Citi's card-linked offer acquisition bets on cost, not on proven category economics

Citi has agreed to acquire Kard Financial and describes the result as strengthening its commerce ecosystem. The word doing the work across this category is media. It is the wrong word, and the distance between it and what card-linked offers actually are explains fourteen years of disappointing outcomes.

Media sits before the purchase decision. Retail media works because the retailer owns the shelf, the search result and the checkout, so the message arrives while the choice is still open. An issuer owns the moment of paying, which is downstream of every decision that mattered.

I have spent the better part of eighteen years on the network side of this problem, working to assemble loyalty ecosystems with merchant-funded and card-linked capability inside them. The conversation that ends these programmes is always the same conversation, and it is never about technology, targeting or delivery. A merchant asks what it is actually buying, runs a holdout, and finds that measured lift does not cover the rebate plus the platform fee. What it has discovered is that it paid to influence a choice already made. Position in the funnel is not a parameter you can tune, and no amount of machine learning moves you upstream of it.

This also reframes the category's success stories, and not in their favour. We do not have evidence that merchant-funded offers work as a standalone business. We have evidence that firms with another reason to run them can sustain them. Amex Offers is a retention feature inside closed-loop economics and has never had to clear on its own. Chase Media Solutions is roughly two years old with nothing disclosed. The one pure-play required to stand unsupported was Cardlytics, and its filings are unambiguous: no profitable year since its 2018 IPO, revenue down from a $309.2m peak in 2023 to $233.3m in 2025, cumulative net losses of roughly $1.15bn, total assets down from $1.26bn to $285.6m, and shareholders' equity negative last year. Operating cash flow turned positive in 2025, which is something. It is not a proof of concept.

The reasonable inference is that this category has never been proven so much as cross-subsidised, and that commerce media is the label applied because commerce media carries a multiple that merchant-funded rebate networks do not.

Set against that, the case for Citi is a case about cost rather than about the category. The purchase price is undisclosed and immaterial, which caps the downside. Citi's rewards expense is a large line, and even a partial shift of it onto merchant marketing budgets registers in the card P&L while interchange compresses. Buying an operating platform with live supply and a team that has done the hard part beats a three-year build. All of that is true, and all of it would be equally true of an option on a business model nobody has yet demonstrated works unsupported.

There is a timing problem underneath it that the announcement does not touch. Card-linked offers are a human-attention product. They require a person to open a banking app, browse, activate an offer, and remember it at the purchase occasion. That asset is being acquired at the moment merchant selection begins moving out of human attention altogether. Where an agent executes the purchase, the inventory that matters is the inventory the agent can see, and a rebate sitting in a cardmember's app is invisible to it.

The strongest response, and it is a real one, is that closed-loop attribution to actual cross-merchant spend is something retail media cannot do off its own property, and that offer discovery inside a banking app does sometimes create an occasion rather than follow one. Both are true. Neither has yet supported a standalone business at scale, which is the whole point.

None of this makes the deal a mistake. Immaterial price, capped downside, real optionality, and a team worth having. It makes the framing a mistake. Citi has not bought a media property. It has bought a rebate distribution network with good technology, at a moment when the category's economics remain unproven and its channel assumptions are being renegotiated by something else entirely. The interesting question is not whether Citi paid too much. It is whether anyone in this category has yet shown that the thing works when nobody is subsidising it.