Guides
Customer lifetime value Canadian ecommerce: an overview
Customer lifetime value Canadian ecommerce retailers report depends on the window, tax-exclusive margin, acquisition cost and customer-data limits.
What to take away
- Customer lifetime value is the gross margin one customer contributes over a set window, reported in CAD and stripped of sales tax.
- Act when the 12 month LTV to acquisition cost ratio drops below 3 to 1.
- A blended LTV cannot name which customers will leave, and a longer window flatters the number.
- PIPEDA caps how long you may keep and reuse customer data, which limits the history you can model.
Customer lifetime value Canadian ecommerce retailers report depends on a consistent window, tax-exclusive margin and customer records they can use. Most LTV guidance is written for US stores. It assumes one currency, one tax regime and no legal limit on customer records.
LTV in Canadian dollars, before tax
Customer lifetime value is the gross margin a customer contributes over a defined window, usually 12 or 24 months, after returns, refunds and fulfilment costs.
LTV components and errors
Component
- Gross margin per order
- Using revenue instead of margin
- Orders per customer
- Counting guest checkouts as new
- Window
- Extending until it looks acceptable
- Returns and refunds
- Left in a separate report
- Tax
- Treating collected tax as revenue
Frequent error
- Gross margin per order
- Orders per customer
- Window
- Returns and refunds
- Tax
US breakdowns often stop at revenue. In Canada, tax has to come out first. GST, HST and provincial sales tax sit in your revenue line until you remit them, and none of it belongs to you. The GST and HST rates that apply depend on the province and the product type.
| Component | Where it comes from | Frequent error |
|---|---|---|
| Gross margin per order | Product cost, shipping, payment fees | Using revenue instead of margin |
| Orders per customer | Order history tied to one customer ID | Counting guest checkouts as new buyers |
| Window | 12 or 24 months from first order | Extending it until the number looks acceptable |
| Returns and refunds | Refund records and chargebacks | Leaving them in a separate report |
| Tax | GST, HST, provincial sales tax | Treating collected tax as revenue |
Two limitations appear at once. The result is a blended average that describes nobody in particular. It also shifts whenever your discount calendar or product mix changes.
Reading the number against acquisition cost
LTV means something only beside the cost of winning the customer. Divide 12 month LTV by blended acquisition cost across every paid channel. A ratio of 3 to 1 is the working floor most operators use.
LTV to acquisition cost ratio
Is 12-month LTV divided by blended acquisition cost at least 3 to 1?
Maintain or raise acquisition spend
Stop raising spend, repair repeat purchase rate
When that ratio falls below 3 to 1, stop raising acquisition spend and repair repeat purchase rate first. More budget at a weaker ratio buys revenue you will not recover inside the window you measured.
A ratio is a decision rule, not a score. Below the floor, extra traffic widens the shortfall.
If your reports cannot produce margin by cohort, fix that before touching media budgets. The Ecommerce analytics review sets out which reports come first.
What the metric cannot tell you
It cannot name the customers who are about to stop buying. It also cannot separate genuine repeat demand from a promotion that pulled next quarter's orders forward.
A longer window raises LTV mechanically. Two teams can report very different figures for the same store by choosing 24 months instead of 12. Cross-team comparisons then mean nothing.
Blended LTV hides whether your strongest cohort is shrinking while a weaker one grows in its place.
Repeat purchase behaviour sits underneath that average. The ecommerce retention marketing guide covers the patterns behind it.
Attribution gaps that distort repeat revenue
Email and paid social both claim the same repeat order. Post-purchase email gets credit for a sale the customer would have made anyway. Platform dashboards treat a repeat purchase as a win for whichever channel touched the customer last, and no report shows the overlap.
PIPEDA adds a second constraint. Under the federal privacy law you may use personal information only for the purposes you collected it for, and you must destroy it once it is no longer needed.
That rules out long lookback windows built on old purchase records. A customer who bought in 2019 and never agreed to profiling should not sit inside a 2027 model.
The practical fix is to model only on customers whose consent covers marketing contact. Your sample gets smaller and your confidence interval gets wider, but the number survives scrutiny.
Win-back sequences are the sharpest case of this tension. The ecommerce email marketing breakdown covers consent, timing and send frequency.
When to stop measuring and decide
- The cohort holds at least 200 customers with a full 12 months of history.
- Returns and refunds have been subtracted from margin, not filed elsewhere.
- Tax lines are out of revenue.
- Acquisition cost covers every paid channel for the same window.
When those four hold, decide. Keep recalculating LTV only when the answer would change a spending choice. Measurement with no decision attached is just reporting.
Before you decide on LTV
- Cohort holds at least 200 customers with 12 months history
- Returns and refunds subtracted from margin
- Tax lines are out of revenue
- Acquisition cost covers every paid channel
Tying a decision rule to a budget cycle is covered in the ecommerce marketing strategy article.
