Earlier this summer, multiple merchants came to us with the same question: “Why am I suddenly seeing a spike in account_closed declines?” Without comparing notes or even talking to each other, these merchants all noticed the same concerning pattern for Visa transactions processed through JP Morgan Chase on cards issued by Citibank. They all turned to us to confirm if the decline code shift was unique to them, a result of Pagos mapping changes, or part of a broader industry shift.

Powered by billions of ingested transaction events, Pagos AI is uniquely equipped to answer this question and identify if this issue was widespread. We quickly noticed not only a consistent increase in declines labeled as account_closed, but also a simultaneous decrease in other decline codes.

We hadn’t changed our decline code mappings, so what did change?

Digging Into the Data

Using Pagos AI and internal diagnostics, we reviewed decline patterns from merchants processing Visa transactions through hundreds of acquirer-issuer pairings, including Chase (PSP) and Citibank (issuer).

Here’s what we found for the Chase-Citibank pairing:

• The share of account_closed declines jumped significantly in July, 2025 across the board

• That increase often corresponded with a slight drop in invalid_account_no_number and not_sufficient_funds declines

• The trend was specific to Citibank, and didn’t appear for any other issuers processed through Chase

• We also saw an increase in refer_to_issuer_1 codes from Citibank, starting in April. That change hasn't hit every merchant in equal measure, but it's following a similar pattern.

Learning Together

Had this shift only appeared for one merchant, it could indicate a new fraud trend or a customer retention issue. By establishing it as an industry-wide change for a specific PSP-issuer combination, however, we can be confident it’s neither. It’s simply a change in how Chase represents declines of Citibank cards.

When a PSP starts reclassifying some declines, that has ripple effects! It changes how you evaluate retry logic and affects how you measure cardholder behavior. It could even impact how you report on business performance. All this to say, you need to know when changes like this occur so you can respond appropriately and update your system to the new status-quo.

With Pagos in their back pocket, Merchants A-D in the visual above didn’t have to waste time and effort trying to figure out what may have happened on their end to cause the decline distribution shift. In a worst case scenario, they may have even drawn the wrong conclusions about their customers or products, and made unnecessary changes.

Side Lessons Learned

Through this analysis, we observed a couple interesting truths about the larger payments industry:

The Unique Relationship Between Merchants and the Payments System

Our analysis sheds light on an often-overlooked complexity of payments: every merchant has their own unique relationship with the broader payments ecosystem. Generalized advice on payments optimization often lacks validation from a view across many merchants and PSPs. 

In our analysis above, there are a lot of consistent variables: same acquirer, issuer, and card brand for transactions processed by large, US-based subscription merchants. Despite this, Merchants A-D saw meaningfully different decline code distributions. That variation becomes even more pronounced when we look at other networks or different acquirer-issuer pairings.

So why don’t we see any strong uniformity across merchants? There are several possible reasons for this:

• Merchants use different Merchant Category Codes, which may trigger different rules from bank or PSP risk models.

• Each merchant has unique average order values, buyer profiles (e.g. demographics), and accepts a different mix of card types.

• Large merchants may have negotiated custom agreements with issuers to optimize decline code usage.

• Issuers may have perceived friction with some merchants that they take into account when interpreting decline codes to manage their risk.

NSF Declines Aren’t as Unambiguous as You May Think

Decline codes have a reputation of being a bit fuzzy. Many hint at an issue, but don’t necessarily point directly to what went wrong. Not_sufficient_funds is considered to be one solid exception. For many merchants, it's one they can always trust to mean exactly what it says: the customer doesn’t have enough money to cover the purchase at this time. As such, they’ll often design logic to retry these declines at regular intervals in hopes the transaction eventually goes through when the account contains more funds (e.g. after payday). 

The apparent reclassification of some declines away from not_sufficient_funds and to account_closed (or other codes) indicates a few things:

1. Chase is improving their process of identifying decline root causes, thus improving the effectiveness of your targeted retry strategy

2. You have to take all decline codes with a grain of salt—a given code doesn’t guarantee 100% that the processor knows exactly what caused the decline; they’re learning and adapting with the market just like merchants are.

3. Issuers are also not always up to date on how they’ve mapped their internal decline decisions to the latest network specifications. This can (and does) result in many generic declines, including the infamous 05 generic decline.

A New Kind of Benchmark

We're building Pagos AI’s capabilities to help you detect issuer-driven changes before they become noise in your support queue or false signals in your analytics. We call this the Pagos network effect: using aggregated data (anonymized and responsibly shared) to spot industry-wide trends faster and give you the confidence to respond with clarity. 

Pair that with our benchmarking tool for comparing your payments performance to that of your industry peers, and you get an unprecedented view into the health of your payments stack!

Contact Pagos today to see how we can help your business optimize payments with benchmark data and industry insights.