Last updated
Customer lifetime value (LTV) is the revenue you expect from one customer over the whole relationship, not from a single order or a single visit. A plain model is average order value, times purchases per period, times the number of periods they stay. In A/B testing, it helps teams connect a term, metric, or behavior to a clearer optimization decision.
Customer lifetime value (LTV) is the revenue you expect from one customer over the whole relationship, not from a single order or a single visit. A plain model is average order value, times purchases per period, times the number of periods they stay. AOV , revenue per visitor , and ARPU are different metrics.
LTV is a forecast about a customer, or the realized total once the relationship is over. AOV answers "how big is an order?" ARPU answers "how much revenue per user in this period?" RPV answers "how much revenue per visit?" LTV answers "how much revenue does this customer represent before they leave?" You can build LTV from those inputs. You cannot swap the names. A one-week experiment almost never observes LTV directly, because the life has not happened yet.
A test can raise the first conversion and acquire people who never buy again. The test metric looks better. LTV, read later on that cohort, does not. Acquisition spending that ignores LTV will pay more for a signup than the relationship returns. That does not make LTV the right primary metric for a button test. It makes it the metric that keeps the program honest after several tests.
Customers in a cohort order twice a year, spend $40 per order, and stay three years. LTV is 2 × $40 × 3 = $240. That uses AOV as one input. It is not an AOV of $240. If a discount test lifts the first order and the same cohort then orders once a year, LTV falls even though the first-order conversion rose. The metrics guide is where AOV, conversion rate, and RPV are compared with each other. LTV is the longer total.
Name the cohort and whether the figure is historical or a forecast. Keep AOV, frequency, and lifespan as separate inputs so a change in one of them is visible. Do not power a two-week test on LTV. Use RPV or conversion rate inside the test, then read LTV on the cohort when enough of the life has elapsed. Definitions of the shorter metrics stay on their own glossary pages and in the metrics guide .
A common mistake is reporting last month's revenue per customer as LTV. That is closer to ARPU for a month. Another is multiplying today's AOV by a lifespan taken from a different product or a different year. The inputs have to describe the same customers.
AOV, conversion rate, RPV, and GMV: The metrics guide for the per-order and per-visit numbers. LTV is the longer horizon, defined here.
No. AOV is the average revenue of one order. LTV multiplies a purchase size by how often the customer buys and how long they stay, or it uses another explicit forecast. A store can raise AOV and still lower LTV if those larger orders happen once and the customer never returns.
Rarely inside the test window. Lifetime revenue is not finished when the test ends. Use a leading metric such as first purchase or revenue per visitor, and treat LTV as the later check on whether the customers you acquired were worth it. The metrics guide covers the leading numbers.
Whichever population you named: all customers, or only customers acquired in a cohort. Mixing this month's AOV with a five-year-old lifespan produces a number that matches neither group. Write the cohort next to the LTV or the figure is not comparable next quarter.
Visual editor, AI-generated variants with MidaGX, redirect testing, and GA4 integration — free forever in the Mida Sandbox. No credit card required.
Still deciding?
A quick screen share on your actual site — no slides, no generic tour. Just your questions answered.
30 min · no commitment · no sales pressure