Short answer: revenue for most of what you do, margin once the decision involves spending money against the number.
Longer answer, because the gap between them is bigger than most people expect.
The easy calculation
How long a customer stays, multiplied by what they spend while they're with you. That's the published definition and it's the one almost everybody uses.
A customer pays $1,000 a month. They stay 36 months. $36,000 in lifetime value. Clean, fast, and directionally right for nearly every question you would ask of it.
The accurate calculation
Now put the cost of serving them next to it.
Say that same customer costs $900 a month to deliver. Support, infrastructure, the humans involved, whatever it genuinely takes to keep them successful. They still stay 36 months. They still pay you $36,000.
They're worth $3,600.
Same customer. Same tenure. Same invoices. One number is ten times the other, and only one of them is money you keep.
That's revenue LTV against margin LTV, and it's the difference between what a customer pays you and what a customer is worth to you.
Before anyone panics
That example is deliberately brutal. A ninety percent cost to serve isn't a software business, it's a staffing agency with a login screen.
In most software companies the cost to serve, setting aside what you spent to acquire the customer in the first place, runs somewhere in the range of 10% to 30%. Not nothing. Not catastrophic. Enough that the two numbers differ meaningfully without differing by an order of magnitude.
Which is why revenue-based LTV survives as the default. At those margins it's wrong in a consistent direction by a consistent amount, and a number that's consistently wrong in a known direction is still useful for comparison.
When the difference starts to matter
Revenue LTV is fine when you're comparing customers to each other, watching whether a cohort is improving, or sizing what your base could be producing.
Margin LTV becomes necessary the moment the number justifies spending. What you can afford to pay to acquire a customer. Whether a segment is worth serving at all. Whether the enterprise accounts everyone is proud of are actually carrying the company or quietly being carried by it.
That last one catches people. Large accounts frequently look extraordinary on revenue LTV and merely fine on margin LTV, because the things that made them large also made them expensive: the custom work, the dedicated attention, the integration nobody else asked for. The revenue view can't see any of that.
Every company measures this differently
Some include only direct cost of revenue. Some load in a share of the success team. Some apply gross margin as a flat percentage across the base and call it done, which is fast and hides the exact variation you would want to find.
There's no settled answer, and pretending otherwise would be its own kind of wrong. What matters is that you know which one you're holding when you quote it, and that you don't compare your margin-based figure to somebody else's revenue-based one and conclude anything at all.
The part that changes the argument
Here's what the margin view does that the revenue view can't, and it points the opposite way from what people expect.
Revenue LTV treats every dollar as identical. A dollar from a new customer and a dollar from an existing one land in the same column and look the same.
They aren't the same. The expansion dollar arrives with no acquisition cost attached, because that was already paid. It usually lands on infrastructure that already exists and a relationship somebody is already maintaining. The marginal cost of the second sale to a customer is nearly always lower than the marginal cost of the first sale to a stranger.
So when you switch from revenue to margin, expansion revenue doesn't merely hold its position. It improves it, relative to new-logo revenue, by more than the revenue view ever showed you.
The honest caveat: this isn't automatic. If the thing you expand into is services-heavy, or requires real delivery work per account, it carries genuine incremental cost and you should check rather than assume. Capacity, tiers, and features tend to be nearly pure margin. Implementation and managed services aren't.
What to do about it
Keep using revenue LTV for the everyday. It's easier to calculate, easier to explain, and good enough for most of the questions you'll ask this quarter.
Build the margin version once, properly, for the decisions that spend money against it. Then look at where your margin actually comes from, split by whether the dollar was a first sale or a later one.
Most companies have never run that split. The ones that do tend to discover that the cheapest, highest-margin revenue in the business is the revenue nobody in the organization is responsible for producing.
Which is a separate problem, and a larger one.
Next Steps
The full metric set, and what each one actually responds to, is on the expansion metrics reference.
The number underneath all of it is what your existing base should be producing and isn't. That's latent revenue, and finding roughly what yours is takes about ninety seconds.
Lincoln Murphy formally named and popularized Customer Success starting in 2010 and has spent 15 years connecting it to expansion revenue and commercial outcomes. Read The Premise.