Predictive Lifetime Value & Why Most Marketing Teams Are Measuring the Wrong Relationship
![[OPT 3] IMG CMO – Editorial – predictive lifetime value and what to measure_-18130](https://thecmo.com/wp-content/uploads/sites/8/2026/06/OPT-3-IMG-CMO-–-Editorial-–-predictive-lifetime-value-and-what-to-measure_-18130-792x446.jpg)
Customer LTV Misconception: Many CMOs miscalculate customer lifetime value, leading to poor marketing and budget decisions.
Predictive Model Importance: Adopting a predictive LTV model shifts the understanding of customer relationships and spending potential.
Content Strategy Shift: Realizing extended customer value necessitates an evolved content strategy focused on retention, not just acquisition.
Data Utilization Challenge: Building a predictive LTV model involves challenges with legacy systems and requires incremental data integration.
LTV Diagnostic Checklist: Assess your current LTV model against critical questions to identify gaps and opportunities for improvement.
Ask most CMOs what their customer lifetime value is and they'll give you a number. Ask them how they calculated it and you'll usually get some version of average order value multiplied by purchase frequency. It's a reasonable approximation and one that’s easy to defend.
Mike Birney, CMO of One Natural Way and Sposey, used it too.
He also built a marketing strategy on top of it, ran acquisition campaigns based on it, and made budget decisions with it. It wasn't until he started pulling the real numbers apart that he realized the figure he'd been working from had never actually captured what his customers were worth.
Want more from The CMO?
Sign up for a free membership to complete reading this article:
Have an account? Log In
"When mom comes through our door and the baby's born, now it's bye-bye, and that's not really the case, because people have two kids, they have three kids," Birney says.
We've really struggled to understand what our LTV looks like because we're not able to really track that well.
-Mike Birney, CMO at 1 Natural Way
Your LTV Number Is Probably Wrong
For families, a second pregnancy, a third, the expanding product line that now includes lumbar support bands, milk storage bags, and telehealth lactation classes. And it missed Sposey entirely, the sister brand that picks the customer up again when kids move toward potty training.
One Natural Way provides breast pumps and pregnancy supplies covered by insurance. The way the business had been measuring it, a customer would come in during pregnancy, get what she needed, and leave. In theory, it’s the end of the relationship..
Except this perspective missed everything that comes after. Years of potential revenue, invisible in the metric the business was navigating by.
If we don't truly understand what that is for our company, how can we make informed decisions moving forward?
It's not a rhetorical question, and the problem isn't unique to maternal health. Research published in the Decision Analytics Journal (2025) applied predictive CLV modeling to a health services portfolio and found that 50% of customers had a predicted value of $363 or less over twelve months, while the top tier reached $10,375.
This is a spread of nearly 30x within the same customer base. When you flatten a range that wide into a single average and use it to set your acquisition budget, you're not using data to power decisions in a meaningful way.
As marketing researcher Peter Fader and colleagues have argued, businesses routinely undervalue customers by anchoring to short-term metrics like Customer Acquisition Cost, which optimizes for cost reduction rather than investment in high-value relationships.
Your LTV number isn't wrong because your math is bad. It's wrong because it was built to measure something simpler than your actual customer relationship.
What a predictive model changes
Birney is now building a predictive LTV tool using AI, designed to model what a full customer journey through One Natural Way looks like — across multiple pregnancies, across both brands, across the years in between. The inputs are historical purchase behavior, average family size, and the product line as it currently stands and continues to grow.
The distinction between historical and predictive LTV matters more than it might seem. Historical LTV tells you what customers have been worth. Predictive LTV tells you what a customer you're acquiring today is likely to be worth — and that changes how much you should be willing to spend to get them.
If your average order value is $200 but a customer acquired during a first pregnancy is worth $1,400 across four years and two kids, channels that appeared marginal suddenly justify their spend. The economics of retention shift too. You start treating the gap between purchases as a relationship to maintain, not a lapse to ignore.
Content as a retention instrument
This is where content strategy stops being a support function and starts doing real work. Once you understand that your customer's value extends well beyond the initial transaction, what you publish and when becomes a deliberate retention decision.
Birney describes One Natural Way's north star as "mom's first always" — being present and useful across the full parenting journey, not just at the point of need. That positioning only holds if you're actually showing up between purchases.
A new mom who orders a breast pump has an array of products that might be valuable to her. She might need nursing support, sleep guidance, or to know that lumbar support bands exist for the back pain that’s been nagging at her since delivery.
And later, she'll be in the market for what Sposey offers. The question is whether she thinks of you when the time comes, and this depends entirely on whether you remain in her orbit.
"All these are additional revenue drivers," Birney says of the expanded product line, "but we haven't been able to really tie that into LTV."
The predictive model is what closes that loop. It tells you which customers are likely to return, when, and for what, and that gives you a content calendar with real commercial logic behind it.
You're not publishing to pump out content. You're sharing content to stay relevant to a customer whose next purchase you've already anticipated.
For most marketing teams this requires a mindset shift. Content produced during the acquisition phase has an obvious job. Content produced between purchases often gets treated as brand awareness. It can be vague, hard to attribute, and easy to cut.
Predictive LTV reframes it as being essential to pipeline. You're nurturing a future transaction with a customer you already have, which is almost always cheaper than acquiring a new one.
The data problem underneath all of this
Building a predictive model sounds cleaner than it is. Birney is candid that pulling analytics into the broader AI equation has been harder than the content and campaign work. Legacy systems are embedded in people's workflows.
Teams build habits around familiar tools and migrating off them is a project with real cost and disruption.
Rather than waiting for a perfect infrastructure moment, he's building incrementally — starting with available data and connecting more sources as the work develops. This approach reflects a broader principle he applies to AI adoption generally where you find one specific pain point, solve that, and build from there.
A predictive LTV model that's operational and 70% accurate is worth more than a perfect one still a work a progress.
Probability-based models outperformed more complex machine learning approaches, largely because the foundational data inputs were sound. Getting the data right matters more than getting the method sophisticated.
Run the diagnostic on your own number
Pull your current LTV figure and stress-test it against these questions: Does it account for product line expansion since you last calculated it? Does it model repeat purchase cycles specific to your customer's life stage, not just your average order cadence? Does it connect to your content calendar in any meaningful way, or are those two systems operating in separate conversations?
If the answers are mostly no, you're not alone. Most LTV calculations were built for simplicity, not accuracy. Your goal isn’t to build something perfect. It's to develop a model more honest and informative than what you have now. And, to let this number start informing decisions it currently has no part in.
Start with one data source you already have, run a basic model and see what the range looks like. Odds are, it's wider than you expected, and this gap is exactly where you have work to do.
What’s Next?
For the latest insights, join us at the CMO Club for interviews with marketing leaders, and resources that can help you drive growth in your business.
