A shopper opens her banking app, sees a tile that reads "10% back at a coffee chain," and taps it. Days later she buys a coffee and pays with the card she always uses. A statement credit appears. She never clicked an ad, never landed on a tracking URL, never carried a cookie from a publisher to a checkout. The reward attached to her card, and the swipe itself told the system a sale had happened.
That is a card-linked offer, and it is the one corner of affiliate marketing that does not care whether the cookie lives or dies. The purchase is the conversion event. There is nothing else to track.
Where card-linked offers came from
The model is older than most marketers assume. Cardlytics, the company that did the most to define the category, was founded in 2008 in Atlanta by two former Capital One executives, Scott Grimes and Lynne Laube. Their idea was specific: banks sit on a complete, deterministic record of where their customers spend, and that record is far more reliable than anything an ad network can infer. If a bank could surface merchant offers inside its own online and mobile banking, the bank would earn new revenue, the merchant would pay only when a real customer bought something, and the shopper would get cash back through a channel she already trusted.
That three-sided arrangement, the bank, the advertiser, and the cardholder, is still the shape of the business. The early years were a grind of bank integrations, because nothing happens until a financial institution agrees to put offers in front of account holders. By the 2010s, Cardlytics powered the offer programs sitting inside large US banks, and the card networks themselves began building comparable infrastructure.
The reason this matters now is not nostalgia. Card-linked offers were a niche curiosity when third-party cookies still worked. They became strategically interesting the moment cookie-based tracking started to decay, and that is the present.
How the mechanics actually work
Strip away the marketing language and a card-linked offer is a deal between four parties: a merchant that wants customers, a platform that runs the offers, a card network or issuing bank that sees the transactions, and the cardholder.
The flow runs like this. A merchant funds an offer, say 8% back, and sets the rules: which merchant identifiers count, which spend threshold qualifies, how long the campaign runs. The offer is distributed into places where people have already registered a payment card, usually a banking app, a card issuer's portal, or a fintech rewards app. The cardholder activates the offer, or in some programs is automatically enrolled. Then she goes about her life.
When she pays with the linked card, the transaction passes through the card network's authorization and settlement stream. The offer platform reads that stream, matches the transaction against the campaign rules by merchant ID, category code, and amount, and if it qualifies, issues the reward as a statement credit or points. The merchant pays a commission only after the qualifying sale clears.
Visa describes this plainly in its own developer documentation. The Visa Offers Platform monitors the VisaNet authorization stream for enrolled, consented cardholders, qualifies purchases against rules covering merchant category, specific merchant, location, and ecommerce activity, then uses a Rewards API to issue statement credits for qualifying transactions. Mastercard Offers runs a network-agnostic version of the same idea, delivering merchant-funded offers across payment types with its own card-linking and redemption layer. American Express runs Amex Offers on its closed-loop network, where the issuer and the network are the same company.
Notice what is absent. No tracking pixel fires on a landing page. No cookie has to survive a 30-day window. No browser extension can overwrite attribution at checkout. The conversion signal is the cleared payment, observed by the entity that processes the payment. As Cardlytics puts it, every dollar is directly attributable to a sale, and the merchant does not pay until the qualifying sale occurs. For an affiliate channel that has spent two years arguing about who deserves the last click, that is a different universe.
There is one more property worth naming. A card swipe in a physical store is the same kind of event as a card payment online. Both move through the same authorization rails. So a card-linked offer tracks an in-store purchase exactly as well as an online one, which is something cookies have never been able to do.
Where it stands now
Two forces have pushed card-linked offers from the margins toward the center of the partnership conversation.
The first is the decay of cookie tracking. Even after Google reversed course on removing third-party cookies in Chrome, Safari and Firefox still block them by default, and AI-mediated shopping journeys increasingly happen with no clear click at all. Content affiliates have watched organic traffic erode and last-click attribution lose its meaning. Against that backdrop, a model where the transaction reports itself, with no intermediary signal to lose, looks less like a curiosity and more like an insurance policy.
The second is that the affiliate networks have stopped treating card-linked offers as a separate world. The infrastructure now plugs into the same plumbing as traditional affiliate programs. Networks including Awin and impact.com have built CLO transactions into their platforms, so card-linked purchases flow into the same dashboards, the same commission structures, and the same validation and approval workflows a brand already uses for content and coupon partners. A bank or rewards app can join a standard affiliate program as a publisher without anyone building a custom integration. The UK's Affiliate and Partnership Marketing Association named card-linked offers one of its 2026 affiliate growth drivers, pointing to the volume moving through the space and the new specialist publishers being built around it.
The publishers, in affiliate terms, are the places that hold the registered cards. Banks are the classic example: Chase Offers and My Wells Fargo Deals are both powered by Cardlytics behind the scenes, and Citi runs its own Citi Merchant Offers. The relationships are not permanent. Bank of America, a Cardlytics partner since 2010, took its BankAmeriDeals program in-house in early 2026, a reminder that a platform's reach depends on bank contracts it has to keep winning. Then there are the card-issuer programs that are also networks, Amex Offers being the clearest case. And there is a fintech tier, the rewards apps and wallets, Cash App Boosts among them, plus card-linked dining programs and point-earning wallets, that exist specifically to get users to link a card in exchange for cash back or points. A useful detail from the practitioner community: programs that run on the same underlying tracking network usually cannot stack with each other, because the same transaction cannot be claimed twice.
Card-linked offers also rely on real platform companies. Cardlytics is the largest and most visible, having extended its reach by acquiring the rewards app Dosh and, in a 2021 deal worth roughly 350 million dollars plus earnouts, the customer-data company Bridg, which adds SKU-level purchase detail to Cardlytics' broader view of where people shop. Newer offer platforms compete on faster bank integrations and cloud-native setups. The card networks themselves sit underneath all of it, as both the rails and, increasingly, direct competitors with their own offer products.
The honest limits
Card-linked offers solve the attribution problem so cleanly that it is tempting to oversell them. The limits are real, and a serious marketer should understand them before shifting budget.
The first is incrementality, and it is the same uncomfortable question that haunts coupon and cash-back affiliates. The card network can prove a transaction happened. It cannot prove the offer caused it. If a shopper was always going to buy that coffee, the 8% back is not new revenue, it is a discount handed to a customer who would have paid full price. The mechanic that makes card-linked offers measurable, observing real purchases, does not by itself separate purchases the offer created from purchases it merely witnessed. Critics of the category have argued for years that weak targeting routinely rewards customers who were already coming, and that is the most frequent merchant objection. The defense is to run proper incrementality tests, holdout groups and exposed-versus-control comparisons, rather than trusting redemption counts. Without that discipline, a card-linked program can look like a triumph and still be paying for demand a brand already had.
The second limit is issuer dependence. A card-linked program reaches exactly the cardholders whose bank or app has agreed to carry the offer, and not one more. The publisher side of this channel is concentrated in a small number of large financial institutions and a handful of platforms. That gives the issuers and platforms real pricing power, and it means a merchant's reach is gated by deals it does not control. It also means the economics can be heavy. Industry critics have noted that the platform and the financial partner together can require a take that is a sizable multiple of the discount the shopper actually sees, which makes the true cost of the channel higher than the headline cash-back rate suggests.
The third limit is merchant fit. Card-linked offers work best where the merchant identifier is clean and consistent, the purchase is frequent and low-consideration, and a small percentage back is enough to nudge behavior. Coffee, quick-service food, fuel, drugstores, everyday retail. The model fits those neatly. It fits a rare, high-consideration purchase far less well, and it struggles when a merchant's transactions are messy in the payment data, for example when a purchase clears under a payment processor's name rather than the merchant's. Rewards also arrive after settlement, not at the register, so a card-linked offer cannot deliver the instant satisfaction of a code applied at checkout.
There is a market-health signal worth weighing too. Cardlytics, the category's bellwether, has been contracting rather than booming. Its total revenue fell about 10% in 2024 to 278.3 million dollars, down from 309.2 million dollars in 2023, and declines continued through 2025, with third-quarter revenue down 22% year over year according to the company's own SEC filings. Card-linked offers being structurally well suited to a cookieless world does not automatically make every operator in the space healthy. Execution, bank relationships, and consumer-incentive costs all still matter.
What to take from this
Card-linked offers are not a replacement for content affiliates, coupon sites, or creator partnerships. They are a distinct instrument with one genuinely valuable property: the conversion event is the cleared payment, observed on the payment rails, so there is no fragile signal between the offer and the sale to lose. That makes them resilient as cookie tracking decays and as AI-mediated journeys erase the click. It also makes them the only affiliate mechanism that tracks an in-store purchase as faithfully as an online one.
The catch is that resilient tracking is not the same as proven value. A card-linked program measures purchases with near-perfect fidelity and still tells you nothing about incrementality unless you build holdout testing into it from the start. Treat the channel as what it is: a high-fidelity, online-to-offline performance instrument with a narrow merchant fit and a real dependence on the banks and platforms that hold the cards. Used inside a brand's broader partnership mix, with incrementality tests deciding whether it earns its budget, it is a sound addition. Used as a vanity metric of redemptions, it will quietly pay for customers a brand already owned.
Council summary
This post argues that card-linked offers are the one affiliate model immune to cookie decay, because the conversion is the cleared payment itself, observed on the card network's own rails, and that this strength is separate from the harder question of whether the offer caused the sale. The council verified the founding facts (Cardlytics, 2008, Atlanta, ex-Capital One founders Scott Grimes and Lynne Laube), the Bridg acquisition at roughly 350 million dollars plus earnouts, and the revenue figures: 278.3 million dollars in 2024, down 10 percent from 309.2 million dollars in 2023, with third-quarter 2025 revenue down 22 percent year over year. One correction was made: Bank of America took its BankAmeriDeals program in-house in early 2026, so it is no longer powered by Cardlytics, and the publisher example was rewritten to reflect that. The Visa Offers Platform mechanics, the named bank programs, and the APMA's 2026 growth-driver call all check out against primary sources. The reader takeaway is to treat card-linked offers as a high-fidelity online-to-offline instrument with narrow merchant fit, and to fund it only when holdout tests prove it drives incremental sales rather than discounts demand a brand already had.
Comments