The Power of Predictive Analytics and Modeling

Machine Learning|8 Minute Read

The Power of Predictive Analytics and Modeling

“Predictive analytics powered by AI can take the data you already have, and unlock immense value from it,” Marketing Artificial Intelligence Institute Director Mike Kaput noted.

Kaput’s right, of course. But there is one significant caveat to his assertion:

To unlock this value, companies must invest in the right predictive analytics tool.

Specifically, companies need a modern, multi-faceted solution that enables not only efficient predictive analysis and modeling, but also substantive action across the business: from building rich, dynamically updated, multi-dimensional segments to orchestrating individualized lifecycle messaging to target prospects and customers.

predictive analytics

Why predictive analytics is a vital resource for companies with big data sets

The use of AI and machine learning at companies of all sizes and in various industries — from retail and ecommerce to financial services and insurance — continues to grow steadily.

Three in five marketing professionals polled by CMO Survey in early 2020 indicated they expect to substantially increase their use of artificial intelligence technologies through 2023.

And that includes systems that offer predictive analytics and modeling capabilities:

More than 50% of marketers said they plan to generate customer insights from predictive technologies in the coming years.

Let’s start on the analytics side.

With in-depth customer insights in hand, marketing and customer experience teams — particularly ones at large-scale companies with hundreds of thousands or millions of customers (and even more data points) in their database — have the requisite business intelligence and historical data to make accurate predictions about future engagement.

(More on that shortly.)

But other business users and teams can also benefit from access to predictive analytics:

  • Customer service and support: Knowing customers’ past interactions, behaviors, and purchases enables service and support representatives to more effectively engage those individuals in a reactive and proactive manner. For instance, they can ask high-volume shoppers to leave reviews and/or rate the company at their (predicted) ‘peak’ level of satisfaction with the business.
  • Digital product and experience: Publishers and retailers are creating compelling digital products and experiences today (e.g., premium content, membership programs, special apps) to boost customer engagement. Predictive analytics helps teams charged with developing and optimizing these programs segment and target the right individuals with promotions for new offerings.
  • Product/inventory management: Retailer Planet Blue uses BlueConic to improve its omnichannel marketing strategy. But the business also utilizes the predictive analytics capabilities in BlueConic to understand which vendors’ apparel sells most in both its ecommerce and brick-and-mortar stores. This sales data, in turn, informs which apparel brands to buy more or less from.

At the end of the day, everyone within a given organization — including, but certainly not limited to, the above-mentioned departments — is charged with improving customer experience (CX) and contributing to accelerated business growth.

With access to real-time predictive analytics — ideally generated and persistently stored in a true single source of truth (see: not a legacy solution) that unifies all customer data — these teams can all work more intelligently (e.g., smarter, perfectly timed messaging) and harmoniously (i.e., cross-departmental data democratization).

predictive analytics

How leading companies forecast future events and customer behavior with a CDP

Before any of the aforementioned teams can make the most of customer analytics, though, marketing and/or data science needs to develop and employ predictive models.

With modern tech, like our pure-play customer data platform (CDP), these teams have two options when it comes to predictive modeling and generating actionable insights:

  • 1) Utilize out-of-the-box (OOTB) models.
  • 2) Build, train, and deploy custom models.

If you have the appropriate personnel (e.g., tech-savvy marketers or a resident data scientist), the latter is certainly a viable option. For instance, BlueConic customers can construct custom models or import ones from their Python library using AI Workbench.

If you’re looking for a quick and easy way to get going with predictive modeling, though — and gleaning customer insights you can activate in real-time, cross-channel lifecycle orchestration efforts — then OOTB models offer an ideal starting point for you.

Here are two OOTB predictive models you can set up and deploy in BlueConic — and some sample action plans you can put into play to make the most of your predictive analytics.

customer lifetime value

Calculating customer lifetime value (CLV)

Based on customers with multiple ‘orders,’ the CLV model in AI Workbench helps BlueConic customers calculate the expected number of future purchases for those individuals.

Moreover, the customer lifetime value model can forecast when an individual is likely to stop buying from a company at a specific point in time in the future, based on purchase data.

The predictive CLV score computed by this model can then be stored in customers’ profiles and persistently updated as they continue to purchase more over time.

As noted, Planet Blue uses BlueConic to grow its business. One of the key predictive analytics-related data points it tracks regularly is CLV scores for online and offline shoppers:

  • Using rich, constantly refreshed profile data in our customer data platform — and without having to rely on internal analysts or data scientists or external agencies — the company uses AI Workbench to calculate CLV as well as recency, frequency, and monetary value (RFM) scores at the individual level.
  • By doing so, the Planet Blue marketing organization can create new, multi-dimensional segments (e.g., ‘Champions,’ ‘Potential Loyalists,’ ‘Needs Attention’) based on that real-time customer behavior and engagement data and, subsequently, develop programs geared toward those distinct segments.
  • This predictive analytics data then informs numerous facets of the business. For instance, marketing can determine how much to spend on ongoing programs and where to spend the most (i.e., specific channels). Meanwhile, in-store retail associates can offer VIP treatment to local, high-CLV customers.

Merging its online and offline first-party data in our pure-play customer data platform and utilizing our OOTB CLV model has clearly paid off (and continues to pay off) for Planet Blue.

Aside from gaining a deeper understanding of its customers, though, the business has also created operational efficiencies across the organization that will positively impact its growth and contribute to its ongoing transformation for years to come.

customer churn prediction

Forecasting customers’ propensity to churn

AI Workbench’s propensity-to-churn model identifies individuals who, based on the same purchase data used to determine one’s CLV score, are projected to ‘exit’ a company’s funnel.

Let’s say your company has a subscription-based business model. In BlueConic, you can create dynamic segments for various tiers and types of subscribers:

  • Customers who’ve been subscribers for specific time periods (one year, two years, etc.)
  • Your highest-value customers in terms of lifetime value, contract size, and/or revenue
  • Usage-based accounting for specific products or services to which customers subscribe

With our churn prediction model, you can then evaluate the risk of attrition for all customers and defined segments. You can then compare those segments against one another.

You can also pinpoint specific data points (called profile properties in BlueConic) associated with a segment to see which correlate most closely to a customer’s likelihood to churn.

Once you know which subscribers are most likely to cancel, you can build a segment for them and focus custom-tailored messaging or offers to only those at risk of churning.

For example, marketers who use BlueConic set up custom Lifecycles to move customers from one lifecycle stage to the next and entice likely-to-churn individuals to remain subscribers with hyper-specific calls to action based on the exact stage they’re in.

(And, in some instances, even cross-sell and/or upsell those individuals.)

customer analytics

What you need to unlock the power of predictive analytics and modeling

Referencing the firm’s Hype Cycle for Digital Marketing and Advertising 2019 report, Gartner VP Analyst Michael McGuire outlined the importance of predictive analytics for mid-sized and large-scale companies’ success in the next decade.

But first, McGuire noted users across the business — marketing, specifically — must become experts at understanding their analytics and utilizing those insights to improve customer experience and elevate their return on investment:

“Event-triggered and real-time marketing will have the biggest impact on marketing activities in the next five years. However, before marketers can realize the benefits of these technologies, they must first become proficient in predictive analytics and delivering personalized communications.”

As exhibited in the predictive analytics examples above, marketing is responsible for the delivery of targeted messaging to prospects and customers based on their first-party data.

But marketing — and the other aforementioned teams across the business — can’t leverage analytics until their C-suite invests in the optimal tech: a pure-play customer data platform.

Thankfully, many executives are starting to understand how applying analytical techniques to glean richer audience insights not leads to better marketing outcomes, but also accelerates business growth and helps teams across their organizations become more agile — particularly during trying, uncertain times like 2020.

customer analytics

Data analytics expert Lori C. Bieda noted for MIT Sloan Management Review why it’s now a strategic imperative for companies to ‘up’ their analytics game and become more agile:

“In order to detect and respond to disruptive events with agility, companies must increase their analytical fitness and develop strong muscle memory when they are put to the test.”

This, of course, applies to predictive analytics.

But Bieda also made an astute point regarding the larger data picture for companies today:

Without the right data infrastructure — the oft-discussed ‘people, processes, and technology’ component — it’s near-impossible to thrive even under the best of economic and business conditions, let alone during ‘down’ periods.

The first step to ensuring this agility — and realizing greater operational efficiencies across your organization — is to secure a pure-play CDP that aids teams throughout the organization make the most of your first-party customer data.

Learn how companies leverage our CDP’s analytics and modeling capabilities to better understand customers and improve marketing outcomes. Request a BlueConic demo today.

blueconic

See what BlueConic can do for you.