Customer Data Alone Can’t Predict Customer Lifetime Value

Brands struggle to predict customer lifetime value due to inaccurate third-party data and insufficient customer data. What they need is a comprehensive database of platform-agnostic behavioral and attitudinal data to target customers with the highest lifetime value.

Richard Purcell

Why Customer Data Alone Cannot Accurately Predict Lifetime Value 

Bad predictive models cost money. In the throes of the Covid-19 pandemic, the housing market turned white hot. In response, Zillow Offers, the home purchasing and house flipping arm of Zillow, cranked up their home purchases.

And then they lost more than $420 million in a quarter. Why? Their algorithmic model for predicting prices and buying and selling homes didn’t accurately predict home price appreciation. Their CEO even said: “Fundamentally, we have been unable to predict the future pricing of homes to a level of accuracy that makes this a safe business to be in.”

Zillow’s mistake? Using customer purchase and irrelevant behavioral data to inform their business strategy. Their algorithm didn’t anticipate the quick cooling of the housing market. 


This is a common mistake — one that prevents brands from targeting their most valuable customers and building a sustainable, long-term customer base. To avoid this mistake, brands must build their models with platform-independent behaviors and attitudes.

Why? Because past purchase decisions are not enough data to find customers with the highest lifetime value. 

Even if brands have access to third party data (in addition to purchase data) to make accurate predictions on customer profitability, it is very costly to obtain and its historical significance may not relate to present-day consumer behaviors. If there’s anything businesses like Zillow learned from the pandemic, it’s that past behavior cannot predict future behavior. A lot of consumers, for example, left urban life behind during the pandemic, changing their fundamental needs and interests in the process.

Let’s look at some of the ways that marketing to consumers has changed, and why you need predictive data that isn’t based on past behaviors.

World Events Impact Consumer Behavior

Covid-19 changed many business B2C models. People stopped going to movie theaters for more than a year. Gyms closed and lost members to virtual workout subscriptions. Grocery stores went big on order pickup or delivery, losing out on impulse in-store purchases. 

Likewise, many subscription services saw a huge increase in demand during the first 18 months of the pandemic. Meal kits and deliveries in lieu of eating out; digital streaming to replace movie theaters; Pelotons take the place of the gym. 

But as the subscription economy booms, how will these businesses retain their customers, especially as consumers gradually return to their pre-pandemic routines? And how will businesses like movie theaters successfully reactivate the customers who might be ready to return? 

Brands Need Reliable, Accurate Data

Brands may be tempted to look at their existing customer data to build retention and growth strategies. But as consumer behaviors change and the economy re-opens, the customer data businesses gathered between March 2020 and October 2021 was dictated by the pandemic and is relevant only to that point in time. As pandemic life slowly recedes, consumers will change their habits and demands again. 

Some brands realize they need more individual insights on their customers to make an impact and turn to tools like Facebook or Google. However, those channels don’t provide individual-level insights. In fact, the insights they provide can be downright false: Facebook knowingly showed inflated engagement metrics for their own gain. Others turn to third party data providers for income data on their customers, however many brands indicate that there is only a 40% match between what the vendor provides and the brand’s actual customers

Similarly, there is no way to confidently take action on data from tools that track location of smartphone users. There’s simply no way to validate that your customer actually visited McDonalds, nor can the data indicate whether they are also a frequent Burger King patron. 

The desire for more demographic and behavioral data clearly demonstrates that customer purchase data simply isn’t enough. Brands need in-depth, platform-independent demographic and behavioral data that will indicate who is worth attracting — and retaining. By platform-independent, we mean foundational behaviors and attitudes outside of the brand. For instance, if a consumer eats McDonald’s every Friday, McDonald’s will only know their behavior on Friday; yet it’s important to know what they eat the other 6 days out of the week. 

Here’s how to find that high-value customer.

Not Every Customer Will Have High LTV

As brands look to reactivate lapsed customers and find new ones, many realize that the cost to acquire and service is too high to make financial sense.

Let’s say a home workout subscription plan that gained thousands of pandemic-era subscribers is looking to mitigate customers who might churn. Based on their customer data, the brand thinks that when a customer stops watching videos, it signals a churn risk.  So the brand reaches out to all 100 customers who stopped watching videos that month and offers them a free month.  

In doing this, the fitness brand treats all its customers as the same because they don’t know anything about each individual customer. There are likely many of these customers who would take the free month and then churn anyway. Why? Because they were never the right demographic, behavioral, or attitudinal fit to be long-term customers, and the brand didn’t know that.

But 30 of those customers are women in households with $100k+ in income, worked in fashion, and showed churn behaviors. These customers’ demographic and behavioral data all indicate that they are high-value customers. From the same data, they learn that this group of customers responds well to text messages from brands. Had the brand sent an email, these high-value customers may never have responded. But in understanding their customer, the brand can learn not just which customers to pursue, but how and when to pursue them.

Know the Customers With the Highest LTV

Leslie Emmons Burthey, VP of marketing at GoodRX, sums up the issue well:

“In a very fragmented data world, brands are stumbling in the dark when it comes to predicting lifetime value. The demise of reliable third-party data means LTV predictions have to be built on scarce customer transactions and usage. More relevant data at the individual level from outside the customer data warehouse means more accurate predictive models, which ultimately means more confidence in business decisions.” 

When brands can cross-reference their customer data with platform-independent behaviors and attitudes, they can gain an understanding of who their ideal, high-value customers are. 

It means that brands do not rely on lookalike audiences or spray-and-pray promotional emails, but data on their specific customers’ actual behaviors. Taking action on known information yields higher-quality conversions, more revenue, and businesses built on long-term customers.