A customer who enters at $80 and one who enters at $180 are not the same business

We've got an eCommerce client who sells one single product. And yet, some people come in at $80 and others at $180. Same product, but two buyers who have nothing in common.

For a long time, we just looked at CPA. How much it costs to acquire a buyer. Full stop. Except two buyers with the same CPA can be worth ten times less than each other over twelve months. And CPA will never tell you that.

That's where customer segmentation comes in. Forget the personas from marketing textbooks, they're useless. Here we start from something concrete, the price at which the customer walks in the door.

The entry price tells you more than the order amount

When a customer discovers a brand, they land on a page, they see a price, they buy (or they don't). That first price already tells you a lot about the person on the other side, way more than the day's revenue.

In the case we have in mind, a wellness product sold in several formats, there are three doors in.

First door, the trial size at $30-40. It's the easiest entry, the one that converts best. Makes sense, the price is low and so is the perceived risk. But this customer is testing. They don't want to commit. They often stay a one-and-done.

Second door, the pack of 8. That one's interesting because it moves people faster towards the pack of 12, and therefore towards repeat purchase. The customer has already crossed the "I'll take more than one unit" threshold.

Third door, the large format at $160-200. That one converts poorly. Normal, nobody drops $180 on a brand they don't know with a snap of the fingers. But when someone does, they stick. For a long time. And in the end, that customer is well worth it even if the page's conversion rate stings.

So you've got the same product, three completely different customers. And it's the entry price that tells you this before you even have a single data point on retention.

Here's an example of a customer segmentation strategy that fits in one line and actually serves you, because it changes what you do the next morning.

Why CPA alone lies to you

I saw it again last week on an account. Meta brings in conversions, CPA looks clean, everyone's happy. And when you look at the real value of those customers three months later, it doesn't hold up.

Meta, when you ask it to "maximise the number of purchases," will go find the easiest path. And the easiest path is almost always the $30-40 trial size. So it fills your account with customers who test and leave.

Your dashboard shows a rising volume of buyers. Your retention, meanwhile, flatlines or drops. Because you bought low-intent volume without realising it.

This is exactly the kind of problem we dig up when an eCommerce audit starts with your P&L and not your campaigns. You look at contribution margin by entry cohort, and that's when you clearly see that the $180 customer, even with a higher CAC, is more profitable over time than three $35 customers who vanish.

Measure value by entry door, not on average

In eCommerce, thinking in averages burns you. If you look at your average LTV, you're mixing the customer who tests and leaves with the one who comes in at $180 and reorders three times. The result is an average that tells you nothing.

This is where most customer segmentation methods fall flat. They hand you big buckets like "the deal hunter" or "the loyalist." Looks great in a deck and doesn't help you decide anything. What we do is cut on a criterion we actually have in the data, the entry door, and we look at each group separately.

For the $30-40 trial size, you look at what percentage makes a second purchase and after how many days. For the mid pack, you check whether it really leads to the larger format and in what timeframe. For the large format, you measure its value at 6 and 12 months.

Once you've got that, you're no longer handling one global LTV but three distinct LTVs, and you can decide for real. Like agreeing to pay more to acquire a customer who comes in at the top, because you know they're worth it.

All of this works on one condition. Your tracking has to hold up and you need to capture the email right at entry, so you can follow the cohort over time. This is where sign-up forms that actually grow your base change the game, because without email you have zero visibility on who stays and who leaves.

What you ask of Meta changes everything

Once you know the $180 customer is worth ten times the $35 customer, the question becomes simple. How do I tell Meta "go find me more of these people."

There are a few ways.

The most direct, you optimise for conversion value rather than count. You stop saying "bring me as many purchases as possible," you say "bring me as much revenue as possible." That alone pushes the algo to chase bigger carts instead of small ones.

You can also send custom value signals. Instead of passing the order value, you pass the customer's predicted lifetime value. A customer coming in at $180, you assign a much higher value in the events you send back to Meta, because you know they'll reorder. The algo learns to hunt that profile.

One thing that helps a lot here is to stop trying to micro-control everything on the targeting side. On Meta, dropping age, gender and manual targeting and letting the algo work on your good value signals almost always beats forcing audiences by hand.

Last point, landing pages. If you want more customers coming in through the large format, you need landing pages and ads that lead straight to that format, not to the trial pack. Too many brands pour all their energy into pushing the cheapest entry because it converts best, then wonder why their retention is garbage.

And the offer you push to each segment

Once you've got your three doors clearly identified, you can stop showing the same thing to everyone. The one who took a trial size, you offer them the pack of 8 when they're running low on their format. Following up three days after their order is too soon. And the one who's already on the pack of 8, you nudge them gently towards the 12. That's where product segmentation and customer segmentation meet, you match the format you push to the customer in front of you.

And the goal is never to force the sale. If you offer the right format at the right moment, the customer just feels like you're saving them time. It's when you throw a big cart at someone who just tested that it reeks of pressure.

How we apply this in practice

If you've got a product with multiple formats and multiple entry prices, here's where to start.

First, pull your cohorts by entry door over 12 months and look at the repeat rate for each one, you'll fall off your chair. After that you calculate contribution margin by cohort, not just revenue, because a large format often has a better unit margin on top of better retention. Then you adjust what you send to Meta, value optimisation, and if you can, values that reflect predicted LTV. And finally, you align your ads and your pages with the customer you actually want. If it's the one who comes in at the top, stop featuring the trial size everywhere.

Aim for the share of good customers, not raw volume. And the price at which they come in tells you, from the very first purchase, which way they'll lean. On one of our fashion accounts, we raised the acceptable CAC by 30% on a specific segment just because we proved those customers were worth it. Volume dropped but profitability climbed sharply on that segment.