Customer Lifetime Value (LTV)

How LTV is calculated

The classic eCommerce version multiplies average order value by purchase frequency per year, gross margin, and expected years of relationship. A customer averaging €80 per order, 2.5 orders a year, at 40% margin, retained for 3 years, is worth €240. Subscription businesses use margin per period divided by churn rate. Sophisticated teams compute it per cohort — customers acquired in the same period, tracked together in the order database — because LTV averaged across all customers hides whether recent acquisition is getting better or worse.

Why LTV lives in order data, not web analytics

LTV needs an identity that survives years: the same customer recognised across every purchase. Order systems have one — an email address or account. Browser-based analytics does not: cookies are capped at days in Safari, rejected outright by 40-60% of EU visitors, and never shared across devices. Any “LTV” a web analytics tool reports is really per-browser value over a cookie’s lifespan, which systematically undercounts. This is a structural limit, not a vendor flaw — and it is why SealMetrics, which measures anonymously and keeps no per-user identifier at all, does not claim to compute LTV. The order database owns the metric.

Where analytics data still distorts LTV decisions

The operative question behind LTV is LTV by acquisition channel: which sources bring customers who reorder, so acquisition budget can follow them. The lifetime margin comes from order data, but the channel label comes from analytics — and if that analytics only observed the consenting minority, the join is biased before it starts. Orders whose acquisition source went unmeasured land in a misleading “direct” bucket, and channel-level LTV inherits the distortion. Revenue attribution on 100% of orders — last-click, at channel level, no consent dependency — gives the LTV model an acquisition input that reflects all customers, not the measurable ones.

What LTV does not tell you

LTV is a projection, not a fact: it extrapolates future behaviour from past cohorts, and a pricing change, product shift or churn spike invalidates the history it rests on. It also says nothing on its own about how much profit acquisition generates — that requires pairing it with acquisition cost per channel. And a high average LTV can mask a wide spread where a small group of loyal customers subsidises many one-time buyers.