Shopify customer lifetime value: a worked Shopify study

One example customer

Example only. linen throw at $68, reordered 1.5 times a year for 2 years.

Revenue. $68 × 1.5 × 2 = $204
Contribution before ads, if every order has the same product and shipping cost. $43 × 1.5 × 2 = $129
First-order contribution before ads. $43

The gap between $43 and $129 is the room to pay more for the first order than the first order returns. It disappears if the repeat rate is a guess, or if you buy every later order with ads as well. Average order value is already in Shopify’s analytics. Frequency is not a setting. It is what your customers did, and a store that launched this year cannot honestly type a multi-year life. The lifetime value calculator will multiply whatever frequency you type. It will not check it.

One multiplication, invented to be extended

This is one invented set of inputs. It is not a cohort, not a store, and not a customer. Say the average order value you type is 54 dollars. Say orders per year is 3. Say years is 2. The calculator's output is revenue, and it assumes exactly the repeat and the lifespan you just typed. No cost is included. If you need a different story, change one input and say you changed it. Do not add a second customer beside this one.

The numbers are example inputs so the product can be checked by hand. They are not a typical order value, not a typical repeat rate, and not a recommended lifespan. Nothing about them should be copied into a media plan as a default.

The base product

Multiply 54 by 3 first. That is 162 dollars of revenue per year under these inputs. Multiply 162 by 2 years. That is 324 dollars of revenue. Check it as 54 times 2, which is 108, then times 3, which is 324. Same revenue. The calculator should show 324 dollars if these are the inputs. It should not show 324 dollars of profit. There has been no subtraction.

Say the sentence in the form the output deserves. If a customer ordered 3 times a year for 2 years at 54 dollars of revenue per order, revenue would be 324 dollars. The word if carries the whole weight. Nothing in the example observed such a customer.

Change only the years

Change years from 2 to 1. Yearly revenue stays 162 dollars, because average order value and orders per year did not move. Total revenue becomes 162 dollars. The result halved because the formula multiplies by years. Customers did not 'churn.' You edited an input. Put years back to 2. Revenue returns to 324.

This is the sensitivity people use to inflate a story. Moving years from 1 to 2 doubles revenue without any new fact. In the example, that move is allowed only because we put it back. On a real sheet, a move like that needs the word assumed, or it needs a cohort that actually stayed. The example has no cohort.

Change only orders per year

With years restored to 2 and the average restored to 54, change orders per year from 3 to 1. Revenue is 54 times 1 times 2, which is 108 dollars. The drop from 324 to 108 is the repeat assumption leaving. Restore orders per year to 3. Revenue is 324 again.

Notice that 1 order per year for 2 years is 2 orders of revenue, not a lifetime in any rich sense. The formula does not add a bonus for loyalty. It counts orders. If someone describes 108 dollars as a relationship, they are adding a story the arithmetic does not contain. Keep 108, while it is on the table, labeled as the one-order-per-year variant of the same example.

Change only the average

Restore the base, then change only the average order value, from 54 dollars to 60 dollars. Revenue is 60 times 3 times 2, which is 360 dollars. The extra 6 dollars of average becomes 36 dollars of lifetime revenue because it is multiplied by 3 and by 2. That scaling is the formula, not a price test. Put the average back to 54. Revenue is 324 again.

The step is here so a price increase is not casually multiplied into a lifetime. You did not learn that customers will pay 60. You learned that the formula will gladly multiply 60. A real average has to come from orders. This 60 is an edit.

Try, and fail, to read the base as profit

The base is 324 dollars of revenue. Suppose someone subtracts an example product cost in their head and calls the result lifetime profit. They have left the calculator. Even a generous reading is wrong if it subtracts product once instead of once per order. Three orders a year for two years is 6 orders. Any cost you care about happens on those orders, or it does not. The lifetime calculator did not ask.

Do the refusal in the note: 324 is revenue. Costs not subtracted. A first order's profit is a different sheet and is not computed here. There is no ad cost in the example, so there is no acquisition story. Adding one would be a new invention. This study stays on the multiplication.

Why 324 is not a spending allowance

If someone offers to spend 324 dollars to acquire the customer in this example, they are offering to spend the entire revenue, with no product, no shipping, no fee, and no certainty that 3 orders a year for 2 years will happen. The offer is absurd inside the example, which is why it is useful. Smaller versions of the same offer, such as spending a large fraction of 324, have the same hole.

The allowed use of 324 is to compare with another multiplication that uses the same formula and the same honesty. It is not an allowable cost per purchase. It is not break-even ROAS. Break-even ROAS is a unit formula. This is a revenue product across assumed orders.

The note with the variants restored

The note lists the base: 54, 3, 2, revenue 324. It lists the one-year variant: revenue 162, then restored. It lists the one-order-per-year variant: revenue 108, then restored. It lists the 60 dollar average variant: revenue 360, then restored. It says costs were never subtracted. It says the inputs are invented.

A reader can rebuild every total. A reader cannot find a recommended budget, a cohort, or a claim that customers order three times a year. If a reader forwards only 324, they have broken the note. Send the rest of the inputs with it or do not send it.

What this multiplication must not become

Do not call 3 orders a year a benchmark. Do not call 2 years a standard lifespan. Do not call 54 dollars a typical average order value. Do not attach the serum, the candle, or the notebook from any other sheet to this 324. Those were different inventions. Combining them would fake a shop.

Do not extend the example into a subscription with a discount, a skip policy, or a churn curve. The formula you are studying has none of those. A more detailed model would be a different model, labeled as yours, and it would not be this calculator's output.

Rebuild 324 and stop

54 times 3 is 162. 162 times 2 is 324. Revenue, not profit. Assumptions are the three inputs themselves, since no cohort was observed. Stop. Another input change would only repeat the same multiplication. One tour is enough to see that the result moves in a straight line and never pays a cost.

When you do this on real orders, the steps are the same and the inputs are counted or marked assumed. The page for the multiplication is /lifetime-value/. The page for the average's meaning is /learn/average-order-value.html. Neither page turns 324 into a fact about your store, because 324 was never your store.

One more hand check, then leave the inputs

Write the base as a single line you can re-read without a screen: 54 times 3 times 2 equals 324 dollars of revenue. Then write the three variants you already restored: 1 year equals 162, 1 order per year equals 108, and a 60 dollar average equals 360. If any of those four totals disagrees with a fresh multiplication, fix the note before you trust your memory of it.

After that check, stop editing. The next honest number has to come from an export, not from another invented input. Three, two, and fifty-four are finished as a drill. They are not seeds for a forecast, and they are not mild defaults you may type when a cohort cell is blank.