Understanding lifetime value (LTV)

The Lifetime value page in Margins groups your customers by the month they first bought from you, then follows each group forward to show what they went on to spend, what they cost to win, and what they were really worth.

Most reports tell you what happened last month. This one tells you what a customer is worth over their whole relationship with you, which is a different question and a more useful one. It's the number that decides how much you can afford to spend on acquisition.

You'll find it in the Lifetime value tab on your Dashboards page.

This article covers what lifetime value is, the questions the page answers, how to read the grid, and every metric on it: what it shows, why it's useful, and exactly how we calculate it.
​

You'll find it in your Lifetime value tab of your Dashboards.

Lifetime value is what a customer is worth across their whole relationship with you, not what they spent the first time they purchased.
​
Here's why the whole relationship matters, in numbers:

If you only look at that first order, you'll decide you can spend about $20 to win a customer. Look at the whole relationship and you can comfortably spend $70, and outspend every competitor who's still only counting the first order.

To answer that question properly, we group your customers into cohorts: everyone whose first-ever purchase happened in the same month. Your January cohort, your February cohort, and so on. We then follow each cohort forward, month by month.

Grouping by first-purchase month is what makes the comparison fair. If you simply looked at revenue per customer across your whole customer base, a strong month of new customers would drag the average down, not because anything got worse, but because you've just added a lot of people who have only bought once so far. Cohorts remove that distortion.

It sets your acquisition ceiling in money you actually keep. Revenue-based LTV overstates what a customer is worth, because it hasn't paid for the goods, the shipping or the returns yet. True LTV has. It's the honest ceiling on what you can pay to win a customer.

It catches a margin problem your P&L will report late. A monthly P&L blends every customer you have. If the customers you started winning in March cost more to serve or return more often, that shows up in a cohort within weeks and in your annual numbers much later.

It tells you when marketing spend comes back as cash. Two channels with identical lifetime value are not equally affordable if one repays you in two months and the other in eleven. That difference is a working capital question, and it's usually the constraint that actually binds.

It separates a volume problem from a quality problem. Flat revenue can mean fewer new customers or worse ones. Those have opposite fixes, and reading down a cohort column tells you which you have.

Rows are cohorts based on the month the made their first purchase ①. Columns are ages, not dates ②. M0 is a cohort's own first month, M1 the month after. So the M0 column holds January for your January cohort and February for your February cohort, lined up so you can compare cohorts at the same point in their life.

That's why the comparison to make is down a column, never across a row. Reading across tells you about one cohort's history. Reading down tells you whether your business is improving.

Up to four reference columns sit before the ages:

Overall closes each row with the cohort's whole life so far, and what it does depends on which metric you're looking at. It either adds the months up, takes where a running total finished, or pools every month together. Hover any Overall cell and the tooltip tells you which.

Cells show a dash where there is nothing to report. Most often that is a month the cohort hasn't lived yet, which is why the bottom right of the grid is empty, your newest cohorts are only a month or two old. But a dash also appears inside a cohort's own lifetime where it simply had no activity that month: no orders means no figure rather than a zero.

Read a dash as "nothing to show here", and use the row's own age to tell the two apart, everything to the right of a cohort's last lived month is the first kind.

Running totals are the exception and behave as you'd expect: Gross LTV, True LTV, Cumulative orders and the payback metrics carry the previous month's total forward through a quiet month rather than breaking the series.

Average is a weighted average, so larger cohorts count for more and the figure describes your business rather than treating a small month and a large one as equals. On running-total metrics it accumulates the average monthly gain, which means it describes the path a typical cohort follows rather than a figure any single cohort has banked yet. Hover any Average cell and the tooltip names the blend.

Total appears only where summing a column means something, on whole-cohort money and order counts.

Cells are shaded from pale to dark by rank within the view you're currently looking at. The scale always uses its full range, so shading is relative, never absolute. Compare shades inside one grid; never between two.

Nineteen metrics, in five groups. Pick one from Metric and the whole grid switches.

These five compare everything against what you paid to win the cohort. Where a cohort has none, the three ratios, Gross LTV:CAC, True LTV:CAC and Return on investment, stay blank rather than dividing by zero and reporting an infinite return.

The two dollar metrics behave differently, and it matters. Cumulative contribution profit and CAC payback subtract whatever marketing we found, so a cohort with none recorded is costed at zero and reads as pure profit. That is the right answer for a month you genuinely ran no ads, and a badly flattering one for a month whose spend simply hasn't been attributed.

A customer joins the cohort of the month their first purchase landed, and membership is decided once. Everything they buy afterwards counts, however it ships.

Only your direct and marketplace sales are on this page. Cohorts are built from your DTC and marketplace channels, your own storefronts and the marketplaces you sell on. Wholesale and other B2B sales, in-person and point-of-sale, and service income are all outside it, both for deciding who joins a cohort and for counting what they spend later. If a meaningful share of your revenue is wholesale or in-store, this page is describing part of your business, not all of it. Your P&L is the one that covers everything.

Only buyers we can identify can join a cohort. A cohort is a promise to follow the same people forward, so an order that arrives without a buyer attached can't take part. That covers guest checkouts and any channel that doesn't hand back a customer identity. The clearest case is Amazon, which identifies the buyer on Fulfilled by Amazon (FBA) orders but not on Fulfilled by Merchant (FBM)
ones, so a customer who has only ever bought through FBM is left out. Including any of these would drag every per-customer figure down, since they could never appear in a later month.

"First purchase" means the first one we can see. A customer counts as new when we hold no earlier order for them. At the very start of your history that isn't the same as their first-ever order: customers who were already buying from you before your data begins are counted as new in the first months we have. So your earliest cohorts are larger than they should be, and their retention and repurchase rates read low. Treat the first few cohorts on a long period as a warm-up rather than a result.

Everything is counted by when your customer ordered, not when the transaction reached your books. A cohort column is a statement about customer behavior, so an order belongs to the month it was placed. This is the single biggest reason these figures won't line up with a P&L month, and section 6 covers the rest.

The same method as your Channel Performance Dashboard:

Returns are your actual returns rather than an estimate, landing against the order they reverse. Delivery rates are measured per store, per month, on a three-month window centered on the month in question, so a cohort's older months are costed at the rates that applied then rather than today's. If you've corrected your selling, affiliate, FBA or Fulfilled by TikTok (FBT) rates in Channel Performance, those corrections apply here too. To learn more about how those costs are built, see Channel Performance Dashboard.

Returns keep arriving after a month closes. About 60% of a month's returns are in by the time it ends and about 94% a month later, so costing a just-finished month on what's visible today would count all of its sales against only some of its returns, and your newest month would look like your most profitable one every single month. For months that haven't finished settling we apply what your settled months measured for that store. Once the month settles, the real figures take over.

Acquisition cost is ad spend split across your stores, affiliate commission at that store's recent rate, and any giveaway cost you've booked, all charged to the month you won the cohort. All of a month's marketing is charged to that month's new customers, with none set aside for retention, which makes your CAC slightly cautious if you run retention campaigns. Marketplace ad spend, and any spend we can't match to a store, is charged to nobody rather than spread across cohorts that didn't earn it.

Stores and SKUs filter on how a customer was won, not on what they bought later. Pick a store and you get the customers whose first order was on it, then everything they went on to buy anywhere. Filtering their later orders too would make your retention look worse than it is. The period you choose sets both which cohorts appear and how long each is tracked, and it always ends at the last completed month.

These figures are built from your booked numbers, but they won't reconcile line-for-line to a P&L period, and they aren't meant to. Four differences account for the gap.
​

Different clock. Lifetime value counts an order in the month your customer placed it. Your P&L recognizes revenue on Finaloop's revenue recognition rules, which is a different question with a different answer. A cohort month and a P&L month are not the same window.

Different depth. Gross profit here stops after product cost and delivery costs. It's before your overheads, salaries, software, rent and every other fixed cost. A cohort that looks healthy on True LTV is not necessarily a profitable business once everything else is paid. Only your P&L answers that.

Different treatment of a month that hasn't settled. For recent months, lifetime value applies your store's settled return rate rather than the incomplete returns visible so far. Your P&L records refunds as they're booked. So the newest month on this page is deliberately more conservative than the same month on your P&L.

Different scope of marketing. CAC here excludes marketplace ad spend and any spend that can't be matched to a store. Your P&L includes all of it as an expense. This is also why the figure agrees with Channel Performance Dashboard, which leaves out the same spend.