Customer lifetime value for a small Shopify store

Customer lifetime value gets asked about more than it gets worked out. This guide separates the two things the phrase can mean, shows what Shopify reports, explains why forecasts need more history than a small store has, and works through how counting profiles instead of people changes the figure.

Two meanings: spend to date and a forecast

Customer lifetime value is used for two different things.

• Spend to date. What a customer has spent with your store so far. It is a record of the past, and you can calculate it from your orders.
• A forecast. What a customer is expected to spend in future. It needs a model of future buying.

The research on forecasting is open about why it is hard. Fader, Hardie and Lee begin from the point that managers want to predict customers’ future purchasing, “which can then serve as an input into ‘lifetime value’ calculations” (abstract). A forecast is a model’s estimate, and it needs enough history to be worth using.

Shopify’s Help Center doesn’t define “customer lifetime value”. Its Customers reports page has no formula for it. What Shopify does give you are the pieces below, and you assemble the figure.

What Shopify reports

Three places show spend per customer, and each has its own scope.

All three are per customer record, which is per profile. Shopify’s Customer cohort analysis report, which this guide calls the retention report, groups customers by the month of their first order, and its “Amount spent per customer” metric is the one Shopify’s changelog describes as giving “insight into your customers’ overall value to your business”.

Why forecasts need history

Shopify, Klaviyo and the research literature all put limits on forecasting. These limits were checked on October 24, 2026.

• Shopify spend projections. Set the metric to Amount spent per customer and turn on Show projections. Shopify says: “If 24 months of data aren’t available, then the Show projections toggle won’t display”. Predictions are compiled over each group’s previous 24 months, and “Projections aren’t a guarantee of future sales.”
• Shopify predicted spend tier. The tier (High, Medium or Low) estimates spending potential from how often a customer buys, how much they spend per order against the store average, how many orders they placed and how recently. Shopify says the store must have made more than a hundred sales. It doesn’t publish where the tiers start and stop.
• Klaviyo predictions. Klaviyo asks for at least 500 customers who have placed an order, at least 180 days of order history with orders in the last 30 days, and some customers who have placed 3 or more orders. It adds that “predictions work best when averaged over many customers and are not expected to be exact for any single individual.”

A store with under 24 months of orders sits below Shopify’s line, and one with fewer than 500 customers who have ordered sits below Klaviyo’s. For such a store, spend to date is the figure to trust, and a forecast is a rough guide.

Per profile or per person

Spend per customer divides revenue by a count of customers. Shopify’s count is of profiles. When one person holds two profiles, the same revenue is divided by one more customer than there are people, and the figure comes out low.

Our reasoning, not a Shopify statement: let p be the number of profiles and n the number of real people, with n no larger than p. Revenue and orders don’t change. Revenue per customer is revenue ÷ p by profile and revenue ÷ n by person, so the profile figure is understated by 1 − n ÷ p. Orders per customer is understated by the same fraction. Average order value (revenue ÷ orders) uses no customer count, so duplicates don’t move it.

Here is a worked example. It is invented to show the arithmetic.

Revenue is the same in both views. Only the count of customers differs, and the profile view reads lower by the share of profiles that are surplus.

MergeGuard’s demo store shows the same split on one customer: Amelia Hughes: 3 profiles with 4, 3 and 2 orders → one customer with 9 orders Demo store (Northsail Supply). Counted by profile she is three modest customers. Counted as a person, her nine orders sit on one record.

What the formula says about big claims

Some pages claim that duplicate profiles understate customer value by a stated percentage. The formula shows what such a claim needs. An understatement of a fifth would need one profile in five to be surplus, and an understatement of a tenth would need one in ten. We found no source that measures how common duplicate customers are on Shopify stores, so we give the formula and leave out a range. Your own count of profiles and people is the evidence that matters.

Refunds

We don’t state how refunds enter the amount spent. Shopify’s API description of amountSpent doesn’t mention refunds, and neither does the segment reference for amount_spent. To find out for your store, pick a customer with a known refunded order and compare their amount spent with the sum of their orders. Write the result next to your window, because it decides whether returns lower the figure.

Reading your result

The steps at the top give you spend per person for one window. Run them twice a year, with the same window each time, so the numbers compare. If the profile figure and the person figure are close, your duplicates are few and the profile report is a fair guide. If they are far apart, use the person figure for decisions such as who counts as a VIP, and read how duplicates change your repeat purchase rate for the same gap on repeat buying.

Where this leads

Customer value feeds decisions about who deserves perks. How to set a VIP threshold turns spend per person into a cut-off you can defend. Reading Shopify’s customer retention report on a small store shows how far that report can go when you have little history.

Counting people for reporting doesn’t require merging anyone. A Shopify merge is permanent, so keep counting and merging as two separate decisions.