Adjusting The Paid-CAC Ceiling For Refund And Return Rates
Apparel and beauty stores routinely overspend on paid acquisition because the CAC ceiling is set on gross orders, not net-of-returns contribution. Here is how to apply the refund-rate haircut so spend matches revenue that actually sticks.
Quick answer
Multiply your gross-order CAC ceiling by (1 − refund rate) to get the net ceiling you can actually afford. An apparel store with a €60 gross ceiling and a 38% return rate has a true ceiling of €37.20 — spending above that loses money on every cohort, regardless of what your ad platform reports.
Adjusting the paid-CAC ceiling for refund and return rates
Discounting your maximum allowable CAC by your refund-and-return rate so paid spend is benchmarked against net-of-returns contribution, not gross orders.
Most paid-acquisition ceilings are built on gross order value: take your contribution margin, divide by your target payback period, and that is the CAC you are willing to pay. The trouble is that an order is not the same as revenue kept. In apparel, 30-45% of units come back. In beauty, 8-12% of orders refund.
A refund-adjusted ceiling rebuilds the math on the slice of revenue that survives returns. The fix is a one-line haircut: multiply the gross ceiling by (1 − refund rate). It rarely changes the model — it almost always changes the bidding strategy.
If your finance team reports on shipped orders and your media buyer optimises to platform-reported purchases, the gap between them is your refund rate — and it is silently eating your margin.
Why the gross-order ceiling overspends
Meta, Google, and TikTok fire the purchase event at checkout completion. They do not retract it when the parcel comes back 21 days later. Your tROAS target, your bid caps, and your CAC ceiling all sit on top of that gross signal.
For an apparel brand with a 38% return rate, every €100 in reported revenue is really €62 in revenue kept. If contribution margin is 55% of revenue kept, you have €34.10 to spend on acquisition — not the €55 the gross model implies. The ceiling you are bidding to is 61% too high.
The bracket-buyer trap
Apparel return rates are not evenly distributed. The customers who order three sizes of the same dress have return rates above 60%. If your paid channel is disproportionately attracting them — fashion-forward Meta audiences are a usual suspect — your channel-level refund rate can sit 10-15 points above the store average. Adjust the ceiling per channel, not per store.
How to detect a refund-blind ceiling
Pull 90 days of paid-acquired orders and join them to the refund table by order ID. Compute refund rate by channel, by campaign, and by first-product SKU. If any of those slices sit more than 5 percentage points above the store average, your ceiling is overspending on that slice.
Two diagnostic signals confirm it. First: cohort contribution margin trends down even though blended ROAS holds steady — returns are draining the back end. Second: payback period in the finance model is 30-40% longer than the media model predicted. Both point to a gross-vs-net mismatch.
How to apply the haircut
Start with your existing gross ceiling — the one derived from your target POAS or contribution-margin payback model. Multiply by (1 − r), where r is the trailing 90-day refund rate for that channel. Re-platform the new value into Meta and Google as your tCPA, or feed it into your bidder as the new target.
Two refinements matter. Apply the haircut on revenue, not on units, because high-value items often refund at higher rates. And use channel-specific refund rates: Meta prospecting on a beauty store might run at 6% refunds while Google brand sits at 2% — a single store-wide haircut over-penalises brand and under-penalises prospecting.
Worked example: apparel store
Gross AOV €95, contribution margin 52%, target 6-month payback. Gross CAC ceiling = €95 × 0.52 = €49.40. Trailing refund rate on Meta prospecting = 41%. Adjusted ceiling = €49.40 × (1 − 0.41) = €29.15. Drop tCPA from €49 to €29 and the channel either gets profitable or gets paused. Either outcome is correct.
Experiment ideas once the ceiling is right
With the haircut in place, the highest-leverage tests shift to reducing the refund rate itself — every percentage point you cut lifts the ceiling. Try size-recommendation widgets on PDPs, video on the model in two sizes, and a returns-deposit on bracket-style orders. Each maps to a measurable refund-rate delta within 30 days.
On the demand side, test creative that under-promises fit or shade. A beauty brand running shade-matching quizzes on cold traffic typically lands at 4-6% refund vs 9-12% on un-quizzed traffic. That single intervention can raise the defensible ceiling by 5-7%, which compounds across every channel feeding the contribution margin-adjusted LTV model.
Frequently asked questions
Channel-specific, trailing 90 days. A blended store rate over-penalises efficient channels like brand search and under-penalises prospecting where bracket buyers concentrate. If you cannot split by channel, split at least by prospecting vs retargeting.
Yes, but with their own refund rates — which are usually lower. Brand search on an apparel store often refunds at 20-25% vs 40%+ on cold Meta. Applying a single store-wide rate everywhere will distort budget allocation away from the channels that already work.
They stack. Contribution margin-adjusted LTV is your true lifetime ceiling; the refund-rate haircut is the first-order adjustment on the first-purchase contribution that feeds it. If you skip the haircut, your LTV:CAC ratio looks healthier than it is.
The (1 − r) haircut handles only revenue lost to returns. Reverse-logistics costs — restocking, refurb, write-offs on opened beauty SKUs — belong in your contribution margin, not in the refund rate. Otherwise you double-count.
Target POAS sets the relationship between profit and spend at a strategic level. The refund haircut is the tactical correction that makes the POAS calculation trustworthy. Without the haircut, the POAS you bid to is denominated in revenue you do not get to keep.
Use a rolling 90-day refund rate, refreshed monthly, and review quarterly for seasonality. Q4 returns spike in January for apparel; if your ceiling runs unchanged from October to February, you will overspend the entire post-holiday window.
Apparel: 45-60 days covers ~95% of returns. Beauty: 30 days. Compute the haircut on a refund window that matches your return policy plus a one-week tail. Shorter windows under-state the rate and re-introduce the overspend you were trying to fix.
Often yes. In apparel, orders above €150 frequently refund at 50%+ because they are bracket purchases. Segment refund rates by AOV band and apply band-specific haircuts to value-based bidding tiers if your platform supports them.
Below 5%, the haircut moves the ceiling by less than your week-to-week noise. Most electronics and supplements stores can skip it. Apparel, beauty, footwear, and home-textiles stores cannot — those categories sit at 8-45% and the haircut is structurally material.
It is the last correction before the number leaves the model and enters the bidder. POAS gives you the profit target, contribution margin gives you the gross ceiling, and the refund haircut converts gross into net. Skip any of the three and the ceiling is wrong by a known amount.
Track CAC, channels, and funnel conversion in one place
Metricuno connects ad spend, funnel events, and revenue so you can see CAC by channel, cohort, and campaign — without stitching together five tools.