The Most Expensive Mistake in Marketing: Mistaking 'Fortune' for 'Fate’

In short: Tier upgrades and downgrades are rare events; Changes in customer trajectory happen every day. 

Yet most CRM programs focus precisely on those rare events. Upgrade rates, the number of customers reaching higher tiers, and high-tier retention—all are centered on those rare events. Meanwhile, the directional changes that occur every day often have no dedicated metric at all. 

Two Completely Different Types of Differences 

To understand this issue, we must first distinguish between two fundamentally different kinds of variation. 

The first type is the difference between people. Some customers naturally buy more; some buy less. This is their inherent rhythm, and it's hard for you to change in the short term. Let's give it an old name: Fate (命) . 

The second type is the change within the same person over time. The same customer buys frequently in the first half of the year, becomes inactive during the second half, and then returns six months later. They're still the same person, but their state is shifting. Let's call this: Fortune (运) . 

The most expensive mistake is confusing these two differences, especially mistaking Fortune for Fate 

Imagine a customer whose purchase frequency suddenly increases.If you interpret that increase as Fate, you may immediately reclassify the customer as "high value" and allocate additional resources to them. Your thinking might be: "This customer was always high-value. We simply failed to recognize it earlier." But the truth may be very different. Perhaps the customer has merely entered a temporary active period and will naturally return to previous spending levels in a few months.

The same data can produce two entirely different interpretations. If you see it as Fate, you invest heavily in a customer who was likely to decline anyway. If you see it as Fortune, you first determine where the customer is heading before deciding whether and how to intervene. The wasted budget comes from confusing the two. 

Interestingly, the method for seeing "Fortune" is far from new. 

In1959, statistician Ehrenberg provided a mathematical description of the fact that "people are inherently different from each other." In1982, an author named Greene explained this math as a metaphor, which he called the "River of Time."  

Imagine each customer is a piece of duckweed floating on a river, each at its own speed. The fast floaters buy often; the slow floaters buy less. You sit on the bank, unable to see the speed underwater. You can only see the duckweed when it occasionally surfaces. That's a purchase. The days below the surface, you can't see; you can only guess based on its rare appearances. 

The beauty of this metaphor is its honesty: the purchase records you have are never the customer themselves, just the few splashes they make when surfacing. 

Greene's math could do one impressive thing: clearly distinguish the fast duckweed from the slow duckweed. Given a purchase history, it could tell you with reasonable accuracy whether this is a fast fish or a slow fish. 

But in a chapter many readers skip, he honestly admitted what his method couldn't do: if the same piece of duckweed changed speed midway, his model couldn't see it at all and would discard it as noise. In other words, he could see Fate, but not Fortune. He knew he was missing a piece, but the tools to fill that gap didn't exist at the time. 

Those tools finally emerged in 2008. Researchers led by Netzer introduced a methodology known as the Hidden Markov Model (HMM) into customer analytics. The idea was simple but powerful. Customers are assumed to move among a number of hidden behavioral states. Each state has its own purchasing rhythm. The patterns of switching between states can be inferred from the sequence of their behaviors. For the first time, Fortune became measurable. Changes in customer momentum could now be modeled rather than ignored. 

Later, in 2018, research by Ascarza added another important insight. His work showed that many customers classified as "high churn risk" are actually impossible to save. No matter what retention efforts are applied, they are going to leave. As a result, large portions of retention budgets are often spent on customers whose behavior cannot realistically be changed. 

From 1959 to today, over six decades, each generation has filled in the gaps the previous one couldn't see. Ironically, most corporate  CRM systems still operate at roughly the level of the 1980s: they can see where a customer is, but not where the customer is going. 

Let's look at a real trajectory. Consider a real customer, anonymized as Customer 847.