LTV on marketplaces: what you can actually measure on Amazon and MercadoLibre

LTV — lifetime value — is the metric that decides whether your acquisition cost makes sense. Without it, knowing that a new customer costs you 258 pesos means nothing: it could be a bargain or a slow bleed, and the difference is how much that person leaves behind before disappearing.

The trouble is that LTV was invented for businesses that know their customers. An owned store knows that Ana’s email bought three times in fourteen months and spent 4,200 pesos. A subscription service knows exactly how many months an average user lasts. On a marketplace you have none of that: Amazon hands you an encrypted alias instead of an email, and MercadoLibre gives you what you need to ship and invoice under the condition — stated in its own privacy notice — that you not use that data for a purpose other than the one that produced the transaction.

Faced with that, there are two wrong reactions. One is to give up and measure nothing, which leaves the business setting budgets blind. The other is to invent an LTV with a tidy formula from the internet, apply it to data that cannot support it and believe the result — worse, because it manufactures confidence.

There is a middle path, and it is what this article is about: identify exactly what Amazon and MercadoLibre give you, estimate an honest figure from that, and — this is the key part — report it with its assumptions written beside it, so nobody mistakes it for an exact measurement.

what LTV is and why it is usually calculated wrong

In its most useful form, LTV multiplies three things: what one purchase leaves you, how many purchases that person makes, and over what period. The simple version that supports decisions is: contribution margin per order × orders per customer within a defined time window.

The two decisions that ruin the calculation are hidden in there.

The first is using revenue instead of margin. An LTV of “3,400 pesos in sales” cannot be compared against a CAC, because product cost, referral fee, fulfillment, shipping and tax all come out of those 3,400. The LTV you compare against acquisition cost has to be expressed in margin, not in top-line sales. It is the most common mistake and the one that inflates results the most.

The second is skipping the time window. “Lifetime” sounds like forever, and forever can neither be measured nor banked. A 12-month LTV is verifiable and actionable. A “lifetime” LTV is a projection built on assumptions nobody revisits. If your business is two years old, you cannot report a five-year LTV.

Glossary: real net margin, with commissions, fees, shipping, tax and advertising already deducted →

the underlying constraint: no identity, no curve

Every LTV calculation rests on being able to follow the same person over time. That tracking is precisely what the marketplace does not hand over, and not out of negligence: the buyer is the channel’s asset, and the relationship is managed from there.

It is worth sizing what that blindness costs, because there is evidence. Bluecore, a retail platform that publishes a benchmark report built from more than a hundred large retailers — US-based and direct-to-consumer, not marketplace sellers — found that retailers identifying more than 40% of their shoppers posted repeat purchase rates 53% above the average, while those identifying under 10% came in 33% below it. In other words, identifying the buyer is not only a measurement question: it is also what enables the actions that produce repeat purchase.

A marketplace seller lives structurally on the wrong side of that curve. It is not their fault and it cannot be fixed inside the channel. What can be fixed is the measurement side: reports exist that, without giving you individual identity, give you enough aggregate behavior to estimate.

what you can actually measure on Amazon

Amazon is by some distance the channel that shows you the most. With Brand Registry enrollment, two pieces are useful.

The first is new-to-brand metrics, which classify an order as new-to-brand when the buyer purchased no product from your brand on Amazon during the previous twelve months. That lets you split your sales into acquisition and repeat. They report natively in Sponsored Brands and Sponsored Display, not in Sponsored Products.

The second sits in Brand Analytics: the repeat purchase behavior report — how many unique buyers purchase more than once, what share of your orders and sales comes from returning buyers — and the loyalty dashboard, which segments buyers into groups such as new, active, at risk and loyal, and estimates a value per segment.

Two honest caveats about these reports. Amazon has repeatedly changed what is available in each marketplace and under what conditions, so confirm in Seller Central what is actually enabled on your Amazon Mexico account before building a report on top of it. And the value estimate Amazon offers is Amazon’s, with its methodology and on its terms: treat it as a directional signal, not as the number you take to your bank.

what you can actually measure on MercadoLibre

There is less here, and it deserves saying plainly: MercadoLibre publishes no equivalent to new-to-brand metrics and no lifetime-value dashboard. Product Ads reports spend, clicks and attributed sales; it does not separate first-time from repeat buyers.

What you do have are your own orders. Each one arrives with the buyer identified by nickname and with the data needed to ship and invoice. If you store that identifier consistently, you can build your own repeat purchase rate over time: what share of January’s buyers bought from you again before December. It is not a usable marketing identity — Mercado Libre’s privacy notice is explicit that a seller receiving buyer data becomes the controller of it and cannot use it for a purpose different from or incompatible with the transaction — but for aggregate measurement it is enough.

One additional limitation should not be buried: one person can buy from more than one account, and one account can be shared. A repeat rate measured this way is a floor, not an exact figure. Reporting it as a floor is correct; reporting it as precision is a lie.

what the public evidence says about repeat purchase

Since LTV depends on repeat purchase, it is worth knowing what has been measured. And again: there is no public repeat-purchase study for Mexican marketplace sellers. Everything available comes from owned ecommerce stores, mostly in the United States, and should be read with that label attached.

• The average that recurs most across independent studies is around 28% of buyers returning to purchase again at an ecommerce store, with an enormous spread by category: roughly 10% in luxury goods and above 65% in grocery. The logic is what you would expect: things consumed quickly get rebought, things that last years do not.
• The Bluecore report cited above, covering more than a hundred large retailers, found a considerably lower average: 16.5%, led by health and beauty (21.5%), then sporting goods (21.2%) and apparel (20.2%).

Two serious sources reporting 28% and 16.5% does not mean one is wrong: they measure different universes over different windows. Which is why these figures are good for exactly one thing — ranking categories and setting expectations of magnitude — and not for setting a target. If you sell consumables, expect high repeat; if you sell furniture, expect very low, and your LTV will depend almost entirely on the first purchase.

The other reference you will see quoted is the 3:1 LTV:CAC ratio. It comes from the SaaS and venture capital world, not from a marketplace measurement. As a unit-economics compass it works: if a customer leaves less than three times what it cost to bring them in, paid growth gets fragile. As a benchmark for your category, it is not one.

how to build an honest estimate

With all that, a defensible estimate takes four steps and fits on one page.

• Fix the window. Twelve months. Long enough to capture repeat purchase, short enough to verify with data you already have.
• Use a cohort, not an average. The new buyers of one specific month. An average across your whole base mixes customers from three years ago with last week’s and tells you nothing.
• Measure in contribution margin, not in sales. What is left after product, referral fee, fulfillment fee, shipping and tax.
• Write the assumptions next to the number. “12-month estimate, March cohort, identification by nickname, Amazon via new-to-brand and MercadoLibre via nickname proxy.” A number with visible assumptions can be argued with; one without them can only be believed.

A worked example closes the idea. If your product’s contribution margin is 190 pesos per order and the cohort bought an average of 1.4 times in twelve months, your 12-month LTV is 266 pesos. Against a CAC of 258, the ratio is 1.03 to 1: the business recovers acquisition and almost nothing else. Get that cohort to 2.1 purchases and LTV rises to 399, a ratio of 1.55. To approach the 3-to-1 the industry treats as a floor you would need close to 4 orders per buyer, or a materially higher margin per order. That calculation, not an external benchmark, is what tells you whether your advertising can grow.

Glossary: unified catalog, one real product with its listings linked; without it you cannot add up repeat purchase for the same brand →

how this reads in iqseller

LTV does not come out of one module: it comes from correctly joining two things that usually live apart, margin per order and catalog structure.

The Profitability module provides the first: net margin per SKU with the COGS you loaded, commissions from the Amazon settlement and from MercadoLibre orders, FBA and Full fees, shipping, and the VAT and withholding breakdown. That is the contribution margin your repeat rate multiplies; using list price instead is the fastest way to inflate an LTV until it stops being useful.

The second comes from the Catalog, with its Parent → Model → SKU tree. A buyer who bought the small size on Amazon and the large one on MercadoLibre is repeat purchase for your brand, but you only see it if both listings hang from the same parent. Without that structure, every variant looks like a separate product and brand-level repeat purchase vanishes from the report. It is, in practice, the technical prerequisite for any multichannel LTV to mean anything.

what not to try to measure

Let’s close with the part almost nobody writes down. Some things simply cannot be measured on a marketplace, and accepting the limit is part of measuring well.

You cannot build an individual lifetime curve per buyer, because you do not have the buyer. You cannot attribute an Amazon repurchase to a MercadoLibre ad, or the other way around. You cannot know whether a customer who stopped buying left for a competitor or simply stopped needing the product. And you cannot measure the lifetime value of a shopper who found you on the marketplace and now buys from your own store, unless they tell you so themselves.

What you can do is measure cohorts, measure in margin, measure per channel and keep the series month after month. Three or four cohorts already give you a trend, and your own trend is worth more than any industry average you have read on a blog — this one included.