Browser Extensions: What They Find vs What They Skim
Honey, Capital One Shopping, Rakuten, and RetailMeNot Genie all promise to find every coupon at checkout. They do not. Here is what each extension actually does, where the cashback math breaks down, and when you should disable them entirely.
Honey doesn't search the web for the best coupon. It ignores everything that isn't already in its database, and its database is built from manual submissions and partner feeds. If you've ever copy-pasted a code from a forum and watched Honey miss it at checkout: that's why. The extension's marketing implies a real-time scrape of the open coupon ecosystem. The reality is closer to a curated, partner-priority lookup table.
This is not a Honey-specific problem. Capital One Shopping, Rakuten, RetailMeNot Genie, and the smaller players all operate on essentially the same architecture: maintain a list of cached codes per retailer, attempt them in sequence at checkout, surface the one that worked. The differences between them are mostly in three places: which retailers they have negotiated affiliate relationships with, how aggressive their attribution-hijacking behavior is, and what kind of cashback layer sits on top.
The differences matter a lot. They just don't matter in the way the install pages suggest.
What Honey actually does
Honey's checkout flow goes like this. When you arrive at a supported retailer's cart page, the extension fires a request to Honey's backend, retrieves the cached list of codes for that retailer, and queues them for sequential testing. The number of codes in the cache is typically between five and forty. When you click Honey's "Try Codes" button, it iterates through the list, applies each one, captures the response, and stops on the first success.
The codes in that cache come from three sources, in roughly descending priority order: direct partnerships with the retailer (Honey is now a PayPal property and inherits PayPal's merchant relationships), affiliate-network feeds, and user submissions. Codes that originated from a third-party blog or forum and were not submitted to Honey simply are not in the database. The extension cannot find what it does not know exists.
The "best price" feature, which Honey markets as a tool that searches the broader web for a lower price on the same product, has been quietly degraded over the past two years. As of early 2026, "best price" returns Amazon and Walmart matches plus Honey's own retailer-partner network. It does not search the open web. Camelizer (the Amazon-specific tool) has more depth on Amazon price history alone than Honey's "best price" has across its entire claimed scope.
The most consequential thing Honey does that is rarely advertised: it overwrites affiliate cookies. When you arrive at a retailer through a coupon site, a creator's link, or a cashback portal, that referrer typically sets an attribution cookie on your browser. Honey, when it activates at checkout, frequently writes its own attribution cookie on top, which means the commission on your purchase routes to PayPal/Honey rather than to the original referrer. This was the subject of a class-action lawsuit in 2024 and remains the extension's default behavior in 2026.
Capital One Shopping vs. Honey
Capital One Shopping (formerly Wikibuy) operates on similar mechanics with two differences. The cashback layer is more generous on average, and the price-comparison engine actually does compare against a slightly broader set of merchants.
Capital One Shopping's cashback rates run 1 to 8 percent at most major retailers, paid in "shopping credits" that redeem against gift cards. The redemption math is reasonable: $1 of credits typically buys $1 of an Amazon, Target, or Walmart gift card, with no minimum threshold for the smaller denominations. This is meaningfully better than Honey's "Honey Gold" program, which redeems at $10 increments only and tops out at slightly lower headline rates.
The price comparison engine on Capital One Shopping pulls from a real index, not just a partner list. I have tested it on roughly thirty product searches and it surfaced a lower price about 40 percent of the time. About half of those lower prices were on retailers I would actually buy from, which puts the practical hit rate around 20 percent. Not nothing.
The catch with Capital One Shopping is the same catch with Honey: attribution. The extension overwrites referral cookies aggressively. If you are shopping through a creator link or a cashback portal already, Capital One Shopping will hijack the commission unless you explicitly disable it on the checkout page.
Rakuten: a different category
Rakuten is not really a coupon extension. It is a cashback portal with an extension wrapper. The mechanic is different in a way that matters: Rakuten has direct affiliate agreements with thousands of retailers, and the cashback is paid in cash (or PayPal) on a quarterly basis rather than in store credit.
The extension activates on supported retailer sites with a popup that says "Activate Cashback." Clicking it sets the Rakuten attribution cookie. If you do not click, you do not get the cashback. The extension does not search for coupons in the way Honey does — it surfaces the publicly-listed promo codes that Rakuten has cataloged for that retailer, but the codes are also visible on rakuten.com without the extension.
Cashback rates at Rakuten run 1 to 12 percent depending on retailer, with periodic "double cashback" promotions that push major retailers like Macy's, Nordstrom, and Sephora to 6 to 10 percent. The redemption is meaningful — quarterly Big Fat Checks of $50 to several hundred dollars are realistic for a household that does most of its shopping through the portal.
The honest tradeoff with Rakuten: you have to remember to click the activation button, and you cannot stack Rakuten cashback with credit card portal cashback most of the time, because the attribution cookie races. Whichever cookie was set last wins. If you click through Rakuten and then check the Chase Ultimate Rewards portal afterward, the Chase portal will overwrite the Rakuten cookie, and vice versa.
RetailMeNot Genie
RetailMeNot Genie is the lightest of the four. It surfaces RetailMeNot's existing public coupon database at checkout, with a "verified" tag on codes that have recently worked. The extension does not have a meaningful cashback layer (small "GeniePoints" exist but the redemption ladder is not worth the storage).
What Genie does well: surface the broader coupon database, which is larger than Honey's. RetailMeNot has been collecting user-submitted codes since 2009, and the deep historical archive sometimes contains codes that other extensions miss. The "verified" flag is not perfectly reliable but it is more accurate than no flag at all.
What Genie does poorly: it does not auto-apply. Every code is a copy-paste action by the user. The extension's value over a manual visit to retailmenot.com is small.
The cashback math vs. card portals
The cashback layer on Honey, Capital One Shopping, and Rakuten exists in tension with the cashback layer on credit card shopping portals — Chase Ultimate Rewards, Amex Offers, Citi Bonus Cash Center, Capital One Offers (separate from Capital One Shopping), and the bank-specific portals like Bank of America's BankAmeriDeals.
The math, simplified: a credit card shopping portal typically pays 2 to 10 percent in points or cash. An extension cashback layer typically pays 1 to 8 percent. They compete for the same attribution cookie. You get one, not both, and the cookie war is determined by whichever was set last in your browser session.
The practical play for a household that wants to maximize: check the credit card portal first for the retailer in question. If the card portal rate beats the extension rate, click through the card portal last and let the extension surface coupon codes only — do not let the extension capture cashback attribution. Most extensions allow you to disable the cashback prompt on a per-retailer basis. Honey's "Pause on this site" function works for this. Capital One Shopping's "Don't activate" button on its popup achieves the same thing.
If the extension cashback rate beats the card portal rate, click through the extension last. Be aware that the extension will also intercept any creator or coupon-site referral, which means you are taking commission away from whoever surfaced the deal to you.
Privacy: what gets sold
All four extensions collect data on what you browse and buy. The disclosures are buried in the terms of service. The data is monetized in different ways.
Honey shares purchase-level data with PayPal, its parent, and uses it for cross-merchant analytics. The data is not sold to third parties in the literal sense, but it is used to inform which merchants Honey courts for partnership and which products Honey promotes through its "Droplist" feature.
Capital One Shopping's data goes to Capital One's analytics infrastructure. Capital One has stated that Capital One Shopping data is not used for credit underwriting. There is no public disclosure on whether it is used for marketing within the bank.
Rakuten sells aggregated, anonymized purchase data to retail intelligence firms. This is disclosed in the privacy policy. The data is technically de-identified, but the granularity (purchase amount, category, timestamp, location) is high enough that re-identification is theoretically possible at the household level. For most users this is a low-risk disclosure. For users who care intensely about purchase privacy, it is a real one.
RetailMeNot's data flow is the least transparent of the four. The privacy policy permits broad sharing with "advertising partners." There is no detailed disclosure of what data leaves the extension.
Performance impact
All four extensions add measurable latency to retailer checkout pages. Honey adds the most — the cart-page activation request typically adds 200 to 600 milliseconds, with the actual code-testing loop adding another 5 to 30 seconds depending on how many codes are queued. Rakuten adds the least, because its activation flow is a single cookie write.
For users on slower connections or older devices, the cumulative impact is noticeable. The extension cart-page experience can take twenty seconds to settle. If you have ever wondered why your cart page felt sluggish after installing an extension, this is why.
When not to install any of them
Three categories of shopping where the extensions actively hurt you.
Smaller D2C brands and Shopify stores. The extensions usually have no codes for these retailers, but they will still hijack any referral attribution. If you reached the brand through an Instagram post, a newsletter, or a coupon site, the extension will likely overwrite that referrer. The brand pays a commission to the extension instead of to the affiliate channel that actually drove the sale. You get nothing.
Limited-edition drops. Sneaker drops, Supreme releases, scarce inventory of any kind. The extension's checkout-modification behavior frequently triggers anti-bot systems, and the cart can be flagged or canceled. The latency added by extension activation is also enough to lose a queue position on a high-demand drop.
Retailers where you have a manually-discovered better code. If you found a 25 percent code on a forum or in a newsletter, the extension may overwrite it with a 10 percent code that worked first in its sequence. Some extensions have improved their "stacking" logic to test against an existing code, but the default behavior is still to apply the first cached code that succeeds, which may not be the best one.
The one trick worth knowing
Some extensions, when toggled into developer-mode or when their popup is closed mid-checkout, will delete the affiliate cookie they set. The mechanic is unintentional but documented in several browser-extension forums. If you have already let an extension capture attribution and want to free the cookie for a card portal click-through, opening browser dev tools, finding the extension's storage entry, and clearing it will sometimes restore the previous referrer. This is fragile and changes between extension versions, but it is the cleanest path back when you have set the wrong cookie order.
The bottom line
Honey is a coupon database with attribution-hijacking behavior. Capital One Shopping is the same with a slightly better price-compare and a slightly better cashback redemption ladder. Rakuten is a real cashback portal with a thin coupon layer on top. RetailMeNot Genie is a code-surfacing tool with no meaningful cashback.
For a household that mostly shops at major retailers and is willing to manually route attribution: Rakuten in the browser, credit card portal as the final click-through, and a manual coupon search in a side tab. The CouponHive search runs the same query Honey would run plus the open web, which means the codes you see at checkout will frequently be better than what any extension surfaces automatically.
If you only install one extension, install Rakuten and discipline yourself to click "Activate" on each shopping session. If you install zero extensions, you will be slightly slower at checkout and slightly richer at the end of the year.