Citi's Commerce Media Ads Rely on Card Data It May Not Control

Citi Commerce Media sells brands access to US cardholders and measures results against their spending. Thin per-customer data and undisclosed partner terms make the usable audience narrower than the launch suggests.

Citi launched Citi Commerce Media inside its US Consumer Cards business on September 23, formalizing an advertising capability it had been piloting. The bank says the business, which spans general-purpose and private-label cards and installment lending, serves more than 70 million US customers and sees 6.5 billion transactions a year across more than 700 spending categories. Brands can buy placements on Citi.com, the Citi Mobile app and paid-media properties off Citi's own sites, then measure whether the customers who saw an ad later spent with them. At about 12:20pm in New York, Citi was flat at $132.56 and Cardlytics, the listed intermediary in the same market, was down 1.1% at $3.54, little changed.

The launch follows Citi's August agreement to buy Kard, a commerce-media and rewards platform that runs personalized offers on bank transaction data. Citi called the deal immaterial and did not disclose terms. Until it closes, the two companies operate separately, so today's product does not include Kard's merchant network.

What 93 transactions a year says about the base

Dividing 6.5 billion transactions by 70 million customers gives about 93 purchases per customer per year, fewer than two a week. A household whose main card is a Citi card typically puts groceries, fuel, subscriptions and restaurants on it and swipes several times a week. An average below two a week means that for a large part of the base, Citi is a secondary card, a travel card, or a store card used at one retailer.

That shapes what advertisers can buy. For customers who use Citi as their main card, the bank sees a broad cross-merchant picture: where they shop, how often, and when they switch. For the rest, it sees a thin slice. Citi's US card spend was $176 billion in the second quarter, up 11% on the year, which works out to very roughly $100 per transaction if the quarter is representative. But the view of any individual shopper depends on how much of their wallet Citi holds.

The partner problem

The bigger caveat is who controls the data. Citi's own description of the business includes private-label cards, and its Retail Services unit issues store cards for retailers such as Home Depot and Best Buy. It also runs large co-brand programs, including the Costco Anywhere Visa, and on April 24 it became the sole US issuer of American Airlines' AAdvantage cards, taking over the portfolio Barclays used to issue.

Several of those partners sell advertising against their own customers. Home Depot runs a retail media network, Orange Apron Media. Best Buy sells advertising through Best Buy Ads across more than 1,000 stores and its website. A retailer with its own media business has an obvious commercial interest in controlling how the bank that issues its store card monetizes the same shoppers' purchases. Partner economics already weigh on the segment: Citi said higher partner payment accruals were one reason US Consumer Cards' non-interest revenue fell 47% in the second quarter.

Citi has not said which portfolios feed Commerce Media or how partner data is treated, and the contract terms are not public. If private-label and co-brand spending is excluded or restricted, the monetizable dataset is mainly the Citi-branded portfolio, a large business but not all of the 6.5 billion transactions. An advertiser buying "cross-merchant" audiences should ask this first.

CTOL view: Partner access will decide the size of this business, and we expect at least one partner to be brought in openly. By December 31, 2027, Citi or a partner will publicly name at least one Retail Services or co-brand partner, such as Home Depot, Best Buy, Costco or American Airlines, whose cardholders' transaction data is used in Citi Commerce Media. Our reasoning is that the branded portfolio alone is too thin per customer to compete with retail media networks, and the obvious deal is a revenue share with a partner that already sells ads. Readers can check Citi's and the partners' press releases and Citi's earnings materials. We are wrong if no such partner is publicly named by December 31, 2027.

Chase set the category; Cardlytics shows the risks

JPMorgan's Chase Media Solutions is the obvious comparison. In its 2025 annual report, JPMorgan said the business reached nearly 74 million customers, up 14% on the year, and "$11 billion in consumer spend," double the 2021 level. The letter does not break that figure down or say how much of it advertising caused, so it cannot be used as a proxy for revenue or compared with Citi's transaction count. What it does show is that a bank-owned media business can grow reach steadily without disclosing revenue.

Cardlytics shows how the model works when a middleman sits between the bank and the advertiser. It sells marketers access to customers of partner banks and pays those banks a negotiated share of billings. In the second quarter of 2026, revenue fell 36% to $36.9 million, billings were $65.5 million and the net loss was $14.9 million, although adjusted EBITDA was positive at $1.7 million.

What happened to its bank partners matters more. Bank of America gave notice in April 2025 that it would not renew, and the relationship ended in February 2026. Cardlytics' top three bank partners accounted for more than 80% of partner payments in 2025, and its agreements are generally terminable on 90 days' notice or more. The large issuers have concluded that they can keep more of the economics by running commerce media themselves. Citi's launch fits that pattern.

How big it can get

Citi's one disclosed result is a campaign for an online retailer that produced "up to 5x incremental return on ad spend." Without the campaign size, control-group design, attribution window, or whether the return is measured on sales or margin, the result is a case study, not evidence of platform economics.

Revenue is the wrong yardstick for scale; profit is the right one. US Consumer Cards produced $4.5 billion of revenue and $852 million of net income in the second quarter, about $18 billion and $3.4 billion annualized. A commerce-media business earning $500 million a year, which would be a strong result, would add under 3% to revenue. But the data and digital surfaces already exist, so most of that revenue would fall to profit. At an assumed 50% after-tax margin, a CTOL estimate, $500 million of revenue would add about $250 million of net income, roughly 7% of the segment's annualized earnings. That is enough to matter to a unit whose expenses rose 10% in the quarter, and it explains why Citi is building rather than renting.

Citi reports third-quarter results on October 13 and is unlikely to break out the business. The disclosures to watch are advertiser count, repeat spend, and whether any partner has agreed to let its cardholders' data into the product.

Citi's Commerce Media Ads Rely on Card Data It May Not Control

What 93 transactions a year says about the base

The partner problem

Chase set the category; Cardlytics shows the risks

How big it can get

Sources

You May Also Like

Subscribe to our Newsletter