The LTV lie

At some point in the last few years, "we have a strong LTV" became the DTC equivalent of "we have a great culture." Everyone says it. Almost nobody can tell you exactly what it means or how they got there.

The number is usually real — technically. The problem is how it was calculated, and more importantly, which customers it's actually describing. Most brands compute LTV by averaging across their entire customer base. The loyal buyers who've been around for three years sit in the same calculation as the one-time discount shoppers who came in during a Black Friday sale and never came back. The average looks healthy. The underlying picture is often not.

The survivorship problem

Here's how the lie typically forms. A brand has been running for three years. They look at the customers who are still buying — the ones who show up in their retention data, their email segments, their loyal customer reports. Those people have high LTV. They calculate the average. The number looks great. They use it to justify their CAC.

What they're actually measuring is survivorship bias. The customers who stuck around long enough to inflate that average are not representative of the customers you're currently acquiring. The brand you were three years ago — different product quality, different positioning, possibly different price point — attracted a different type of buyer. You've been planning around their behavior ever since.

Meanwhile, the cohort you acquired six months ago is churning faster. The cohort you acquired during your last big promo is largely gone. The blended average masks both of those facts entirely.

A blended LTV number is an average of your past. It tells you almost nothing about the customers you're paying to acquire today.

What cohort LTV actually shows you

Cohort analysis groups customers by when they were acquired — a January cohort, a Q3 cohort, a post-rebrand cohort — and tracks how that specific group behaves over time. It's a slower, less satisfying way to look at the data. There's no single number that tells you whether things are good or bad. But it's the only method that shows you the truth.

When you look at LTV by cohort, a few things become visible that a blended number hides. First, you can see whether recently acquired customers are behaving differently from older ones. If your Q1 cohort is tracking 30% below your Q1 from two years ago at the same time horizon, that is a structural problem — and it's happening right now, not in the historical average. Second, you can see which acquisition channels are producing genuinely different customer quality. Your email-acquired customers probably have a different 12-month LTV than your Meta-acquired customers. Your subscription buyers look different from your one-time purchasers. Most brands know this directionally. Very few have actually measured it.

Tools like Triple Whale, Northbeam, and Afterwake have all built cohort-level LTV views into their platforms. Shopify's own analytics has a basic cohort report. The data is available. The bottleneck is usually the willingness to look at a number that might be worse than the one currently on the slide deck.

The revenue vs. profit problem

Even brands doing cohort analysis correctly often have a second issue: they're calculating LTV on revenue, not gross profit. These two numbers can tell completely different stories. A customer who spends $200 over their lifetime sounds like a solid LTV. But if your COGS, shipping, returns, and transaction fees eat $140 of that, the actual gross profit LTV is $60. If you paid $80 to acquire them, you lost money on the relationship — even though the revenue LTV looked fine.

This matters more than ever right now. With acquisition costs up 40–60% over the last two years, the margin for error between revenue LTV and profit LTV has essentially closed. You can no longer afford to plan acquisitions around a number that flatters the top line.

The practical fix

You don't need a data science team to do this better. The starting point is just changing what you report. Instead of "our LTV is $X," get into the habit of saying "our 12-month gross profit LTV for customers acquired in Q3 last year is $X." That specificity forces the right questions: is that number going up or down quarter over quarter? Does it differ by channel? What does the payback curve look like — are they profitable in month 3, month 6, or never?

The payback period framing is particularly useful for growth marketers because it connects directly to spend decisions. If your Meta-acquired customers pay back their CAC in four months on average and your TikTok-acquired customers take nine, that has real budget implications — regardless of what the blended LTV says. L.E.K. Consulting put it simply: LTV metrics should always specify the customer set and the time frame. Without those two things, the number is just a story you're telling yourself.

The brands getting this right aren't necessarily doing anything technically complicated. They've just committed to measuring the customers they're acquiring now, not the idealized average of the customers they used to have. It's a small shift in framing. The implications for how you spend, what you discount, and which channels you scale are anything but small.