
You copy a code, paste it at checkout, and the cart rejects it. The page you found it on still lists it. We run a deal site, so we wanted to see how that happens from the other side. In August 2026 we took a hard look at our own database and found tens of thousands of offers we still called active that our coupon feed no longer carried.
Here is what we found, what we changed, and what we still can't do. We can't see inside other coupon sites, so this is about ours.
In our database, dead offers stayed listed because the record never showed they had ended. On August 19, 2026, 94% of our active offers had no end date, and our coupon feed reports a removal only once, so a missed one stays missed. When we compared against a fresh full copy of that feed on August 20, 2026, we expired 56,729 offers it no longer carried. We now expire offers the feed reports as suspended or past their end date, and each deal page shows when the offer was last confirmed. We still do not test codes at checkout.
Most offers arrive with no end date
As of September 28, 2026, we have pulled 681,368 offers from affiliate feeds: CouponAPI, CJ, Rakuten, Awin and Sovrn. Each feed syncs once a day. Some offers come with an end date. Most do not.
End dates make expiry easy. The date passes, we mark the offer expired. The trouble is how rarely that date exists. On August 19, 2026, 485,243 of the 514,701 active offers in our database, or 94%, had no end date at all.
An offer with no end date looks the same on its last day as on its first. Nothing in the record says it is over. So the obvious fix, checking the date, could only ever catch a small slice of the dead ones.
Removals are sent once
Our coupon feed is incremental. Each call reports what changed since the last call: new offers, changed offers, and offers that were removed. It does not hand over the whole catalogue every day.
That works until a removal gets missed. A missed removal is not sent again. The offer sits in the database marked active, and nothing ever arrives to contradict it. We had missed a lot of them, and part of that was our own doing: from February to August 2026, our import ignored the removal flag the feed was sending. We fixed that on August 20, 2026.
What a full recount turned up
On August 20, 2026, we pulled a fresh full copy of that coupon feed and compared our whole database against it. Any offer we held as active that the feed no longer carried, we marked expired. Earlier that same day we had also stopped carrying 26 stores, for reasons unrelated to dead codes, which is why the starting count below is far lower than the August 19 figure.
| Date | What we measured | Result |
|---|---|---|
| August 19, 2026 | Active offers in our database with no end date | 485,243 of 514,701 (94%) |
| August 20, 2026 | Offers expired because the full feed no longer carried them | 56,729 |
| August 20, 2026 | All our active offers, before and after the recount | 71,232, then 17,397 |
About four in five of all the active offers we held that day were gone from the feed. Every one of them was still marked active on our side.
That is a dead code seen from the inside. The store ends the offer. The removal never lands. The listing stays up because nothing tells it to come down, and a shopper finds out at checkout.
What we do now
- An offer is marked expired when the feed reports it as suspended, or when its end date passes. Feeds sync once a day, so a change reaches us at the next sync, not the moment a store makes it.
- Fresh offers go first. The Real Deals board, our homepage and each store's Approved Specials favor offers the feed has confirmed in the last 14 days. Favor is the accurate word. Something older can still turn up there, so the date on the deal page still matters.
- Every deal page shows when the retailer feed last confirmed the offer. If that date is weeks old, treat the offer as unconfirmed.
What we still cannot do
We do not test codes at checkout. Nobody here, human or software, types a code into a store's cart to see whether it goes through. Our method page puts it plainly: "Neither a score nor a Gazoobie Pick badge proves that a price beats other shops, that a coupon works at checkout, or that every offer in our database has been reviewed."
So a code can pass every check we run and still fail on your cart. The store may have ended it since our last sync. It may exclude the item you picked. It may be for new customers only.
If a tested code matters most to you, some services say they do the testing. Coupons.com says "an editor then runs it through the brand's real checkout" (its verification page). RetailMeNot says its team "verifies codes manually," and its Rewards app pays a $5 credit when a code fails, under conditions set out in its terms. Honey and Coupert both run browser extensions that say they try codes at checkout for you. We went through several of these in our look at RetailMeNot and its alternatives.
How to spot a dead code before checkout
None of this needs a special tool. It needs a minute before you copy anything.
- Look for a date. A recent "last checked" or "last confirmed" date is worth more than a long list of codes. Every Gazoobie deal page shows one. No date at all tells you nothing either way.
- Read the terms before the code. Minimum spend, excluded brands, new customers only, app only, one use per account. A live code can still refuse your cart.
- Check the store's own site. Look at its homepage and its promotions or sale page. If a code promises far more than anything the store is advertising, be skeptical.
- Read the code itself. One with an old year or a past holiday in its name probably belongs to a campaign that has ended.
- Be careful with codes that look personal. Welcome codes and referral codes can be tied to one email address or account, so a copy posted online may be refused.
- Apply it early. Enter the code in the cart before you type in card details. If the total did not drop, the code did not work, whatever the message says.
- Check the price, not only the code. A code taken off a high starting price can still leave you paying more than another store charges. Here is how to tell whether a discount is a good price.
The price a discount starts from
Dead codes fail loudly, at checkout. Reference prices are a separate and quieter problem. Where a feed supplies both a reference price and a sale price, we check whether the claimed discount adds up.
We also track how sale prices behave over time. In our sale price study of priced listings from our affiliate product feeds, August 7 to September 19, 2026, 3,004 of 4,157 discounted listings seen on 30 or more separate days (72.3%) never changed price once. The bigger the advertised discount, the more likely the price never moved at all. The study names no retailer. Treat a percentage off the way you treat a code: as a claim to check.






