The leaky bucket is the gap between customer acquisition and customer retention. New customers come in through paid media. Lapsed customers leak out through poor retention. For most DTC brands at $5M to $30M, the leakage is large enough that 60 to 110 percent of paid media spend is going to replace customers already acquired. Here is the five-number diagnostic to calculate yours.

Most DTC founders we audit can tell us within a week if their CAC has moved. They cannot tell us, with any precision, what their lapsed customers have cost them in the same period. One of those numbers gets a slack channel. The other gets a shrug.

That gap is the leaky bucket. It is the single largest hidden cost in DTC at $5M to $30M, and it is the reason so many brands at this stage feel like they are scaling activity but not profit. Acquisition is filling the bucket faster than ever. Retention is leaking it out the bottom faster than ever too. The bucket level looks stable. The cost of refilling it keeps climbing.

This is the diagnostic we run with founders when we sit down to open the books. Five numbers. About 45 minutes if you have the data clean, longer if you do not. The output is a single figure: what your retention leakage costs you, every month, in CAC you have to keep spending to stand still.

Why this is a 2026 problem, not a 2022 problem

The leaky bucket has always existed. What has changed in the last 24 months is the cost of each litre of water leaving it.

CAC has continued to climb across categories. Most brands we work with are paying 25 to 60 percent more per acquired customer than they were in 2022. The platforms are not getting cheaper. Audience targeting is consolidating into broad and Advantage+, which has flattened the audience-driven CAC lever. iOS is still suppressing signal. None of these trends are reversing.

So every lapsed customer in 2026 costs more to replace than they did in 2022. The leakage is the same. The price of standing still has gone up. Brands without a tight retention engine are paying more, every quarter, just to keep their topline flat. That is the structural problem the leaky bucket audit surfaces.

The five numbers you need

You will need these for the last 90 days, ideally pulled together in one view. If you do not have them, that is the diagnostic right there.

Number 1. Blended CAC

Total marketing spend across all channels and partners, divided by total new customers acquired in the period. Not platform CAC. Not paid-only CAC. Blended. Every dollar going out, every new customer coming in.

Number 2. Average contribution margin per first order

Average order value of a new customer minus product cost, shipping, payment fees, discount cost, and any acquisition-attributable cost (free shipping promos, welcome discounts). Most brands have AOV. Far fewer have this number, and it is the more important one.

Number 3. Repeat purchase rate at 90 days

Of the customers who made their first purchase 90 days ago, what percentage have come back to buy a second time? Pull two cohorts: customers who completed a post-purchase flow and customers who did not. The gap between those two numbers tells you what your retention engine is actually doing.

Number 4. Revenue per subscriber, trailing 90 days

Total revenue from your email and SMS programme over the last 90 days divided by your average subscriber count over the same period. This is the cleanest single read on whether your owned-channel engine is compounding or stagnating.

Number 5. Lapsed customer cohort size

Customers who bought once or twice in the period 90 to 180 days ago and have not bought since. This is the leak. The bucket level you have to refill to stand still.

The calculation

Once you have all five, the leakage cost calculation is straightforward.

Start with your lapsed customer cohort size from number 5. Multiply that by your blended CAC from number 1. That figure is what it cost you, originally, to acquire the customers who have now leaked out. It is also, roughly, what it will cost you to replace them this quarter to keep the topline flat.

Subtract the contribution margin those customers produced for you while they were active. Most brands have not modelled this cleanly, but a reasonable proxy is the number of orders they placed multiplied by your contribution margin per first order (from number 2). If they made it to a second order, you can lift that figure by 30 to 50 percent for the second-order margin uplift.

What is left is your net leakage cost. The amount of money your business has burned through acquisition spend that retention failed to convert into a profitable customer relationship.

For a brand acquiring 1,000 new customers a month at a £40 blended CAC, with a 70 percent lapse rate inside 180 days, that figure sits somewhere between £200,000 and £300,000 in net leakage per quarter. Six figures, every 90 days, hidden across two reports that do not talk to each other.

What each number tells you on its own

Even without the full calculation, each of the five numbers individually surfaces a different problem.

If your blended CAC is more than 30 percent above your contribution margin per first order, you are running a payback timeline that finance will be uncomfortable with the moment they see it. The fix is either a sharper acquisition mix or a stronger retention programme that pulls second-order timing forward.

If your contribution margin per first order is lower than you expected, your discount strategy is probably eating margin you cannot see in the topline. Most brands underestimate this number by 15 to 25 percent because welcome discounts, free shipping, and post-purchase incentives never make it into the marketing dashboard.

If your repeat purchase rate at 90 days is below 25 percent, your retention engine is either underbuilt or operating on a 2020 playbook. We wrote about the specific flows that move this number in Klaviyo CAC Payback.

If your revenue per subscriber is flat or falling quarter on quarter, your owned channel is not compounding. That is usually a segmentation problem, a flow architecture problem, or a list-quality problem. Sometimes all three.

If your lapsed customer cohort is more than 60 percent of customers acquired six months ago, you are running an acquisition-only business model dressed up as a brand. The fix is structural, not tactical.

Why most brands have never run this

Three reasons, and they reinforce each other.

The data lives in different places. Acquisition data sits in ad platforms. Retention data sits in Klaviyo and Shopify. Contribution margin lives in finance. Pulling all five numbers into one view requires either a dashboard nobody has built yet, or a 45-minute spreadsheet job nobody has prioritised.

The reporting cadence is wrong. Most brands review marketing weekly or monthly. The leaky bucket only shows up clearly on a 90 to 180 day window. By the time it shows up in the monthly P&L, three quarters of work has already happened against the wrong assumption.

The accountability is split. Acquisition agencies are accountable for new customers. Retention agencies are accountable for repeat revenue. Nobody is accountable for the leakage between them. We wrote about this structural gap in the connected agency model piece.

None of these are insurmountable. All of them are why this calculation almost never gets run, and why it surprises founders when it does.

What to action this week

Three moves, in order, if any of this resonated.

Pull the five numbers for your last 90 days. If you only manage four of the five before next Friday, that is still more than 90 percent of DTC brands at your stage have ever calculated. Number 5 (lapsed customer cohort) is the one most teams have to build from scratch. Worth the time.

Take the net leakage figure and put it next to your last quarter's paid media spend. The ratio tells you something uncomfortable. Most brands find that net leakage is 60 to 110 percent of their paid media bill. That means roughly half to all of your acquisition spend last quarter went to replacing customers you already paid to acquire. That conversation tends to change priorities quickly.

Decide what the next 90 days of marketing investment looks like with leakage as the number you are managing against. For most brands at $5M to $30M, the answer is a meaningful shift of budget from acquisition to retention work, supported by a cleaner reporting setup that connects the two.

Where to go next

Webtopia and our sister brand Oaks run acquisition and retention as one connected programme, with leakage as the headline metric. We do this because the gap between acquisition and retention is where most growth at $5M+ gets lost, and almost no agency in the space is set up to close it.

If you want a leaky bucket audit run on your own account, book a call and we will walk through your numbers with you. If you want a sharper view on the retention side of the equation, Retention Basics is the longer-form companion to this piece.

Frequently asked questions

What is the leaky bucket in DTC?

The leaky bucket is the gap between customer acquisition and customer retention. New customers come in through paid media (filling the bucket) and lapsed customers leak out through poor retention (draining the bucket). For most DTC brands at $5M to $30M, the leakage is large enough that 60 to 110 percent of paid media spend is going to replace customers already acquired.

How do I calculate retention leakage cost?

Take the number of customers who bought 90 to 180 days ago and have not bought since (the lapsed cohort). Multiply by your blended CAC to get the acquisition cost that has not yet paid back. Subtract the contribution margin those customers produced while active. The result is your net leakage cost, usually six figures per quarter for a brand acquiring 1,000+ customers a month.

What is a healthy repeat purchase rate at 90 days?

For a DTC brand at $5M to $30M with a well-built retention engine, a 90-day repeat rate above 30 percent is healthy. Below 25 percent suggests the post-purchase flow architecture is underbuilt or operating on an outdated playbook. The single biggest lever is the post-purchase email flow in the first 30 days.

Why is acquisition without retention more expensive in 2026?

CAC has climbed 25 to 60 percent across categories since 2022, while the audience and targeting levers inside ad platforms have flattened. This means every lapsed customer in 2026 costs more to replace than they did three years ago. Brands without a tight retention engine are paying more each quarter just to keep their topline flat.