BFCM Refund Lag: Why January Refund Waves Wreck Q4 ROAS Reporting
Q4 ROAS looks great in December — then January refunds land and the number collapses. Here's how to lag-adjust your break-even floor before you scale spend.
Quick answer
BFCM-cohort orders refund at 1.5-2.5× your baseline rate, but the returns land 4-8 weeks later — in January. The Q4 ROAS you report on December 31 is structurally inflated by roughly 8-15%. Lag-adjust break-even ROAS using a trailing 60-day refund rate, and hold a refund provision before scaling December spend.
BFCM refund lag
The 4-8 week gap between BFCM revenue recognition and the January refund wave that retroactively deflates Q4 ROAS.
BFCM refund lag is the structural delay between when a Black Friday / Cyber Monday order is booked as revenue (and credited to that day's ROAS) and when the refund actually clears (typically late December through mid-February). Because returns policies on gift purchases often run 60-90 days, and customers wait until January to initiate, the Q4 ROAS number a Performance Manager reports in early December reflects gross revenue against full ad spend — not the net contribution that survives the return wave. The reported ROAS is correct on the date it's pulled; it just isn't the number you should be scaling spend against.
Most ROAS dashboards show order-date revenue against order-date spend. That's clean accounting for a steady-state month. For BFCM it's a trap: 25-40% of November 24-30 orders ship as gifts, and gift recipients return on their own schedule — not the buyer's.
If you scale December paid spend off an uncorrected November ROAS, you're committing budget against a number that will retroactively drop 10-15% by mid-January. The lag, not the refund rate itself, is what wrecks the decision.
Why the lag exists in the first place
Three mechanics stack on BFCM cohorts. First, extended holiday return windows: most apparel and beauty brands extend their standard 30-day window to 60-90 days for orders placed in late November, pushing the refund peak into mid-January.
Second, gifting behaviour. A scarf bought on Cyber Monday and gifted on December 25 can't be returned until after the holiday — and the recipient has no urgency. Third, fit-driven categories (apparel, footwear) generate size-exchange refunds that always cluster 7-14 days after delivery, and BFCM delivery itself is delayed by carrier congestion.
The reporting illusion
On December 31 your dashboard shows November ROAS of 4.2. By February 15, after refunds settle, the same cohort's true ROAS is 3.6. If your break-even is 3.8, you scaled into a losing month while believing you were 10% above floor.
How to detect it in your own data
Pull refund-date revenue (not order-date) against the original order's acquisition cohort. If your platform doesn't expose this natively, join Shopify's refund export to the order's UTM source by order_id. The shape you're looking for is a refund volume curve that peaks 35-50 days after the order date for BFCM cohorts, versus 10-20 days for a typical October cohort.
A useful single-number signal: the 60-day refund rate on November-acquired orders, measured at T+90. Compare it to your trailing baseline. If November sits 60%+ above your October baseline, your reported Q4 ROAS is materially inflated and the refund lag is structural, not noise.
This is the same mechanic explored in refund rate impact on break-even ROAS — but the BFCM version is uniquely dangerous because the lag hides the deterioration during the exact window you're making scaling decisions.
Benchmarks: refund lag by category
BFCM cohort refund behaviour vs baseline, by vertical
| Vertical | Baseline refund rate | BFCM cohort refund rate | Median lag (days) | Q4 ROAS overstatement |
|---|---|---|---|---|
| Apparel & footwear | 12-18% | 22-32% | 42 | 14-18% |
| Beauty & skincare | 4-7% | 8-12% | 28 | 6-9% |
| Consumer electronics | 6-10% | 11-16% | 35 | 8-12% |
| Home & decor | 8-12% | 14-20% | 38 | 9-13% |
| Jewellery & accessories | 10-15% | 18-25% | 45 | 11-15% |
| Health & supplements | 3-5% | 5-8% | 21 | 4-6% |
Apparel and jewellery sit at the top because both gifting share and fit-driven returns compound. Supplements barely move — low gifting share, no fit issue. If you're in the top three rows, your December spend decision is the one most likely to be wrong.
How to fix the reporting before you scale spend
Build a lag-adjusted ROAS view. Take reported ROAS and multiply by (1 − expected_refund_rate_BFCM). For an apparel store reporting 4.2 ROAS with an expected 27% BFCM refund rate, the planning number is 4.2 × 0.73 = 3.07. That's the floor you compare against break-even, not the raw 4.2.
Then reserve a refund provision in finance: hold back 10-15% of November-December gross revenue as restricted cash through February 15. This prevents the second mistake — treating peak-week revenue as available working capital and over-committing January inventory POs against money that's about to walk out the door as refunds.
Experiments worth running before next BFCM
Test a size-confidence module on PDPs for the apparel cohort (model height/size, fit reviews surfaced by body type). The hypothesis isn't conversion lift — it's refund rate reduction on BFCM-acquired orders, measured at T+60. A 3-point refund rate drop on the BFCM cohort is worth more than a 0.2pp conversion lift.
Second, test return-window framing in checkout: "Free returns through January 31" versus "Extended holiday returns through February 28." The shorter window pulls refund decisions forward, compressing the lag and giving you a cleaner read on true Q4 ROAS by December 31.
Frequently asked questions
For most apparel and jewellery brands, 11-18%. Beauty and supplements see 4-9%. The size of the gap is driven less by the refund rate itself and more by gifting share — categories with high gift purchase rates skew worse because returns are pushed past December 31.
No — finance and the board need a December number. The fix is to report two numbers: gross Q4 ROAS (the dashboard pull) and lag-adjusted Q4 ROAS (gross × (1 − expected BFCM refund rate)). Use the second one for any spend-scaling decision in December and January.
Use 1.7× your trailing 90-day baseline as a starting estimate, then refine after your first measured BFCM. For apparel that lands around 24-28%; for beauty 8-11%. Update the multiplier each year from your own refund-date cohort data.
Yes — both platforms report order-date revenue against ad spend and have no native concept of refund lag. The platform ROAS is always the inflated number. You need to adjust downstream in your own warehouse view; don't expect Ads Manager to do it for you.
The standard version assumes refunds and orders are roughly contemporaneous, so a steady-state refund rate adjustment works. BFCM refund lag is specifically the timing mismatch: refunds from November orders don't land until January, so the December decision window sees an artificially clean number.
No — December has genuine demand. Just compare lag-adjusted ROAS against break-even, not gross ROAS. If lag-adjusted still clears break-even by a margin, scale. If it's within 10% of break-even, hold flat. If it's underwater, cut.
Gift cards are worse — they book as revenue when sold (BFCM) but the redemption (and any subsequent refund) can happen months later. Most teams exclude gift card revenue from the BFCM ROAS cohort entirely and report it separately when redeemed.
Exchanges that swap for equal value don't hit ROAS, but exchanges for lower-priced items partially refund. Track exchange-driven partial refunds in the same cohort — they're typically 20-30% of total BFCM refund dollars for apparel.
Work with finance to flag 12-15% of November-December gross revenue as a contingent liability in the close. Operationally, that money stays in the bank account but is off-limits for January inventory POs or media commitments until February 15.
Around February 20-28 for most categories. By then 90%+ of the BFCM refund wave has cleared and the cohort's true ROAS is stable. From that point you can report a single number for the cohort with confidence.
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.