Why Most "Verified" Codes Aren't: An Audit of 1,000 Listings

When a coupon site shows you a "verified 3 hours ago" badge, ask yourself: by whom? In our audit of 1,000 listings across the top six coupon aggregators between January and March of 2026, the answer was usually "a script that pulled the code from another site that pulled it from a third site." Not a single click test. Not a single browser session. Just a timestamp generated by a cron job.

Of those 1,000 codes, all of which carried some flavor of "verified," "tested," "working," or "checked today" badge at the moment we collected them, 438 produced a successful discount on a real test order. Another 162 produced a discount that was smaller than advertised — typically a stacked sitewide discount being represented as a category-specific one. The remaining 400, exactly 40 percent of the sample, did nothing. They were either expired, geo-restricted to a market we weren't shopping from, tied to a minimum cart we hadn't met, or had simply never existed in the first place.

That is the headline number for this audit: when a major coupon site tells you a code is verified, it has roughly a 56 to 60 percent chance of working as advertised. The badges, in other words, are about as reliable as a coin flip with the coin slightly weighted in your favor.

How "verification" actually works at most sites

To understand why this happens, you have to understand the supply chain behind a coupon listing. Most aggregator sites do not have an editorial team typing codes into a checkout form. Most of them ingest data — through some combination of the following:

Affiliate network feeds. Networks like CJ Affiliate, Rakuten Advertising, Awin, ShareASale, and Impact provide structured feeds to publishers. Those feeds include the merchant's currently authorized promo codes, expiration windows, and terms. In theory this is the cleanest source. In practice, the feeds lag — merchants change codes faster than feeds refresh, and several of the networks we examined update on a cadence of 24 to 72 hours. A code that was killed Tuesday morning may still be in the Friday feed.

Scraping competitors. A surprising amount of the coupon ecosystem is sites scraping each other. We watched a single typo — a code listed as "SPRING25!" with an exclamation point that the merchant's checkout did not accept — propagate across four of the six audited sites within 36 hours. None of them caught the typo. None of them tested it.

User submissions. Some sites allow shoppers to upload codes. These are nominally moderated, but the moderation typically involves checking that the field isn't empty and that the code matches a regex. There is no test order.

Press releases and merchant announcements. When a retailer issues a "GET 20% OFF WITH CODE FALL20" press release, every aggregator picks it up. The aggregators rarely circle back when the campaign ends.

The "verification" that gets stamped on top of all this is, in the cases we examined, almost always automated. The script checks one of three things: that the code field still exists in the affiliate feed (which tells you nothing about whether it works), that a competitor site still has the same code listed (which tells you only that the competitor hasn't checked either), or, in a small minority of cases, that an HTTP request to the merchant's coupon-validation endpoint returns a non-error response. That last one is the closest to actual testing, but it has a fatal flaw: many merchant endpoints return 200 OK for any code, then deliver the discount-or-not logic at the cart stage. Hitting the endpoint isn't the same as completing a checkout.

The audit method

We built our sample by pulling, on three randomly selected days in February 2026, the top 20 listings for each of 50 stores across six major aggregator sites. The 50 stores were chosen by 2025 US e-commerce search volume — Amazon, Target, Walmart, Macy's, Nordstrom, Kohl's, Old Navy, J.Crew, Sephora, Ulta, Best Buy, Home Depot, Lowe's, Wayfair, and a long tail down to specialty retailers like Allbirds and Bombas. We removed duplicates within a site (the same code listed three times under three category tags) but kept duplicates across sites, because cross-site duplication is itself a finding.

Each of the 1,000 unique listings was tested by a researcher in a fresh, cookie-cleared Chrome session, geolocated to a US state matching the listing's stated targeting (or to the broadest "US" if no state was specified). The researcher built a cart that satisfied any stated minimum, applied the code at checkout, and recorded the outcome at the final review screen — without completing the purchase. Every test was timestamped and screenshotted.

We did not contact any of the audited sites in advance. We are not naming them in this article because the goal here is not to litigate against any particular publisher; the pattern is industry-wide, and singling out two or three would imply the others are clean. They aren't.

What "verified" actually looked like

A few patterns emerged from the failures.

The zombie code. A holiday code from late 2025 — "BLACKFRI30" or similar — still listed as "verified" in mid-February. We found 73 of these in the sample. These are the easiest failures to detect from the user side: if a code's name references a season or holiday that has passed, treat it as suspect.

The phantom percentage. A listing claims "30% off sitewide" but the code, when applied, produces 15 percent off, or 30 percent off only one category, or 30 percent off only with a stacked email-signup. Eighty-one of our successful applications fell into this bucket — they "worked" in the sense that something happened, but they did not deliver the advertised value. We classified these as partial successes in the 162 figure cited earlier.

The first-order-only code. Twenty-nine codes worked perfectly — for an account that had never ordered from the merchant before. The listings did not disclose the new-customer-only restriction. From the perspective of a returning shopper, these are non-functional, and a verification process worth the name should at least flag the restriction.

The geo-locked code. Forty-seven codes were valid only in specific states or for specific shipping addresses, again undisclosed. International shoppers reading US coupon sites encounter this constantly, but we found plenty of US-to-US restrictions too — California-only codes listed as national, Hawaii and Alaska excluded without note.

The minimum-cart shell game. Listings frequently understate the minimum order. A listing says "20% off orders $50+" but the actual code requires $75. This is a small lie that compounds: shoppers add a filler item to clear what they think is the threshold, then are stuck either abandoning the cart or buying the filler at non-discounted prices.

The "code not required" surprise. Forty-four of our 1,000 listings produced a discount that was already applied automatically at the merchant's site, with no code needed. The aggregator, having found a sale, fabricated a code listing around it — sometimes with a code field that the merchant accepted as a no-op, sometimes with an empty code that the aggregator labeled "click to reveal," which on click revealed nothing more than a tracked outbound link. These are not strictly fraudulent, but they are misleading. The shopper believes a code-bearing exclusive exists where none does.

What real verification looks like

We talked to editors at three publications that do test their codes — two niche personal-finance sites and one membership-based deal community — and the workflow is genuinely labor-intensive.

A real verification process, at minimum, involves a test order in a fresh browser session, geolocated to the targeted market, with a cart that satisfies the stated minimums and reflects a realistic basket for the merchant's typical customer. The tester applies the code, captures the discount delta at the final review screen, and records the result with a timestamp. Codes that pass are tagged with the test date; codes that fail are removed from the listing or annotated with the failure mode.

For high-traffic codes, retesting happens daily or every other day, because merchant promo logic changes. For long-tail codes, weekly is typical. The editors we spoke with all flagged the same operational problem: a single tester can verify maybe 80 to 120 codes per shift, depending on merchant checkout complexity, and most aggregators are listing tens of thousands. The math doesn't work without a team of dozens, and the unit economics of an ad-supported coupon site usually don't support that headcount. So most don't have it.

How to spot a stale listing yourself

A few heuristics held up well across our audit.

Check the code's name. If the code references a season, holiday, or campaign name that's expired, the code is almost certainly dead. "SUMMER" codes in February. "BTS" (back-to-school) codes in March. "VDAY" codes after February 14. The merchants kill these on schedule even when aggregators don't update.

Check the comments or success-rate widget. Several aggregators surface user thumbs-up/thumbs-down votes. A code with 200 recent thumbs-down and 12 thumbs-up is dead, regardless of what the badge says. We found a strong correlation between user-reported success rates and our test results — when users said a code didn't work, our researchers usually agreed.

Check the merchant's own site. This sounds obvious, but: most merchants will tell you, on a "current promotions" or "deals" page, what codes are actually live. If the aggregator's "verified" code isn't on that page, it's probably not live.

Check the disclosure. Listings that include specific restrictions ("excludes clearance," "first orders only," "minimum $75 cart") are more often real than listings that promise "20% off everything." Real merchant codes almost always have restrictions; aggregator-fabricated ones often don't, because the fabrication is doing the simplification.

Check the timestamp claim against the rest of the page. If every listing on a store page says "verified 2 hours ago," the timestamps are almost certainly script-generated. Real verification cadence varies by code — some get checked twice a day, some once a week — and a homogeneous timestamp is a tell.

What we changed in our own listings

CouponHive is not exempt from the structural problems described above. We use affiliate-network feeds, like everyone else, and we have a finite editorial budget. What we have done in response to the patterns we found in this audit is publish the date and method of the most recent check on every listing, distinguish between machine-validated and human-tested codes with separate badges, and surface user-reported failures within an hour rather than burying them. We also removed the homogeneous "verified hours ago" widgets from listings that hadn't actually been retested in that window — if a code was last tested four days ago, it now says four days ago.

The broader takeaway from spending three months looking at this is that the coupon-aggregation industry has been telling shoppers a story about freshness and verification that the underlying process does not actually deliver. The badges are theater. The supply chain is industrial-scale copying. And until the badges mean what they say — until "verified" maps to a real test order with a real cart and a real timestamp tied to that test — shoppers are right to assume that any given code is roughly a coin flip, and to plan their checkout strategy accordingly. Stack two or three codes if you can. Save the cart and try again tomorrow if the first one fails. And don't take the badge at face value, regardless of the brand on the badge.