What is customer lifetime value?

Customer lifetime value is two numbers sharing one name.

One is a measurement. It is what a customer has already spent, and your platform holds it.

The other is a forecast. It is what that customer spends next, and it is the half that every spending decision actually rests on.

They get printed side by side, in the same currency, under the same label. Only one of them can be checked.

Customer lifetime value, as published on 2026-08-21. Klaviyo splits it three ways — historic, predicted, and the sum of the two. Omnisend splits it two ways and publishes no time window for its prediction. Shopify’s analytics reference carries no field with the name at all; it measures amount spent and uses the phrase only to say what the number is for. Salesforce defines it as revenue, then publishes a formula that subtracts cost. Bloomreach and Yotpo multiply by profit margin. Drip and IBM do not.

What is customer lifetime value?

It is the total money one customer is worth to a business across the whole relationship.

Salesforce states it plainly: “Customer lifetime value (CLV) is the total revenue a business can expect from a customer throughout the entire relationship.”

That sentence is not where the disagreement lives. The definitions broadly agree.

The disagreement starts at the arithmetic, and the same page says so: “There’s no single formula for customer lifetime value, but this is the one most widely used:”

Is CLV the same as LTV?

Yes. Three abbreviations circulate for one metric.

Drip puts it directly: “You’ll see CLV and LTV used interchangeably. They mean the same thing in this context.”

It records a convention as well — LTV for subscription and SaaS economics, CLV for ecommerce — and then says it is not kept: “but plenty of people swap them freely.”

Yotpo publishes a third: “Customer lifetime value (CLTV, or LTV for short) is the numerical value for how much a customer is worth to your company across the duration of the relationship.”

So the initials do not settle which quantity is meant. That is settled by the formula, and Yotpo says the formula is unsettled too: “Surprisingly, there is no one standard formula to calculate customer lifetime value.”

Which half of the number can you look up?

The backward half. Every platform that shows the metric shows what the customer has already spent.

Klaviyo calls it “Historic CLV” — “The total value of all previous orders an individual has made, taking into account any refunds and returns.”

Omnisend calls it “Historical CLV” and defines it as “Total amount spent to date”.

That figure is a fact about the past. It is the only part of customer lifetime value that is not an assumption.

What do the figures look like on one profile?

Klaviyo prints them together on a profile. These are the values in its own documentation:

The measured part is $401 of the $500. The rest is a forecast about a year that has not happened.

Does every platform count the same money?

No. Three references, three treatments of what counts as spent.

Klaviyo nets refunds and returns into its figure.

Omnisend drops the order instead: “Only orders with paid or partially paid statuses count toward CLV. Orders with other payment statuses are excluded from the calculation.”

Shopify’s broadest cohort field goes the other way and bundles everything in — “Amount spent (subtotal, taxes, shipping, sales reversals, discounts, fees, and more), from customers in the given cohort”.

Shipping and tax are not customer value. In that field they are counted anyway.

Is there a field actually called customer lifetime value?

Not on Shopify’s analytics field reference. What that page holds is spend.

Total amount spent is “Amount the customer has spent across all time”. Amount spent per customer is “Average cumulative amount customers in the given cohort have spent at your store”.

Where the phrase does turn up, it is in the Uses column — advice about what to do with the figure, not a definition of one. Customer amount spent carries “Guide marketing investment decisions based on customer lifetime value.” Total amount spent carries “Calculate sustainable customer acquisition costs based on lifetime value.”

So on that reference, lifetime value is a purpose. The fields underneath it measure the past.

What is the predicted half?

An estimate of money that has not arrived.

Klaviyo publishes it as “Predicted CLV”, defined as “A prediction of how much money a particular customer will spend in the next year.”

Omnisend publishes “Predicted CLV” as “Estimated future revenue based on purchase behavior.”

IBM draws the same line in general terms: “The historic CLV looks at how much an existing customer has already spent with the business. The predictive CLV is an estimate of how much a customer might spend.”

How far ahead does the prediction reach?

One year, on both platforms that publish a window.

Klaviyo says it twice. The profile field is next-year spend, and the segmentation page repeats it: “Sum the Predicted CLV of all members of a segment and you will get the expected revenue from customers in this segment for the next year.”

Shopify’s predictive dimension uses the same window. Predicted spend tier is “Label specifying how much the customer is predicted to spend in the next year”.

Omnisend states a method and no time window: “Omnisend calculates this by comparing the contact’s purchase history, order frequency, and spend to similar customers in your store.”

Lifetime and next year are different questions. The dashboards answer the second one.

Is the prediction meant to be right about one customer?

No, and the vendor says so.

Klaviyo: “predictions work best when averaged over many customers and are not expected to be exact for any single individual”.

The same page shows why. One profile can carry an impossible order count — “you may see 1.43 as the number of predicted orders for a particular customer”.

Read one profile and that is nonsense. Read a thousand and it is a forecast.

Is there a published formula behind the prediction?

No. Klaviyo describes a model: “Klaviyo automatically builds a Customer Lifetime Value (CLV) model using your company’s data and retrains the model at least once a week.”

Omnisend describes a comparison against similar customers.

So the individual prediction cannot be reproduced. Neither vendor publishes arithmetic that would let you derive one profile’s figure, or see which inputs moved it.

What is the number actually for?

Two decisions, and they lean on different halves of it.

The first is how much you can pay to acquire a customer. Shopify puts that use straight onto its spend fields: “Calculate how much you can afford to spend acquiring new customers.”

That attaches a forward-looking decision to a backward-looking field. It holds when the past is a fair guide to the next year, and it is the version most stores can actually populate.

The second is who gets which treatment. That one runs on the forecast.

How do you use it to decide who gets what?

By sorting people on what they are expected to be worth, then treating the groups differently.

Klaviyo’s own example is a store whose average order value is “around $15”, building “a segment of customers who are predicted to spend no more than $5” and sending it a discount.

That is the low end of the range being handled on its own. Churn risk sits beside the value field, exported as a decimal: “Churn risk will be exported into your CSV as a number between 0 and 1. For example, 0.45 would correspond to a 45% churn risk.”

What does the software not decide?

What any of those people are sent.

The platform sorts the list and scores the risk. It does not decide what a low-value customer should be offered, what a discount is worth giving away, or what a campaign is about.

That judgement sits upstream of every field on this page, and it is where the hours go — before anything is designed, written or scheduled.

How do you average it across a segment?

Klaviyo publishes a recipe, and it is worth reading closely before using it: “You can calculate the average customer value of a segment by averaging Historic CLV and Total CLV.”

Total CLV is already historic plus predicted. Averaging the historic figure with that sum returns the historic figure plus half of the predicted one.

Run it on the profile above and it gives 450.5 — above the measured $401, below the total $500.

That is a half-weighting of the forecast rather than an average of two independent quantities. It is a defensible thing to want. It is not what the word average suggests.

Does “value” mean revenue or profit?

Both, depending on who published the formula. It is the widest split of all.

The platforms measure revenue. Omnisend: “Customer Lifetime Value (CLV) shows the total revenue a customer brings to your business throughout their customer journey.” No margin term appears on the platform side at all — not in the two definitions published under this name, and not in the spend fields of the reference that publishes no definition.

Drip’s formula is revenue too: “CLV = average order value × purchase frequency × customer lifespan.”

Yotpo publishes the same three factors and a fourth: “x Profit margin”.

Bloomreach starts from profit: “CLV can be calculated by multiplying the average annual profit of a customer by the average duration of customer retention.”

And Salesforce, having defined the metric as revenue, publishes “CLV = (Average Revenue Per Customer × Customer Lifespan) − Total Costs to Serve” — a formula that returns what is left after cost.

What does the margin term do to the number?

It divides it, and one guide’s own example shows by how much.

Bloomreach works its example as 25 (AOV) * 2.67 (F) * 0.41 (GM) * (1/0.6) = $45.7.

Drop the 0.41 and the same inputs give 111.25. That is 2.44 times the published answer, because 2.44 is one divided by 0.41.

The same gap runs in reverse on Drip’s example, which ends at “$300 × 5 = $1,500.” Apply Yotpo’s “50% profit margin” to it and the answer is 750.

Between those two published formulas the margin term is the only difference in structure. It is worth half the answer.

How long is a “lifetime”?

Between one year and fifteen, depending on who you read.

The platforms say one year. The guides pick a lifespan and multiply by it.

IBM’s coffee-shop worked example is “The formula: CLV = USD 5 (average sale) x 100 (annual visits) x 5 (years) = USD 2500”. Its other two examples assume four years and fifteen.

Drip assumes five. Yotpo assumes seven — “customers shop with this brand for seven years” — and lands on “the average customer lifetime value is $350.”

None of those is wrong. Every one of them is chosen.

Where does the lifespan figure come from?

Two methods, and they do not measure the same thing.

Yotpo takes it from order dates: “(First order date – Last order date) ÷ 365 days per year = Individual customer lifespan”

Run that subtraction the way round it is printed and you get a negative number, so take the later date minus the earlier one. Those individual spans are then averaged: “Sum of customer lifespans ÷ Total number of customers = Average customer lifespan”.

Bloomreach derives it from churn: “Customer Lifetime Period = 1/Churn Rate”, worked as “Churn Rate: 60% -> Customer Lifetime Period: 1,67”.

A dates-based lifespan can only be measured on customers who have already finished buying. A churn-based one can be computed today, and it inherits whatever the churn rate was defined as — and that definition is not settled either.

Does anyone discount future money?

No. The published formulas add up expected future spend at face value, with no present-value adjustment in any of them.

Over a one-year horizon that barely matters. Over a fifteen-year one it is most of the answer, and a projection and a measurement end up printed as the same kind of number.

What does it take before a platform shows you the number?

Klaviyo publishes four conditions, and they are not trivial for a young store: “At least 500 customers have placed an order.”

“You have at least 180 days of order history and have orders within the last 30 days.”

“You have at least some customers who have placed 3 or more orders.”

Plus an ecommerce integration, or the API sending orders. The same conditions gate segmentation by CLV.

What if a store is too new for that?

Then it has the backward half and nothing else, which is an honest position rather than a broken one.

Omnisend’s gate is lower — “CLV data is only visible for contacts who have placed at least one order.” — and its surface is narrower: “There is no aggregate report showing total CLV or average CLV across all customers.” Nor can you segment on it: “It’s not currently possible to segment customers based on CLV statistics, but we plan to add this option in the future.”

A store below those thresholds can still work out an average order value and a purchase frequency by hand. What it cannot do is borrow a prediction its data has not earned.

Which number answers which question?

What should you check before comparing your figure to anything?

Three things, in this order.

Whether it is revenue or profit. Bloomreach’s own example puts that gap at 2.44 times.

What horizon it covers. A dashboard figure covering one year, held against a guide’s figure covering seven, is not a comparison.

What counted as an order. Refunds netted in, refunds dropped, or shipping and tax bundled — three answers across three references.

When does the definition stop mattering?

When you hold one of them still.

Customer lifetime value is worth far more against your own store last quarter than against any published figure. What breaks a comparison is a change of basis, not a change of label.

Pick a definition. Write down its horizon and whether it carries margin. Then keep it.