Modeling & Predictive Analytics

Forecast your future with predictive analytics

Wouldn’t we all like to see the future? Anyone who predicted and planned for the Covid-19 pandemic would have weathered the storm better than those of us who were surprised–they would have stocked up on toilet paper and hand sanitizer, for one thing!

One study found that 84% of marketers use predictive analytics in some way – the AI systems that drive these analytics can handle more data, utilize more nuanced information and work with more industries and niche audiences than ever before. Marketers know they need to harness this capability before their competition drives them off the road.

What is predictive analytics?

Predictive analytics models assess historical data to discover patterns, observe trends, and use that information to predict future trends.

It used to be that if you wanted to analyze the data–say, of weather patterns–you’d need to bring out paper charts and do the math yourself. But now with computer modeling, sensors all around the world automatically feed in data about wind speed, temperature, storms and more every day. The system takes this massive amount of data, runs the calculations, and shows an approximate prediction for weather events coming up. Predictive analytics doesn’t stop there– it pairs real-time weather data with historical customer behavior information and other contextual data to predict the best time to send out the right messages before a storm hits–whether that’s highlighting snow shovels at the hardware store or recommending insurance before hurricane season.

The Predictive Intelligence Benchmark Report found the average lift in conversion rate for sessions influenced by predictive intelligence to be 22.66%.1 It allows you to be nimble, make decisions based on real time data, and personalize your offerings to find the right audience each time.

What can you do with predictive analytics?

Savvy marketers know that predictive analytics can help you achieve a lower CPA, identify cross-sell opportunities, realize a higher LTV, and get your arms around program ROI. But what does it take to get there?

Predict future consumers

To lower CPA, predictive modeling helps you keep your costs low and still acquire quality customers. To do that we use systems like lead score models, CPA models, and look-alike models.

Lead score and CPA models both use scoring to raise certain perspectives or customers who have valuable or desirable traits to the top of your list so you can connect with them first.

With look-alike modeling, AI systems take attributes from your current best customers and goes to find new customers who match, or who are near matches. Maybe you have a whole untapped consumer base that looks just like your current customers–but they use a different social media platform than the one you’ve invested in. The AI is predicting which people will want what you offer so that you can reach them where they are and lower your overall CPA.

Predict and personalize consumer needs

Amazon’s front page shows you what you’ve recently purchased, what products you were recently searching for, and also products they think you might like. This kind of dashboard puts items the consumer is more likely to buy in front of them first. Which is more likely, that someone training for a marathon needs elite quality athletic shoes or bicycle wheels and accessories?

But this can get even more personalized, and more nuanced than just product prediction. If you’re worried about customer’s dropping off, a retention model can help you make certain changes:

  • Website navigation: Accommodate different users with navigation buttons, CTA’s, colors and web page set up. People who use Apple devices are used to certain buttons being on the left side of the screen, whereas android users are used to basic function buttons being on the right. Predict which set up a user will find easiest to navigate and provide them with that experience.
  • Checkout process flow: Certain customers may want one-click checkout, while others prefer to review their cart and go through the purchase process one step at a time. Some will want a chat box or a live representative to help them through, especially for large or complicated orders.
  • Back-in-stock messaging: Back-in-stock emails yield a high revenue per open ($2.43),2 and show click-to-open rates at 19%, as well. On the other hand, transactional, marketing, abandoned cart and other post-purchase drip email campaigns only saw click-to-open rates of 1-5%.
  • Channel access: If your customer works the night shift, they may be watching TV before they go to sleep in the early morning, unlike your other 9-5 customers. Predict where people will be and when they will access their social media, email, and CTV.
  • Relevant ads and discounts: If a section of your customers are first time homeowners, what will they need? Maybe they’ve never had to call a plumber before. You can set up personalized ads that offer first time homeowners a discount and a walkthrough of how to maintain their home’s system.

An LTV model or a share of wallet model can also help you use all these strategies to increase the average customer purchase and identify which at-risk customers are worth investing in.

Improve your account based marketing strategy

Instead of sifting through and marketing to a large base of individual consumers, sometimes millions of people at a time, account based marketing selects just one business, identifies and engages with key decision makers, and then converts. Your first party data may not have all the information on the company or key shareholders and leaders in that company, making your efforts to reach those individuals slow and aimed at a general audience. An in market model or a reactivate model can help you increase ROI by maximizing budget and reaching out to the right people at the right time with the right message.

Enriching your data with a 3rd party list can also help–like Data Axle’s B2C link, which connects professional and personal information so that you can connect with them at home, at work, and based on what’s important to them. These models can be developed and run periodically so that you can adjust your trajectory based on what’s really happening.

What is predictive analytics built on?

Clean data, accurate high-quality data.

Machine learning algorithms detect patterns and can be trained over time to respond to new data or values, giving you valuable insights. They use an enormous amount of data. Data with inconsistencies, duplicates or gaps will tangle the AI machine learning systems into an unusable mess, and you won’t be able to trust the insights that come out of them. The insights you get out of ML systems is only as good as the data you put in.

For instance, banks need to access reliable, real-time data on credit risk to help guide new account acquisition and make informed loan decisions. With intent data and predictive modeling, they can accurately prioritize prospect accounts. With out of date data they may give more credit to someone who has just opened five new credit lines in the last month.

The quality and security of the data matters, too. Data Axle offers over 400 attributes to help you segment audiences and make nuanced decisions. With Data Axle data in your CRM, you can develop customized predictive models. Optimized predictive models ensure that you contact the right prospects, at the right time, on the right channel with the right messaging/offers. And none of our data relies on cookies. It’s all permissions-based information that you can use with confidence.

Predictive marketing has the potential to transform the way brands do business. However, in order to reap the benefits, companies need to have a solid foundation of quality data, machine learning technology and personalization strategies.

Talk to us today about how we can help your company forecast their next success.


Natasia Langfelder
Content Marketing Manager

As Content Marketing Manager, Natasia is responsible for helping strategize, produce and execute Data Axle's content. With a passion for writing and an enthusiasm for data management and technology, Natasia creates content that is designed to deliver nuggets of wisdom to help brands and individuals elevate their data governance policies. A native New Yorker, when Natasia is not at work she can be found enjoying New York’s food scene, at one of NYC’s many museums, or at one of the city’s many parks with her two teacup yorkies.