Feature Prioritization Through Customer Lifetime Value

A strategy to unlock future customer loyalty.

Many businesses market incorrectly. Intuitively (or formally), they know the three steps of marketing a product are Segment-Target-Position:

1. Segment: Divide the market into Mutually Exclusive, Collectively Exhaustive (you may hear your management consulting friends refer to MECE) units;
2. Target: Choose the segment that your product aligns with the most; and
3. Position: Explain why your product is superior to the next-best alternative for your chosen target segment.

Though simple in theory, this framework is clearly not foolproof, as evidenced by the countless failed products released to the market. Personally, I think it is because many businesses mess up the first step of the process in defining how they will segment the market.

The Pitfalls of Demographic Segmentation

Unfortunately, many (frankly, bad) marketers — or business partners, product managers, product owners, analysts, etc. — fall back onto demographic-based segmentation. Demographic-based segmentation is an absolute no-no.

With demographic-based segmentation, your best-case scenario may mirror the US rollout of Honda’s small, five-door hatchback, the Fit. The Fit was brought from overseas to appeal to young car buyers who were looking for an automobile that was small, affordable, safe, reliable, fuel efficient, and — most of all — spunky.

Instead of finding its intended, youthful audience, the Fit found a market with older buyers who happened to be looking for the exact same characteristics in a car. My ex-mother-in-law had a Fit — not what Honda was going for! Luckily for Honda, this happy accident kept the Fit in US dealerships for about fifteen years; however, they were not able to reach young buyers as they originally hoped.

A word of caution, the Honda Fit illustrates how a target audience might be misaligned in a product’s go-to-market strategy, but it is just as easy (and much more cringe) to miss on a product designed for a particular demographic. See BIC’s pink pens.

Psychographic Segmentation: An Alternative Approach

Thankfully, a respectable alternative to demographic segmentation gained popularity in the 1960s and 1970s: Psychographic Segmentation.

Psychographics take into account buyers’ beliefs, values, motivations in order to home in on potential customers. Often, these segments are born from a few rounds of marketing research that culminate in a survey that is run through a K-means cluster analysis, which is sent to Marketing for naming.

Naming? Yes, naming. Marketing understands that firm leadership has a tough time remembering the difference between Segment 1 and Segment 4, but they form attachments to either Cost Cutting Carlos or Frugal Frida. Our familiarity with (and subsequent eyeroll following) these goofy names is a testament to the prevalence of psychographic segmentation. And for good reason: psychographic segmentation is a quality alternative to demographics.

One small problem. What I say in a survey and what I do in real life oftentimes don’t align. A slightly bigger problem is that I’m not unique in that divergence. In fact, it’s so common for people to have a gap between what they say and what they do that researchers have coined a term for it: the say-do gap. To be clear, the say-do gap doesn’t render psychographic research worthless, but it does necessitate a mindful understanding of its limitations and continued search for alternative approaches.

Value-Based Segmentation: Tracking Customer Behavior

Enter value-based segmentation! Whereas psychographics focus on what survey respondents say, value-based segmentation tracks what customers actually do. It pays attention to how much money a customer brings to the business, and how much it costs the business to acquire that customer.

However, value-based segmentation isn’t as mature as psychographics, and as a result, there are still a lot of unknowns in terms of best practices as well as many naïve approaches. A few things to understand about customer lifetime value are that it should:

1. Be based on margin, not revenue;
2. Account for time value of money; and
3. Be future oriented.

The first point should be self-explanatory. If Adam buys a $100 item with a 5% margin at the same time that Betty buys a $100 product with a 70% margin, then Betty is a more-valuable customer even though Adam contributes the same revenue as her.

The second bullet is a little bit more complicated, but it should still make sense intuitively. If Charlie buys a $20-margin item today, and Debra buys a $20-margin item in a year, then Charlie is a more-valuable customer than Debra even though they contribute the same margin because a dollar today is worth more than a dollar tomorrow (given it can be reinvested).

Future orientation is what turns value-based segmentation into the Wild West.

The purpose of business isn’t about rewarding past behavior, but it is an effort to maximize the time value of future interactions. The same way that the Yankees offering Babe Ruth a new contract today provides no value, a company rolling out the red carpet to customers based solely on their historical purchases doesn’t benefit the business. Often, history can be indicative of the future — we’ll get there in a moment — but at some point, all customers will eventually stop doing business with you. Therefore, the past can’t be what drives our decisions; it must be future expectations.

A Case Study: Creature Comforts Brewery

An unconventional example of future focus comes from an Athens, Georgia-based craft brewery, Creature Comforts. The brewery launched in 2014, and its flagship Tropicalia IPA quickly became supply constrained across Georgia. At some point, a group of Georgia natives, who were in Atlanta for an extended project with their California company, brought their coworkers the beer, and everyone loved it. The director of the project was quickly fascinated to hear about the cult following and had to know more. His assistant reached out to the brewery for a meeting to learn about their young company.

Of course, the brewery obliged and ended up sending the project team on their way with some free beer and, upon request, a 2XL shirt. The director’s name was Joe Russo, the beer and the shirt would be sported by Thor (played by Chris Hemsworth) on screen in Marvel’s Avengers: Endgame movie, which grossed more than $1 billion worldwide. Beyond providing free product placement in a blockbuster movie, Joe Russo was rumored to be an investor in the brewery’s cross-country expansion into a new space in Los Angeles. While his historical purchases certainly did not position him as a top customer, Joe Russo’s future time value made him central to Creature Comfort’s corporate strategy.

Statistical Models and CLV Calculation

For those of us who don’t have Hollywood beating down our doors, statistical models can estimate an individual’s future value to a company based on their past purchases. Without going into too much detail, I’ll explain how one such model — the beta-gamma beta-binomial gamma-gamma — works. Based on a transaction log that can be as simple as a customer ID, timestamp, and purchase amount, the model creates RFM slabs for cohorts. RFM is a term from direct marketing (think: mailers) that stands for Recency, Frequency, Monetary Value. Essentially, it’s an easy way to segment customers based on what they’ve actually done. We break customers into cohorts, or groups, based on when they were acquired, understanding that different internal and external factors may impact why a customer first purchased when they did.

The model assumes that customers have predetermined propensities to stay “alive” (i.e., they’re still available to be customers — they haven’t moved; they haven’t won the lottery; they haven’t deceased), which we’ll represent as a “death coin”. Every time there is a buying opportunity, customers flip their death coins. If the coins land on “death” then they are no longer customers ever again; they have won the lottery and moved to Monaco. However, if the coins land on “alive” then the customers subsequently flip another coin, a “purchase coin”, which represents their predetermined propensity to buy during the period; it can land on either “buy” or “don’t buy”. These two coins are based on customers’ past recency and frequency of purchases. The monetary value piece basically assumes that past check size will be indicative of future check size.

It’s important to point out that this model assumes a customer’s death coin, purchase coin, and check size are intrinsic to them. In other words, there is little a company can do to extend a customer’s life, increase their propensity to purchase, or grow check size. For the most part, these things are what they are. Buying behavior doesn’t change.

Strategic Focus on Top Customers

Understanding that our actions cannot drive meaningful changes in buying behavior, we abandon efforts of transforming average customers into great ones. Instead, our strategy should shift toward finding as many fantastic customers as possible. Layering on top of the 80/20 rule, which assumes that 20% of customers generate 80% of firm value, it becomes increasingly clear that we should narrow the focus of our products to these top customers. Consider how casinos comp some guests’ stays but certainly not mine.

You may ask yourself how to identify these individuals. A good CLV calculator is going to provide the most granular detail. It is important to understand how the model works. It is also helpful to know that it may look more like a consulting arrangement rather than a product or subscription. The basic calculations are pretty much the same for most products and companies, but models will almost always need minor tweaks here or there.

In lieu of the time and resources required to build a model, which should be updated each period you introduce a new cohort (depending on your business, this may be weekly, monthly, quarterly, or annually), you might be able to hack it. Though obviously imperfect, you can directionally estimate future purchases based on the first part of the RFM framework — Recency. Customers with the most recent purchases are likely going to buy again. Given that purchases are often streaky, customers may very well buy in the very near term, maintaining a high time value of money.

Loyalty Programs and Customer Engagement

Shifting gears to bring all of this home, it has to tie back to agile software development. Focus on top customers is oftentimes most visible via companies’ loyalty programs. These programs can provide customers with a number of benefits (e.g., discounts, saved preferences, etc.), but my favorite incentive is access, which can be either exclusive, guaranteed, or early, for top customers.

Perhaps America’s favorite loyalty program, Amazon Prime, boasts nearly 170 million members who spend 2.3x more money annually (estimated $1,400 vs. $600) than non-Prime members. As a result, Amazon tries to deepen its relationship by offering free shipping, video and music streaming, photo storage, sales (i.e., Prime Day), and more. Many of these benefits are offered only to Prime members and provide a great illustration of exclusive access.

Guaranteed access has become a popular perk in travel communities. Specifically, most hotel loyalty programs offer guaranteed rooms with a minimum notification period. Similarly, chipmakers (e.g., Intel, Nvidia) often reserve units for top customers, such as Microsoft and AWS, guaranteeing that they remain on the bleeding edge of computing, even while their suppliers work out kinks while ramping production.

As an aside, these top customers are regularly tapped to provide input on product decisions. This practice may be most evident in the sports world. For instance, Gatorade was launched to assist University of Florida Gators student athletes during competition. The trickle down is that if it is designed and works for world-class athletes, then it will serve just fine in my recreational pursuits. Likewise, Nike is famous for partnering with the world’s best athletes. I promise if a basketball shoe is designed for Michael Jordan or LeBron James then it will be more than sufficient for me.

A technological example of this strategy is evident through Epic Systems, a privately held, Madison, Wisconsin-based software company that houses medical records for nearly 80% of US patients. After decades of steady growth, the company suddenly exploded when medical heavyweights like Kaiser Permanente and Cleveland Clinic signed on as customers. Afterwards, the halo effect trickled down as many large- and medium-sized health systems determined that if Epic was designed for industry leaders then it would work just fine for them.

Involving Top Customers in Product Design

In addition to involving your best customers in product design, it is rewarding for them and wise for customers to give early access. In a technology setting, this could be through the use of beta programs. The sequence of events may look something like a feature request followed by collaborative design then an MVP customer and finally a beta for like-users.

Betas give customers new functionality as well as bragging rights and recognition. In return, product teams receive early feedback from their best customers who are okay with bugs. It can occur on a small scale, or it can be massive. The Apple Beta Software Program is separate from the Apple Development Program, public to anyone, and grants members access to pre-release software. The tech giant further leverages the program to tap select developers for invitation-only initiatives, allowing these individuals to play important roles for launches of, say, operating systems.

So what?

So I want to encourage decision makers who prioritize backlogs to stop including enhancements for the stakeholders who scream the loudest or to quit bringing in features that will benefit the most number of users.

Instead, I want you to focus on your best customers. Identify what those with the highest future time value of money will find important, and make sure to include that this sprint. Figure out what that is by giving them access, be it exclusive, guaranteed, or early. Be okay if 80% of your user base misses the value or only gets the trickle down, and reward the future behavior of your best customers. In doing so, you will grow and find some new ones along the way.

This article is a preview of the upcoming Navalia Pulse — an anthology of thought-provoking opinions on technology and its critical role in organizations’ future. Sign up here to receive updates on when it will be available.