Hey,

Here's a combination that should be impossible.

Every team on the account was hitting their target. Paid was buying new customers at the agreed cost. Email was sending on schedule.

Nothing broke anywhere.

And the business was still getting smaller. Each and every month.

I was on a call about this account a few weeks ago. The agency partner who runs its paid media had found it himself. That's the only reason anybody found it.

His words: "we've been getting new customers all year, but then they're not coming back."

 

What was actually happening

New customer acquisition was fine. That's what makes this one so hard to catch.

The number that moved was the second-order rate inside twelve months. It had slipped by a couple of points year over year.

A couple of points sounds like rounding. Nothing that'll have a meaningful impact.

Run it through a year of orders and it stops sounding like rounding.

Say a hundred new customers used to produce twenty second orders. Now they produce eighteen. That's two orders lost per hundred customers.

At ten thousand new customers a year, that's two hundred orders you used to get for free.

What you lost is a slice of every cohort, forever. And it compounds.

 

Why nobody caught it

This is the part worth sitting with.

Look at who was responsible for what.

The paid team's number was cost per new customer. That was on target, so paid reported a good month.

The email team's number was sends, opens and attributed revenue. Also fine.

Nobody owned the twelve-month repeat rate.

It sits between two departments. So it belongs to neither. So it moves for a year before anyone looks at it.

Every individual scoreboard was green. The business was still shrinking.

 

The decision underneath it

Here's the cause. It was a resource allocation nobody revisited.

More than 90% of their Meta spend had gone to new customers.

That was the right call when they made it. Acquisition was the real constraint back then. Then it just kept running, long after the constraint moved.

The partner described retention spend as "always starved." It existed. It ran during sales. It never got a real budget.

So the brand spent a year buying customers and almost nothing keeping them. The acquisition numbers looked great the whole time.

 

What it does to the number you're allowed to pay

Here's where this turns into a budget problem.

Your maximum cost per new customer comes out of what a new customer is worth. Most brands set that number once. Then they treat it as a fact about the business.

It's a fact about one cohort.

If your twelve-month repeat rate falls, the lifetime gross profit of a new customer falls with it. Your ceiling on acquisition cost drops at the same time.

Nobody sends you a notification.

So you carry on buying at a cost that cleared the bar eighteen months ago. The ad account still reports hitting target, because the target is the stale one.

That's how a brand runs efficient against its own benchmark and unprofitable in reality. For a year. With no single number looking wrong.

 

The check, and it takes less than an hour

Open your orders export. Build one table.

Group customers by the quarter of their first order. Four quarters is fine. Eight is better.

For each cohort, work out the share who placed a second order within 365 days. You need full twelve-month windows. So your most recent complete cohort is the one from four quarters ago.

Then read down the column.

A flat line means your retention is holding and you can keep buying.

A line sloping down means each new customer is worth less than the one before. Either your acquisition cost comes down to match, or your repeat rate comes up.

Most brands have never built this table. It takes less than an hour. Right now, before you spend heavily into Q4, it's the most useful hour available to you.

You only need the orders export. No new tooling, no tracking fix, no agency. Every brand I've asked already has the data sitting there.

 

What to do if the line is dropping

Three things, in order.

First, work out whether it's the product or the follow-up. Look at second-order rate by first product purchased. If one entry product repeats much worse, you've been acquiring customers into a dead end. Your best-performing ad might be the cause.

Second, check what share of your paid budget touches existing customers. Under 10% with a falling repeat rate means you've found a lever you aren't pulling.

Paid's main goal should always be new customer acquisition. But if a little more budget makes each customer worth more to the business, that's worth doing.

Third, look at your email program by audience. Count how many flows and campaigns are written for someone who has never bought. On most accounts it's nearly all of them.

That last one surprises people. The partner on this account said it plainly. A lot of their email was stacked toward new customers too, even though they had a huge existing customer base they could have been emailing.

So the brand had two channels, both pointed at acquisition, and a repeat rate nobody was feeding.

After that, recalculate your maximum cost per customer off the newest complete cohort. Not the number in your spreadsheet. If the two differ by more than about 10%, you've been running to the wrong target.

 

The part that matters for Q4

You're about to acquire more new customers in eight weeks than you normally do in six months.

If your repeat rate is dropping, every one of them is worth less than the equivalent customer last year. And you'll be making next year's budget decisions on their first orders.

Build the cohort table before the volume arrives. Afterwards you can't separate the Q4 cohort from the trend.

There's one more reason to do it now.

A Q4 cohort behaves differently from the rest of your year. Gift buyers and discount buyers repeat at lower rates than someone who found you in March and paid full price.

So measure your Q4 intake against your annual average and it'll look like retention collapsed in January. What actually happened is that you bought a different kind of customer for eight weeks. They need their own line on the table.

 

Say you're past $1M a year and spending $30K a month or more on ads. If you want a second pair of eyes on where your problem sits, I keep a few free 30-minute calls open each week.

Talk soon,
Andrej

PS: Reply "cohort" and I'll send you the exact table layout, including the formula for the 365-day window. It's the one I build on every audit.

PPS: We finally caught the one responsible →