Customer Cohort Analysis: The Exact Playbook We Use to Find Where Scaling Brands Leak Money

Published: 3 May 2026

I had a call last month with a DTC founder doing £3.2 million a year. Revenue up 40% year on year. Meta ROAS sitting at a comfortable 4.2x. Google pulling its weight. Email flows humming along.

On paper? Everything looked brilliant.

Then we pulled the cohort data.

Turns out, customers acquired in Q4 2025 had a 12-month repeat purchase rate of just 9%. The Q1 2025 cohort? 31%. That's not a dip. That's a cliff. And it meant the "growth" they were celebrating was almost entirely propped up by one-time buyers who would never come back.

The business wasn't scaling. It was inflating.

This is what happens when you run your brand on averages. And if you're a 7 or 8-figure brand pushing hard to scale... there's a decent chance the same thing is happening to you right now.

Jump Ahead

• Why Your Dashboard Averages Are Lying to You
• What Customer Cohort Analysis Actually Is
• The Exact 4-Step Framework We Use
• What the Data Actually Reveals (Real Examples)
• The Three Leaks That Kill Scaling Brands
• How to Act on This Data Today

Why Your Dashboard Averages Are Lying to You

Here's the problem with aggregate metrics. They smooth everything out. Your blended CPA looks fine. Your overall LTV looks healthy. Your ROAS is above target.

But underneath?

One channel is bringing in high-quality buyers who repurchase three times in 12 months. Another channel is bringing in bargain hunters who bought once because of a 30% off ad and will never return. When you average those two together, both look "fine." Neither looks alarming. And that's exactly why you miss it.

Let's do the maths.

Say you acquired 5,000 customers last quarter. Your blended CPA is £28 and your average LTV is £112. That's a 4:1 LTV:CAC ratio. Textbook healthy.

But break it down by cohort:

• Meta Prospecting cohort (2,800 customers): CPA £22, but 12-month LTV of just £64. That's a 2.9:1 ratio.
• Google Brand Search cohort (1,200 customers): CPA £18, LTV of £189. That's 10.5:1.
• Meta Retargeting cohort (1,000 customers): CPA £44, LTV of £143. That's 3.25:1.

Same quarter. Same "healthy" blended numbers. But the channel you're pouring the most budget into (Meta Prospecting) is actually your worst performer on a unit economics basis. And the one that looks "expensive" (retargeting at £44 CPA) is delivering far better customers.

Averages hide this completely.

What Customer Cohort Analysis Actually Is (And Why Most Brands Skip It)

Cohort analysis is dead simple in concept. You group customers by when they were acquired (usually by month or quarter), then track how each group behaves over time.

That's it.

You're not looking at "all customers" as one blob. You're asking: "Of the 2,000 people we acquired in January 2026, how many purchased again by March? By June? By December? And how much did they spend?"

Then you compare January's cohort to February's. To March's. To the same month last year.

According to Eightx's 2026 ecommerce benchmarks, average customer acquisition costs have increased 18-24% across most verticals in the past two years. When acquisition costs are climbing that fast, understanding which customers are actually worth acquiring becomes the difference between profitable scaling and expensive vanity metrics.

So why do most brands skip it?

Because it's not on the default Shopify dashboard. It's not in your Meta Ads Manager. It requires pulling data from multiple sources, stitching it together, and actually sitting with uncomfortable truths about which channels, campaigns, and offers are bringing in customers that stick... and which ones aren't.

Most brands would rather not know.

Most brands looking at their blended metrics while their cohort data tells a completely different story.

The Exact 4-Step Framework We Use With Clients

This is the framework we run with every brand we work with. It takes about 2-3 hours to set up properly the first time, and then maybe 30 minutes a month to maintain. For brands doing seven figures plus, this is non-negotiable.

Step 1: Pull Your Cohorts by Acquisition Month

Group every customer by the month they made their first purchase. Not their most recent purchase. Their first. This is your acquisition cohort.

You can pull this from Shopify (Analytics > Reports > Customers over time), from your CDP, or from a simple SQL query if you've got a data warehouse set up. If you're using tools like Saras Analytics or Triple Whale, most of this is pre-built.

You want at least 12 months of cohorts. 24 is better.

Step 2: Track Repeat Purchase Rate at 30, 60, 90, 180, and 365 Days

For each monthly cohort, calculate what percentage made a second purchase within each time window. This is your cohort retention curve.

Here's what you're looking for:

• 30-day repeat rate: Are people coming back quickly? This often signals product-market fit and strong post-purchase flows.
• 90-day repeat rate: This is your bread and butter metric. For most DTC brands, if someone hasn't purchased again within 90 days, the probability drops dramatically.
• 365-day repeat rate: Your true retention number. This is what determines whether your LTV projections are fantasy or reality.

Plot these on a simple spreadsheet with months down the left side and time windows across the top. You'll see patterns immediately.

Step 3: Layer in Acquisition Channel

This is where it gets genuinely interesting. Take those same cohorts and split them by the channel that acquired them. Meta Prospecting. Google Brand. Google Non-Brand. TikTok. Email capture. Organic.

Now you're not just asking "are our January customers retaining?" You're asking "are our January Meta Prospecting customers retaining at the same rate as our January Google Non-Brand customers?"

I can tell you right now... they almost certainly aren't.

We consistently see 2-4x differences in retention rates between channels within the same cohort. That has massive implications for how you allocate budget. If you're interested in how different DTC marketing strategies impact customer quality, this is exactly the lens you need.

Step 4: Calculate True LTV:CAC by Cohort and Channel

Now pull it all together. For each cohort, by channel, calculate:

• True CAC (include agency fees, creative costs, tools. Not just ad spend)
• Observed LTV at 90, 180, and 365 days (actual revenue, not projected)
• Gross margin adjusted LTV (because revenue means nothing if your margins are 20%)

According to MHI Growth Engine's LTV:CAC analysis, a healthy ratio sits at 3:1 or above. Below 1:1 and you're literally paying people to buy from you. Above 5:1 and you're probably under-investing in growth.

But here's the key... you need this ratio at the cohort level, not the blended level. A blended 3.5:1 can hide a channel running at 1.2:1 that's eating your margin.

What the Data Actually Reveals (With Real Numbers)

I want to show you what this looks like in practice because the patterns are remarkably consistent across brands we work with.

The "Cheap Customer" Trap

We worked with a health supplements brand doing about £5 million annually. They'd been aggressively scaling Meta spend because the CPA was low. £16 per customer. Incredible, right?

When we pulled cohort data by channel:

• Meta Broad Prospecting: £16 CPA, 90-day repeat rate of 11%, 12-month LTV of £38
• Google Non-Brand Search: £34 CPA, 90-day repeat rate of 28%, 12-month LTV of £127
• Organic/Direct: £0 CPA (effectively), 90-day repeat rate of 41%, 12-month LTV of £203

The "expensive" Google customers were worth 3.3x more than the "cheap" Meta customers. The brand had been optimising for the wrong metric for over a year. They'd been celebrating low CPA while slowly destroying their unit economics.

We shifted 30% of Meta Prospecting budget to Google Non-Brand and invested in better organic content. Within two quarters, blended LTV increased 22% while total customer acquisition only dropped 8%.

Net result: significantly more profit from slightly fewer customers.

The Seasonal Cohort Problem

Another pattern we see constantly. Black Friday and Q4 cohorts almost always have worse retention than Q1 or Q2 cohorts. Makes sense when you think about it. You're acquiring people with heavy discounts during peak shopping season. Many of them are one-time gift buyers or deal-hunters.

One fashion brand we analysed had a Q4 2025 cohort with a 7% 90-day repeat rate versus a Q2 2025 cohort at 24%. Yet they kept using blended annual numbers to project growth.

Their "record Q4" was actually creating a retention hole that took six months to show up in the P&L.

The Three Leaks That Kill Scaling Brands

After running this analysis across dozens of brands, the same three leaks show up again and again.

Leak 1: Channel Quality Decay

As you scale spend on any channel, customer quality tends to decline. Your first £10k/month on Meta hits your warmest lookalikes. Your next £50k hits colder audiences. By the time you're spending £150k/month, you're reaching people who barely know your category exists.

Cohort analysis makes this visible. If your January Meta cohort retained at 25% but your June Meta cohort (after you tripled spend) retains at 12%... that's channel quality decay in action.

The fix: Don't just track spend and CPA by channel. Track cohort quality by spend level. There's almost always a spend threshold where quality falls off a cliff. Find it. That's your efficient frontier.

Leak 2: Over-Discounting on Acquisition

This one is brutal. Brands offer 20-30% off to acquire first-time customers, then wonder why those customers never buy at full price. You've anchored them to a discount. You've trained them to wait for the next sale.

Run your cohorts split by acquisition offer. Compare "20% off first order" customers to "free shipping" customers to "no offer" customers. I've seen the retention gap be as wide as 15-20 percentage points.

The fix: Test value-add offers instead of percentage discounts. Free gift with purchase. Exclusive access. A bundle deal that maintains margin. The goal is to remove friction from the first purchase without destroying the price anchor.

Leak 3: Ignoring the Post-Purchase Experience

Most brands spend 95% of their energy getting the customer. Then... silence. Maybe a generic order confirmation. Maybe a "rate your purchase" email two weeks later. That's it.

Meanwhile, that customer is going to Reddit to see what other people think. They're checking if they overpaid. They're looking at competitors.

This is where the full customer journey matters. What happens in the 7-14 days after first purchase is arguably more important than the ad that got them there. A strong post-purchase email flow, an unboxing experience that surprises them, proactive support... these things move 90-day retention rates by 5-10 percentage points.

That might not sound like much. But on a base of 10,000 customers per quarter, a 5 percentage point retention improvement at an average order value of £65 is an extra £32,500 per quarter. From customers you've already paid to acquire.

How to Act on This Data Today

If you've made it this far and you're thinking "right, we need to do this"... here's exactly where to start. Today. Not next quarter.

1. Export your customer data by first purchase date for the last 12 months. Shopify, your CDP, whatever you use. Get first purchase date, total spend to date, number of orders, and acquisition source if you have it.
2. Build a simple cohort retention table. Months down the left. Percentage who made a 2nd purchase within 30, 60, 90, 180, 365 days across the top. Google Sheets is fine for this.
3. Look for the cliff. Where does retention drop off dramatically? Is it consistent across cohorts or did something change? If retention dropped after a specific month, go back and look at what changed. New channel? New creative? Bigger discounts?
4. Split by channel. Even a rough split (paid vs organic vs email) will reveal massive differences in customer quality.
5. Recalculate your LTV:CAC at the cohort level. If any channel is below 2:1 on actual (not projected) 12-month LTV, that's your leak.

This isn't theoretical. This is literally a spreadsheet exercise you can do this afternoon.

The Bigger Picture

Here's what I see over and over with brands doing £2-10 million. They've got someone managing paid ads. Someone handling email. Maybe an SEO agency. And none of these people are talking to each other about customer quality.

The paid team is optimising for CPA. The email team is optimising for open rates. The SEO team is optimising for traffic. Everyone's hitting their individual KPIs. And the business is still struggling to scale profitably.

Because nobody is asking the cohort question. Nobody is connecting the dots between acquisition source, customer quality, retention, and true lifetime value.

That's the work that actually matters at this stage. Not another 5% improvement in ad ROAS. Not another A/B test on a subject line. The real leverage is in understanding which customers are worth acquiring, where they come from, and what keeps them coming back.

Everything else is optimising the wrong thing.