Holiday returns: a January validation checklist

The busiest selling weeks of the year produce the busiest returns weeks of the year, six to eight weeks later. The affiliate commission on those orders was tracked in November, sits pending over the holidays, and — unless somebody does something about it — is approved and invoiced while the parcels are still coming back.

This is the checklist for that fortnight. It assumes you run one or more affiliate networks as an advertiser, that you refund in Shopify, and that nobody on your team wants to spend the first week of January reading a CSV. Every date below is worked out from the network's own published rule; the sources are at the end.

Why January is the expensive month

Three things line up badly.

The volume is concentrated. A month's worth of commission can be tracked in four days. Whatever your usual pending list looks like, the one you are reviewing in January is several times larger, and the proportion of it that will be returned is higher than your annual average — NRF and Happy Returns put the online return rate at about 19.3% of sales in 2025, before any holiday effect.⁸ We have no sourced holiday-specific figure to give you, so take your own November and December numbers from last year rather than a benchmark.

The refund lands after the decision. Gift returns are the slowest kind: bought on 27 November, opened on 25 December, posted back in the first week of January, refunded when your warehouse catches up. On a 30-day validation period, the commission on that order was approved before the customer even opened the box.

Nobody is at their desk. The deadline that matters most falls between Christmas and mid-January, which is exactly when the affiliate manager is on leave and the finance close is late. Networks do not wait for you to come back: "If a commission isn't reviewed before the period ends, it will be automatically approved in line with your program's auto-validation settings."²

When a Black Friday sale auto-approves

Black Friday 2026 is 27 November, Cyber Monday is 30 November. Here is when the commission on a sale tracked on Black Friday stops being yours to change, by network rule:

Read the first row again. On CJ's standard cycle, Black Friday and Cyber Monday transactions lock on 10 December — corrections for the previous calendar month go in before the 10th of the current one.³ If your January validation is the first time you look at the season, CJ is already closed. Sales from 1 December onwards get until 10 January.

Your own dates will differ: the periods in the table are each network's published default or a worked example, not your contract. Validation periods by network has the defaults, the maximums and where each number is set.

The eight-step routine

Do steps 1 and 2 in early December, the rest in the first working week of January.

1. Write down every network's deadline for the season. One line per network: the rule, the number and the date a 27 November sale settles. This is the whole plan; everything else is execution.
2. Send the CJ corrections before 10 December. If you run CJ on the standard cycle, November's transactions cannot wait for January. Extensions buy another cycle on transactions you cannot judge yet, and only on transactions still in NEW or ACCEPTED.⁹
3. Export pending transactions from every network, on the same day. One file per network, covering 1 November to today. Doing them on the same day is what makes the duplicate check in step 6 possible.
4. Export Shopify orders from a week before Black Friday to today. Include the financial status, the refunded amount and the cancelled date. Starting early catches the order tracked on 20 November and refunded in January.
5. Match on the order reference and sort into four piles. Refunded in full or cancelled; partially refunded; found and kept; not found. Normalise the reference first — #1041 and 1041 have to match.
6. Check the two things the season creates. Duplicates: one order claimed on two networks, which no single network can see, because the same shopper clicked through a deal site and a creator link in the same week. And codes: a creator's Black Friday code posted to a coupon site is the classic January discovery.
7. Decline the full refunds, amend the partials, and write a real reason. A full refund or cancellation is a decline (a correction on CJ, a reversal on impact.com). A partial refund is an amend carrying the new sale amount — on Awin you change the sale amount and the commission is recalculated from it.¹⁰ The reason needs what happened, when, and the number.
8. Upload before the earliest deadline, then check the upload actually took. Awin's batch history lists failed rows; impact's batch answers a late row with ALREADY_LOCKED.⁴ A file that half-failed is worse than one you never sent, because you will believe it worked.

What January validation misses

The six that come up every year.

Refunds that land after the deadline. The structural one, and the only one you cannot fix in January — see the next section.

Partial refunds treated as declines. A customer who returned one of three items kept two. Declining the whole commission is not a correction, it is a mistake in your favour, and publishers notice the pattern.

Exchanges and store credit. With an exchange the customer kept the order value, so there is usually nothing to decline. Store credit is a policy choice. Decide which it is, write it into your programme terms, and apply it the same way every cycle rather than case by case in a busy week.

Gift orders returned by someone else. A return processed under a different name or a gift receipt still refunds the original order, and the affiliate commission is on the original order. Match on the order, not on the customer.

The order that is not in your export. Almost always the export did not go back far enough, or the network's reference is your order name and your export has the order id. Widen the export and check again. Never decline something because you could not find it.

Test and staff orders. Peak season is when teams place the most of them, and a staff order through an employee's own affiliate link is the awkward conversation you want to have in January rather than in an audit.

For the wording on any of these, fifteen decline and amend reasons are written out ready to copy, and de-duplication across networks has the method for step 6.

If your period is shorter than your returns window

This is the one thing on the page worth changing permanently.

If your published returns window is 60 days and your Awin validation period is 30, then every December return is structurally too late: the commission is approved before the customer's right to return has even expired. No amount of January diligence fixes that, because the transactions are already gone.

Awin's own advice is to set the period to your returns window plus about a week.¹ The extra week is the gap between a parcel arriving and the refund being recorded in Shopify, which over the holidays is longer than usual. Where a network lets you choose — Awin, Webgains, Tradedoubler, impact.com, and Rakuten through TrueLock — do the arithmetic once, in a quiet month, and ask for the change before the next season rather than after it. The ceilings are real: 67 days on Awin Access, 50 on Webgains.¹⁶

If you run an extended holiday returns policy — "anything bought from 1 November can be returned until 31 January" — your validation period has to cover that, or the policy is a commission subsidy.

Doing this before AffCheck ships

Plainly: AffCheck is pre-launch. There is no Shopify App Store listing to install today, and this guide is not asking you to wait for one.

What exists now is the free check: it runs in your browser, reads a network export and a Shopify orders export, and gives you the list from steps 5 and 6 — refunds, partial refunds, cancellations, duplicates across networks, codes used by a publisher they were not issued to, and references with no matching order. It suggests; you decide what to decline, in your network, under your own programme terms. Files never leave your device, which is the only sensible arrangement for a spreadsheet full of order numbers.

Without any tool at all, the method is a single sheet: every network's pending rows stacked in one tab with a network column, the Shopify export in another, XLOOKUP on a normalised order reference, and COUNTIF on the reference to find anything claimed twice. It takes an afternoon in January. It is also, every year, the afternoon that pays for itself.