Most e-commerce brands know what customer lifetime value means. Very few measure it in a way that actually changes how they spend money. CLV gets quoted in strategy decks and dropped in board meetings, but the decisions that flow from it, who to acquire, who to retain, where to invest, are usually made on shorter, simpler signals. That gap between knowing the metric and acting on it is where most brands quietly lose ground.

Key Takeaways

  • Most brands calculate CLV on average order values that hide which customer segments are actually profitable.
  • Acquisition cost and CLV are rarely compared at the segment level, which distorts payback period assumptions.
  • Retention investment is often allocated to customers who would have stayed anyway, not those at genuine risk of leaving.
  • CLV models that ignore product margin, return rates, and support costs overstate the value of a customer by a wide margin.
  • Brands that model CLV by acquisition channel make materially better budget decisions than those that aggregate across all channels.

Why do most CLV calculations produce the wrong answer?

The most common approach is also the most misleading. A brand takes average order value, multiplies by average purchase frequency, multiplies by average customer lifespan, and calls it CLV. The problem is that average customers do not exist.

A cohort of customers acquired through paid social in Q4 looks nothing like a cohort acquired through organic search in Q2. Their repeat purchase rates differ. Their return rates differ. Their average order values differ. Collapsing them into one number produces a CLV figure that accurately describes no one.

Shopify data from 2023 showed that for most direct-to-consumer brands, the top 20 percent of customers by lifetime value generate between 60 and 80 percent of total revenue. If you are averaging those customers in with the 80 percent who buy once, your CLV number is structurally wrong before you have even started using it.

What costs do brands routinely leave out of the calculation?

Gross revenue per customer is not CLV. The figure that matters is what a customer contributes after the costs that customer generates have been subtracted.

Three cost categories tend to disappear from most calculations:

  • Product returns. In apparel e-commerce, return rates between 25 and 40 percent are common. A customer who buys four times but returns half of every order has a much lower real value than the order count suggests.
  • Customer support cost. High-volume or high-complexity customers generate disproportionate support tickets. For brands running lean support teams, this cost is invisible in the CLV model but very visible in operating margins.
  • Fulfilment variance. Customers who order single items cost more per order to fulfil than customers who bundle. If your fulfilment economics favour larger carts, a high-frequency single-item buyer can look valuable on paper but be marginal in practice.

Accounting firm KPMG noted in a 2023 retail report that brands operating with CLV models that exclude returns and support costs tend to overspend on retention programs by 15 to 30 percent relative to the actual margin contribution of the customers being retained.

Are brands retaining the right customers?

This is the question most retention programs never ask.

The default approach is to send win-back campaigns to customers who have gone quiet, offer loyalty discounts to repeat buyers, and build email flows triggered by days-since-last-purchase. The logic seems sound. The execution is often backwards.

Most at-risk customers who receive a win-back discount fall into two categories. The first group was going to churn regardless. A 15 percent discount accelerates one more purchase, extends the perceived relationship by 30 days, and then they leave anyway. The second group was not actually at risk. They buy seasonally, or they had a life event, or they were simply slow. The discount cannibilises revenue you would have received at full price.

The customers who generate the most retention ROI are those in a specific middle band: genuinely at risk, responsive to intervention, and high enough in predicted lifetime value to justify the margin cost of the incentive. Identifying that band requires a predictive model, not a days-since-last-purchase trigger.

Brands using RFM segmentation (Recency, Frequency, Monetary value) get closer to this, but RFM is a descriptive model, not a predictive one. It tells you what a customer has done. It does not tell you what they are likely to do next.

How does acquisition channel distort CLV assumptions?

One of the clearest CLV mistakes is treating all acquisition sources as equivalent once a customer is in the database.

Customers acquired through paid search tend to have higher first-order values and lower repeat rates. Customers acquired through word of mouth or referral tend to have lower first-order values and significantly higher lifetime values. Customers acquired through influencer campaigns often show strong short-term engagement followed by sharp drop-off.

If you are averaging these cohorts together when making acquisition budget decisions, you are almost certainly overspending on channels that look efficient on CAC but underperform on CLV, and underspending on channels that look expensive per acquisition but generate your most loyal customers.

A mid-sized Australian homewares brand that Lenka Studio worked with had this exact problem. Their paid social CAC looked competitive in isolation. When we helped them segment CLV by acquisition channel, they found that customers from their email referral program had a 12-month CLV almost 2.4 times higher than their paid social customers. They had been starving the referral program of budget because the CAC looked higher. The payback period calculation had been using blended CLV, not channel-specific CLV.

What does a more honest CLV model actually include?

A useful CLV model does not need to be complicated. It needs to be honest about what it includes and what it excludes.

The inputs that make the biggest difference to accuracy:

  • Gross margin per order, not revenue. If your margin varies by product category, use the category mix per customer cohort.
  • Net return rate by cohort. Some customer segments return at twice the rate of others. This should be reflected in the model.
  • Predicted purchase frequency, not historical average. Historical averages include customers who have already churned. A survival curve model gives you a more accurate forward-looking figure.
  • Acquisition channel as a segmentation variable. Run the model separately for each major acquisition source before blending.
  • A discount rate. Money received in year three is worth less than money received today. Most SMB CLV models skip this. For high-consideration purchases with long repurchase cycles, it matters.

Why does CLV get ignored when it matters most?

The honest reason is that CLV is a long-horizon metric in a short-horizon business environment. Most e-commerce brands are reporting weekly on channel performance, monthly on revenue targets, and quarterly to investors or boards. CLV plays out over 12 to 36 months. That makes it structurally uncomfortable to act on when short-term numbers are under pressure.

There is also a measurement delay problem. If you change your retention strategy today, you will not see the CLV impact clearly for 6 to 12 months. That makes attribution difficult and makes the strategy easy to abandon before the evidence arrives.

Brands that overcome this tend to have one of two things: either leadership that genuinely understands cohort economics and protects the longer-horizon view, or a data infrastructure that surfaces leading indicators of CLV early enough to be actionable. Metrics like 90-day repeat purchase rate, second-order margin, and referral rate within the first 60 days all correlate strongly with eventual CLV and arrive fast enough to influence quarterly decisions.

What does this mean for how you allocate marketing spend?

If your CLV model is wrong, your CAC targets are wrong. If your CAC targets are wrong, your channel mix decisions are wrong. The error compounds quickly.

A brand targeting a 3:1 LTV:CAC ratio with an overstated LTV figure might actually be operating at 1.8:1 when costs are properly attributed. That is not a margin optimisation problem. That is a structural problem that looks like it is working until it suddenly does not.

The practical implication is to audit your CLV model before you optimise your media spend. Most brands do it the other way around. They optimise channels first and wonder why the profitability does not improve.

If your brand is at the stage where this kind of audit would be useful, it is also worth assessing how CLV trends connect to broader brand health. A declining CLV often shows up as a brand perception problem before it shows up in the financials. The free brand health score assessment from Lenka Studio can help you surface those early signals before they become revenue problems.

When is CLV the wrong thing to optimise for?

Not every business stage makes CLV the primary metric.

In the early stage, when you are acquiring your first 500 customers, CLV data is thin and cohort sizes are too small to be statistically meaningful. Optimising heavily for CLV at this stage can cause you to abandon acquisition channels before you have enough data to judge them fairly.

In categories with very long repurchase cycles, like furniture or high-consideration electronics, CLV models based on repeat purchase frequency are less useful. A different framing, such as net promoter behaviour or referral rate, often provides more actionable signal.

The goal is not to make CLV the centre of every decision. The goal is to make sure your CLV assumptions are accurate enough that the decisions you are already making are based on real numbers.

Frequently Asked Questions

What is a good customer lifetime value for an e-commerce brand?

There is no universal benchmark because CLV depends heavily on product category, margin, and price point. A more useful ratio is LTV:CAC. Most healthy e-commerce businesses target a ratio of 3:1 or higher, meaning the lifetime value of a customer is at least three times what it cost to acquire them.

How often should e-commerce brands recalculate CLV?

CLV models should be reviewed at least quarterly, and rebuilt from scratch annually. Customer behaviour shifts, channel mix changes, and product pricing evolves. A CLV model that was accurate 18 months ago may be significantly off today.

Is CLV useful for small e-commerce businesses with limited data?

Yes, but with appropriate simplification. Even a basic CLV model that distinguishes one-time buyers from repeat buyers gives you more useful signal than a blended average. As your customer base grows past a few hundred cohorted customers, you can add more segmentation variables.

How does return rate affect customer lifetime value?

Return rate reduces both the revenue and the margin contribution of a customer. A customer with a 40 percent return rate who appears to generate $800 in annual revenue may be contributing only $280 in net revenue after returns and associated fulfilment costs are removed. This is one of the most commonly omitted variables in SMB CLV models.

Can CLV improve through product strategy, not just marketing?

Yes, and this is underused. Product bundling, subscription structures, and category expansion all increase average order value and repurchase frequency. Brands that improve CLV through product and category strategy often find it more durable than improvements achieved through loyalty discounts alone.

If your e-commerce brand is working through any of these questions and you want a team that has seen how CLV models connect to real acquisition and retention decisions, the team at Lenka Studio is happy to talk through what that looks like for your specific business. Reach out and we can take it from there.