommend that you review the dataset you used to make sure it meets requirements.
Understanding Predictive Traits
After initial trait computation is finished, navigate to the Prediction tab. You’ll see the Explore your prediction section.This allows you to visualize and browse your prediction data. From here, you can create custom Audiences from the group for further use, as well as send this data to Destinations.
To see more information about the quality of your prediction, navigate to the Understand your prediction section. Here, you’ll see statistics related to your predictive model.
Area Under the ROC Curve (AUC): Measures the model's ability to distinguish between positive and negative instances. AUC is useful for comparing different models or variations of the same model.
Lift quality: Evaluates the effectiveness of a model in generating positive outcomes compared to random chance.
Log loss: Measures the accuracy of a model's predicted probabilities, providing a more nuanced assessment of the model's calibration and probability predictions.
These metrics will help you better evaluate the effectiveness and performance of your predictive model.
Conclusion
In this recipe, we learned how to harness the power of Twilio Segment to predict a customer’s readiness to buy. We walked through the basic steps for setting up Twilio Segment, creating a Predictive Trait in Twilio Engage, and understanding the information which comes from the Predictive Trait.
By utilizing the insights derived from Twilio Segment, financial institutions can enhance their targeting efforts, engage with customers at the right moment, and drive revenue growth.
With this foundation, you should have a solid understanding of how Twilio Segment's AI capabilities can assist you in anticipating customer readiness to buy. Now, you can embark on your own journey to leverage predictive analytics and deliver personalized experiences to your customers in the ever-evolving FinTech landscape.
For more information on Predictive Traits and ideas on how to integrate them into your workflow, see the documentation or schedule a demo today.
Getting started is easy
Start connecting your data with Segment.