The two-column answer: when credits win and when the card wins
Credits win when the money is going straight back into the same account, when the amount is small, and when speed matters more than the route the money travels. The card wins when the order is finished, when the amount is large, or when you may not order again.
The dividing line is simple enough to write on one line: credits for a balance that stays, the card for a balance that leaves. Everything else in this article is an exception to that line.
Our own three refunds split two to one in favour of credits on speed alone, with credits arriving inside 30 minutes twice against card times of 4, 9 and 21 days (Recorded, 3 refunds, orders placed 2026-05-02 to 2026-08-11).
Add a third question to the two-column answer: whether you will order again inside the same quarter. A shopper who will is better served by credits whatever the amount, and a shopper who will not is better served by the card whatever the speed.
1. Write down when you expect to order again. 2. Write down the size of the refund. 3. Choose the card when the first answer is never. 4. Choose credits when the second answer is small.
Minutes inside the account against days across two banks
Credits move inside one ledger. The account adds a figure and the balance changes, with no bank, no card scheme and no statement date in the path, which is why the recorded arrivals were measured in minutes rather than days.
1. Cancel or adjust the order. 2. Confirm the refund route at that moment. 3. Watch the credits balance rather than the statement. 4. Screenshot the balance once it changes.
A card refund crosses two institutions. The first release is the slow part and the second is the statement cycle, which is why an order cancelled on day one can appear on a statement three weeks later without anything having gone wrong.
Compare the two routes on certainty rather than on speed alone. A credits refund inside the account is close to certain once it is processed, while a card refund depends on a statement cycle you do not control, and certainty is worth something on an amount you are counting on.
The expiry clock, and what a balance does not tell you
A balance that never expires is worth face value and a balance with a deadline is worth less. Whether credits carry an expiry date, and how long it runs, is something our own records cannot settle (Not verified).
Treat the question as live until you have seen an answer on your own account. The cost of assuming wrongly is asymmetric: a balance spent this month is worth its full figure, and a balance discovered after a deadline is worth nothing.
A practical habit replaces the missing rule. Spend a balance within the same month it appears, or leave a note with the date it arrived, and the question stops mattering either way.
Read the balance against your own note. Where a refund of £40.00 produces a balance that reads £40.00 on the same day, the route has behaved; where it reads £34.90, the difference is a coupon that did not return, and that is worth knowing before the next order rather than after it.
Treat a large balance as cash with an unknown deadline. The uncertainty is the reason a four-figure balance sitting unused is a worse position than a smaller one spent this month, whatever the expiry rule turns out to be.
A balance can only be spent in one place, and that is the strongest argument on either side of this question. Credits that land in minutes are worth nothing to a shopper who has finished with the account, while a card refund that takes three weeks is worth exactly what it says on the statement whenever it finally arrives.
What a refund does to a coupon that was already used
A coupon reduces the amount paid, so a refund can only return what was paid. Where a code took £5.00 off a £45.00 order, the refund returns £40.00, and whether the code itself becomes usable again is Not verified on our side.
1. Note the amount actually paid rather than the basket total. 2. Note which coupon was applied. 3. Compare the refund with the paid figure. 4. Record any difference as a coupon that did not return.
Plan without the code. An order built around a coupon that will not return is a different order, and treating the code as consumed before the refund question comes up keeps the arithmetic honest.
Ask for the coupon question in writing while the order is live. The answer costs one message and it removes the guesswork from every refund afterwards, which is worth more than the coupon itself on any basket above about £50.00.
Ask the expiry question in writing and keep the reply with the order number. An answer that names a period turns a balance into a dated figure, and an answer that names none is worth the same as no answer, because a rule nobody has written down cannot be relied on when a refund is queried three months later.
Three exceptions that break the simple rule
Exception one: a refund larger than the payment. Where a goodwill credit was already added to the same order, a full refund can return more than the card was charged, and the surplus lands as credits whatever route was chosen.
Exception two: a partly shipped order. When two of four lines have already left, the refund covers the lines that never shipped, and the freight is not apportioned back to them in any way we could predict.
Third in the list is a card that has since been replaced. A refund sent to a closed card can come back and re-enter the account as credits after a delay we have not measured (Not verified).
The second exception is a partly shipped order, where two of four lines have already left and the freight is not apportioned back to the lines that never shipped. The refund then covers goods rather than goods and carriage, and the difference can be a whole opening rung.
Third on the list is a card that has since been replaced. A refund sent to a closed card can return to the account as credits after a delay we have not measured, which turns a card choice into a credits outcome without anyone asking (Not verified).
The fourth exception is a refund spent before it was checked. A credit that arrives and is used on the next order cannot be queried afterwards without reopening the whole calculation, so a balance read on the day it changes is worth more than the same balance read three orders later when the figures have been mixed together.
Keeping a credits balance visible between orders
A balance that sits behind a login is a balance you will forget. Once a month, open the account, read the figure, and write it beside the date in a note you already keep, which takes under a minute.
Compare the figure with what you expected. Where a refund of £40.00 produced a balance of £34.90, the £5.10 difference is the part worth chasing while the order is still recent.
Keep the route decision with the figure. A note that reads 40.00 credits on 2026-08-11 is worth more than a memory of having chosen credits once, because it dates the decision as well as the amount.
Re-read the note at the end of the month rather than at the end of the order. A balance that has been sitting for three weeks is a decision waiting to be made, and the note is what turns it back into a number rather than a vague sense that something is owed.
The data point behind this note
In 3 recorded refunds the credits route delivered inside 30 minutes twice and the same evening once, against card times of 4, 9 and 21 days (Recorded, 3 refunds).
Rates last checked 2026-10-02. Where a figure is community-reported we say so; where we could not verify it, we write Not verified instead of estimating.