The amount of data that exists is exponentially increasing every year. Yet, companies (and marketers) aren’t always sure of the best way to use their collected data. Creating effective segments and personalizing messages can feel like an uphill battle. A bright spot in recent years has been the ability to use predictive modeling to create ‘smart segments‘. However, creating the foundation for these segments can still be challenging. This post will examine how Simon tackles predictive modeling and some predictive modeling use cases.
What is the Predictive Modeling Process?
You can’t know what will happen in the future, but you can try to predict it. Machine learning (ML) has amplified marketing technology to create, process, and validate new models to forecast outcomes. Still, these models are built on top of your unique customer data to ensure that the predicted results are specific to your business.
Step 1- Gather Your Data
Companies have a lot of data. Maybe even too much data. Most of this data lives across various siloed systems throughout an organization. CDPs like Simon Data help alleviate this challenge by collecting and consolidating all your customer data into one place. Then, Simon bridges these disparate pieces of data into a single customer view that marketers can access. Once your data is neatly in one place, you can begin orchestrating more effective cross-channel journeys that drive better engagement and ROI.
Step 2- Analyze Your Data
Data alone does not drive action. Marketers still need to figure out who, how, and when to engage with users across their customer lifecycle. Yet, with so much data, determining how to message different users effectively is still challenging. The process costs also marketers time and results in missed opportunities relevant to the customer’s journey.
Simon Predict speeds up the analysis of your data by giving you the crucial information needed to drive effective actions in a matter of minutes. This process is possible in the bespoke design of Predict’s three distinctive ML models- churn propensity, likelihood to purchase, and product recommendations. Additionally, each makes it easier for marketers to increase the value of their customers by improving their personalization. The goal is to create meaningful moments that strengthens your customer relationship. Below, we share some of the many predictive modeling use cases to show how each works.
The Churn Prediction Model
Simon Predict’s churn propensity model scores each customer from 1 to 100, rating them on their likelihood to disengage. Knowing which customers you are about to lose is invaluable information that helps marketers de-risk this likelihood. Imagine targeting customers that are likely to churn in retention campaigns, where they receive a unique promotion. Or, imagine omitting these high-risk users from certain communications, like regular emails that contribute to negative customer experiences.
Let’s pretend that you are an e-commerce pet supply business that runs on a subscription model for this use case. Your goal is to help increase the number of lifetime value (LTV) users by running a retention campaign. With Simon Predict’s churn propensity model, you see that several users are at risk for churning. Therefore, this is the right time to create a positive experience and convert them into satisfied, LTV users. To kickstart a more promising customer experience, you choose to run a segmented campaign. Then, this campaign sends users at risk for churning down a unique customer journey that gives them a discount. Yet, knowing what specific deals to offer varying levels of churn risk users is still hard to nail down.
By quickly running experiments within Simon’s journey management tool, you can easily test what types of promotions maximize engagement and ROI. Using Simon Predict and Simon Journeys together, you see that providing a 35% discount on auto-ship pet food to 40% churn risk users and a 20% discount on auto-ship pet food to 30% churn risk users gives the greatest return. Additionally, by encouraging users to enroll in auto-ship programs, you helped turn these users from being disengaged customers into regular buyers that guarantee purchases.
The Purchase Protection Model
Likelihood to Purchase
Like the previous use case example, Simon Predict’s likelihood to purchase model scores each customer on how close they are to making their next purchase. Simon helps marketers improve their odds of converting users along their customer journeys by providing marketers with this intel. Now, marketers access unique insights that influence more effective and personalized campaigns.
Let’s follow the same pet store in the example of Simon Predict’s churn propensity model. Now that the pet store has increased their number of LTV users, they want to increase their total revenue by improving their number of purchases. To help start the initiative, the marketers want to run a “dog days of summer” sale campaign that will target brand awareness and purchase propensity. First, users are segmented based on their purchase likelihood score. Then, we send users down different customer journey branches based on those scores. An example of these branches is scaling discounts. The more likely they are to buy a dog bone or a bag of catnip, the lower the discount. Additionally, this purchase prediction grouping allows businesses to drastically improve conversion rates while influencing positive customer experiences.
Understanding your customers’ likelihood to churn and purchase is vital information drastically influencing your marketing tactics. But, Simon Predict doesn’t stop there. Simon Predict offers a product recommendation model that helps marketers tailor their content with information unique to each customer experience to take personalization a step further.
Let’s round out our pet store’s “dog days of summer” sale campaign by maximizing the moment each user receives their scaling promotional email. In addition to offering unique discounts designed to increase their likelihood to purchase, you can populate campaigns with tailored product recommendations. For example, a customer is browsing a tug-of-war rope for their dog or abandons new cat food in their cart. These customers get a deal that corresponds to their likelihood to purchase. They also get a personalized message that shows products you know they’re interested in, further increasing your chances of turning interest into conversion.
Customizing Your Predictive Data Modeling
These predictive model use cases are only a few of the many ways we have seen clients successfully use Simon Predict. The three bespoke models provide vast opportunities for countless use cases based on your campaign goals. Still, in situations where the out-of-the-box capabilities don’t match your business needs, Simon Predict’s custom models option exists. This fourth option is built on your data and tailored to your unique use cases. Yet, when building out your models, the key things to remember are to be specific about your goals and set test groups to compare results. Simon Predict will help you do the rest. If you keep those things in mind, you’ll be seeing better engagement and ROI in no time!
To see a personalized demo on Simon Predict, request a demo today!