How customer lifetime value is worked out
Start with what a typical customer spends in one order, multiply that by how often they buy in a year, and extend it over the number of years the relationship realistically lasts. That gives lifetime revenue. Subtract the cost of the goods and of serving them, and you have lifetime value in the form that is actually useful — profit, not turnover.
Two refinements matter. Money arriving years from now is worth less than money today, so long horizons should be discounted rather than added up at face value. And refunds and cancellations have to come out, or the figure counts revenue the business gave back.
There is a simpler version worth doing first. Look at a group of customers who bought in the same month, and track what that same group has spent since. That is a cohort, and it produces an observed figure rather than a modelled one. It is slower, and far harder to argue with.
Why customer lifetime value matters
It decides how much you can afford to pay for a customer. A business judging its advertising on the first order alone will pause campaigns that are profitable over the relationship and keep ones that are not, because the first order is the least representative purchase a customer ever makes. Read next to customer acquisition cost, lifetime value is what tells you whether the whole model works.
It also changes where money goes. If repeat purchase is where the value sits, then email, service quality and retention are acquisition investments, not afterthoughts — they are what makes the first sale affordable. And it changes who you target: the audiences that produce the highest lifetime value are rarely the ones that produce the cheapest first order.
For a business in Nepal serving both local and international customers, the split is often stark. The two groups can differ enormously in repeat rate and order size, and a single blended figure hides which side of the business is actually funding the other.
Where lifetime value goes wrong
Optimism is built into the arithmetic. Assume a longer relationship or a higher repeat rate than the data supports and the figure inflates quickly, which then justifies bidding the business cannot sustain. The forecast is usually wrong in the direction its author wanted.
Using revenue instead of margin is the second error, and it is common because revenue is easier to find. It makes low-margin products look like good customer sources.
The third is averaging everything together. Most shops have a small group of customers worth many times the typical one, and a long tail worth very little. One blended number describes almost nobody and hides both. Segment by product, channel and first purchase before drawing conclusions.
What to do about it
Calculate it on margin, from real cohorts, for a period you can actually observe rather than a lifetime you have imagined. Then use it to set an acquisition target you are willing to defend, remembering that paying more today for a customer who pays back later requires cash you have to have.
Feed the segments back into your advertising: build audiences from your highest-value customers, exclude the segments that never repeat, and let value rather than volume shape the budget. That decision belongs in a marketing strategy discussion rather than inside a single campaign, because it usually changes what the campaigns are for.
