Amazon customer lifetime value: how to calculate LTV you can trust

Amazon customer lifetime value (LTV) is the money a group of new customers brings in over a fixed time after their first order. To calculate it, group customers by the month they first bought, and add up margin per customer at fixed ages like 90, 180 and 365 days.

Most guides stop at one formula: average order value times purchases per year times years as a customer. That number is easy to compute and almost always too high. On Amazon it also hides a bigger problem. You do not know who your customers are, so even counting them takes care.

Below: why Amazon LTV is hard, what Amazon gives you to work with, a worked cohort example, and the traps that inflate the number.

Why LTV on Amazon is harder than on your own store

On your own store, every order has the shopper’s real email. You can see that the same person bought in March, May and August.

Amazon does not give you that. The buyer email in your order data is an alias. Amazon’s Selling Partner API (the feed that sends order data to software) describes the field as “the anonymized email address of the buyer” (Orders API model). Amazon staff have told sellers that each buyer gets an anonymized alias for each seller relationship (Amazon Seller Forums post).

That alias still works for LTV inside Amazon. Because it is tied to one buyer and one seller, you can link that buyer’s orders with you over time. What it cannot do is match that person to your Shopify store, your email list or an ad platform.

Two more things show up in real data:

• The buyer field arrives late. In the data we see, it often fills in some time after the order. A fresh order may have no buyer ID yet.
• It can be missing for stretches. We have seen whole months where most orders came through with no buyer ID, next to months that were fine.

Revenue, orders and units do not need the buyer ID. Anything that counts people does: customers, new versus returning, repeat rate, LTV and CAC (the cost to get a new customer). So every Amazon LTV starts with a coverage check, covered in the steps below.

What Amazon gives you to work with

Amazon offers three reports that touch customer value. None of them is a cohort LTV, but each is useful next to one.

New-to-brand metrics in Amazon Ads

Amazon Ads counts an order as new-to-brand when the shopper has not bought from your brand in the past 12 months (Amazon Ads new-to-brand metrics). Reports show new-to-brand orders, sales and their share of the total.

Two limits. These are ad-attributed orders only, not every new customer. And Amazon says new-to-brand data covers Sponsored Brands and Sponsored Display, not Sponsored Products (the ads that promote one product in search results). In the ad accounts we see, Sponsored Products is often most of the spend.

Amazon also offers long-term sales, an estimate of the extra sales a campaign can bring your brand over the next 12 months (Amazon Ads long-term sales). Amazon says it is based on the past 12-month return of similar shopper actions. Treat it as Amazon’s model, not a measured LTV.

Brand Analytics repeat purchase report

Brand Registry is Amazon’s program for brand owners, and Brand Analytics is its set of reports for them. Registered brands can pull a repeat purchase report. Per product and per week, month or quarter, it shows orders, unique customers, the share of repeat customers, and repeat purchase revenue and its share (SP-API analytics report types).

It is a good check on your own numbers. But each row covers one period. It does not follow a group of first-time buyers forward, so it cannot tell you what a January customer is worth by July.

Subscribe & Save metrics

If you sell through Subscribe & Save, Amazon’s Replenishment API returns subscription metrics (Replenishment API reference). They include active subscriptions, subscription revenue, and the share of subscriptions still active 30 and 90 days after they start. There is also average revenue per subscriber and non-subscriber over the past 12 months. A subscriber lifetime value metric shows total spend on your catalog over the past 24 months.

This tells you how much subscribers are worth next to one-time buyers. It covers subscribers only, and it is revenue, not margin.

The right way: cohorts at fixed ages

A cohort is a group of customers whose first order fell in the same month. Cohort LTV answers a plain question: how much margin did the average customer in this group bring within N days of their first order?

Four rules make the number trustworthy:

1. Use first purchase ever. A customer is new only if they have never bought from you before, not just new in the window you picked.
2. Use contribution margin, not revenue. Contribution margin is what each order leaves after product cost, Amazon fees and fulfillment. See contribution margin for Amazon and Shopify for how to build it.
3. Measure at fixed ages. 30, 60, 90, 180 and 365 days after the first order. Include the first order itself.
4. Only fill in an age a cohort has fully reached. The last customer in the cohort must have had the full N days. Anything younger stays blank.

A worked example

Illustrative numbers only. A brand sells body lotion on Amazon. Each $22 order leaves about $7 of contribution margin. That is after product cost, the referral fee (Amazon’s commission on each sale) and the FBA fee (Amazon’s charge to pick, pack and ship). Closed data runs through August 31, 2026.

Each cell is cumulative contribution margin per customer, first order included.

The June cohort’s 90-day cell is blank. A customer who first bought on June 30 reaches 90 days on September 28. Filling that cell now would mean guessing. A guess placed in the same table as measured numbers gets read as measured.

What the table does tell you:

• Newer cohorts can be compared at the same age. At 60 days, January reached $9.20 and June $8.40. That is a fair comparison.
• July starts lower. It was a deal month. Discounted first orders leave less margin, and deal shoppers may come back less. Watch its 60-day cell before drawing a conclusion.
• The 365-day number comes from one cohort. It is a single reading, not a trend.

Now compare the simple formula. Average order value of $22, times 2.5 orders a year, times 3 years, gives a “lifetime value” of $165. That is revenue, over a lifespan nobody measured. The measured 365-day margin is $17.40.

Compare LTV to CAC at the same age

LTV to CAC compares what a customer is worth with what it cost to get them. Both sides need the same basis.

Say the brand estimates it spends $12 in ads per new Amazon customer. The August 2025 cohort also brought in $55 of revenue per customer by 365 days. Using the table:

The same customers read anywhere from 0.86x to 4.58x. The revenue line looks best and says least, because it ignores the cost of the product and Amazon’s fees. The margin lines show when the ad money comes back: somewhere between 90 and 180 days.

There is no industry number that makes a ratio good. A brand with cheap cash and a long view can live with a slow payback. A brand that needs its money back in a quarter cannot. Set your own target, write down the age it applies to, and track your cohorts against it.

Amazon CAC is itself an estimate. Total ad spend divided by new customers counts spend that also reached repeat buyers. It also credits ads with new customers who found you on their own. Say which version you use, and keep it the same every month.

Where it goes wrong

These are the traps we see most often in real Amazon customer data.

The simple formula overstates LTV

Average order value times frequency times lifespan has three problems. It uses revenue, not margin. Frequency often comes from your best, most active customers. And lifespan is a guess, often several years for a brand that has not existed that long. Each one pushes the number up.

Counting repeat rate before a cohort is old enough

A customer who first bought in July has had two months at most to reorder. If a bottle lasts 90 days, almost none of them have run out. Their repeat rate looks terrible. It is not low, it is unfinished. Compare cohorts only at an age every one of them has reached.

Customer counts that dip in the last few weeks

Because the buyer field fills in late, recent weeks show fewer customers than there really were. Orders per customer then seems to jump. Nothing changed in how people shop. Leave recent weeks out of any customer count until their buyer coverage stops rising, or label them as still filling in. This is also one of the most common ways AI tools get ecommerce numbers wrong. A model can invent a story about shopper behavior to explain a dip that is only missing data.

A missing month is worse. It will not fill in by waiting. It shows up as a month with normal orders and very few customers. Customers who only bought in that month then vanish from your cohorts.

Gifts and shared household accounts

One Amazon account often buys for a whole household. Gifts ship to someone who never appears in your data. So an “Amazon customer” is really an account, and one account can stand for several people. There is no fix for this in Amazon data. It is a limit to state, not a problem to solve.

Revenue LTV against CAC

CAC is paid in real dollars. Revenue LTV includes the cost of the product and Amazon’s cut, which you never keep. Dividing one by the other makes every channel look profitable. Use contribution margin before advertising on the LTV side, since ad spend is already on the CAC side.

Comparing ages that do not match

A 365-day LTV against CAC from last month’s ads mixes a mature cohort with today’s costs. A 60-day LTV against a payback target set for a full year mixes the other way. Match the age, and match the time period the CAC came from.

How to calculate Amazon LTV, step by step

1. Check buyer coverage per month. For each of the last 18 months, find the share of orders with no buyer ID. Look for months far above their neighbours. Check per month, because an all-time average hides a bad month.
2. Exclude what is not ready. Mark months with poor coverage as unusable for customer counts. Leave out recent weeks whose coverage is still rising.
3. Assign every customer a first-order month. Use their first order ever, across all the history you hold. Keep cancelled orders out.
4. Attach contribution margin to every order. Subtract product cost, referral fees, FBA fees and refunds. Leave advertising out for now.
5. Build the table at fixed ages. Cumulative margin per customer at 30, 60, 90, 180 and 365 days. Leave a cell blank until the whole cohort has reached that age.
6. Compare with CAC at a stated age. Pick the age that matches how fast you need your money back. Put CAC from the same months next to it.
7. Say how old every number is. “180-day margin LTV, January 2026 cohort, 2,400 customers” is a number someone can act on. “Our LTV is $165” is not.

Synthesis labels every customer number with its caveats. Two examples: “this cohort is too young to judge repeat rate” and “Amazon buyer emails arrive late so recent customer counts run low.” LTV is never stored as a fact. It is computed fresh from the brand’s own data each time it is asked for.