Author: Ivaylo Bahtchevanov

Predictive Lifetime Value

When marketing resources are constrained, it is critical for businesses to identify and focus on the future high value customers that will have the biggest impact as these users represent the biggest opportunity for the business.

Traditionally, customer lifetime value is computed by looking at historical purchasing data for your customers. You take the average revenue from a given customer (average purchase value * average number of purchases) and multiply that across their average expected lifetime (e.g. before they churn).

While leveraging historical data provides a meaningful signal, this approach has severe limitations. When looking at current customers, it assumes the distribution of the past will resemble the distribution of the future, which is not necessarily the case. What’s more, it generalizes purchasing behavior across users and ignores variance between users of a given group. This approach also doesn’t give you a good understanding of how valuable new or prospecting customers are to your business.

This is where predictive lifetime value becomes useful - i.e. leveraging the power of machine learning to predict how much spend any given customer will perform over a specified period of time. You can forecast spend at a user level, or capture revenue for a broader segment or cohort.

Key Benefits

Predicting the future lifetime value of customers can provide a wide range of benefits for businesses.

Improved customer segmentation

By segmenting both current and existing customers by future expected value, you can analyze each segment, identify specific behaviors by demographic, and determine what sets your higher value customers apart. You can then systematically find more potential customers with the same profiles. This is useful both in prioritizing resources to maximize ROI and in providing highly personalized content or outreach. Most of Amazon’s sales come from the recommendations that are derived from their customer segmentation and a thorough understanding of what each user is willing to buy.

Improved Marketing, Sales, and Growth Strategies

By knowing the lifetime value of customers, businesses can tailor their marketing strategies to specific customer segments. For example, businesses can offer targeted promotions and discounts to high-value customers to incentivize them to continue purchasing from the business. They can determine who to invite to live events or in-person activities. For sales teams, they can improve productivity and efficiency by focusing their efforts on accounts which will bring the most revenue down the line. For existing customers, you can determine who should receive upgraded services, loyalty programs, or other perks. Retailers and ecommerce businesses such as Victoria’s Secret improved customer engagement by offering members-only benefits such as exclusive content, early access to upcoming sales, tips, and more.

Improved Customer Retention and Profitability

By focusing on retaining high-value customers, businesses can reduce customer churn and increase revenue over time. Knowing the lifetime value of customers can help businesses understand how much they can spend to acquire new customers while still making a profit. This knowledge can help businesses allocate resources to retain existing customers and improve customer loyalty. When customers feel valued and appreciated, they are more likely to remain loyal to a business and continue making purchases. This in turn leads to improved profitability - Bain & Co showed that a 5% increase in retention could lead to over 25% increase in long-term profits by keeping the right customers.

The Kumo Approach

Kumo’s goal is to make it very easy for teams to quickly generate predictions that will help you understand future potential outcomes based on a set of actions. This is accomplished by building a graph from your enterprise data, learning the relations and interactions between all the entities in your tables, and providing a no-code interface for orchestrating many ML models concurrently with a single query.

By using the Kumo Predictive Query, you can perform a comprehensive what-if analysis that enables you to determine the optimal strategy from a set of potential actions and identifies the best path forward based on a desired outcome. You can flexibly test multiple hypotheses with a series of queries.

As discussed above, LTV is a very useful input for an entirely new set of problems. Once you’ve built the customer segmentation based on LTV, you can then run subsequent queries to build your strategies in personalizing content, improving net retention, tailoring outreach and notifications, and many more growth optimizations,

Most importantly, because you can launch many queries concurrently with no additional overhead, you can tackle each of these problems with minimal time to value. This allows you to capitalize quickly on business opportunities before they pass.

To learn more about how Kumo can build your marketing and growth strategy, reach out to us directly!

Cookies settings