Portfolio Health: Returns, Chargebacks and Retrievals

Agents are trained to watch volume, and volume is the metric that tells you least about whether your book is in trouble. A portfolio can grow every month while quietly accumulating the three ratios that decide whether it keeps its pricing, keeps its processing, and keeps existing. Those ratios are chargebacks, ACH returns and retrievals, and they are worth more of your attention than the residual statement is.

The chargeback ratio, and which one is being measured

The card networks monitor disputes as a ratio, and the first thing to establish is which ratio your program and the networks are actually using, because there are two and they differ materially. Count-based is disputes divided by transactions; volume-based is disputed dollars divided by processed dollars. A merchant with a handful of very large disputed tickets can look fine on one and alarming on the other.

Both matter, and the threshold that triggers a monitoring program is lower than most agents assume, well under one percent. What makes this dangerous for a portfolio rather than for a single merchant is that the monitoring is per-MID, but the reputational and operational consequences roll uphill: an acquirer watching a portfolio with several merchants in monitoring programs starts pricing that portfolio differently, and the merchants who did nothing wrong pay for it.

ACH returns are the ratio agents forget

Bank payments are cheaper to accept and the return economics are completely different from cards, which is exactly why they catch people out. NACHA monitors return rates by category, and the administrative and unauthorised categories carry far lower tolerances than the ordinary insufficient-funds return does.

The practical distinction: a customer who does not have the money today is a normal business problem and returns under a code that is monitored loosely. A customer who says they never authorised the debit is a different category with a much tighter ceiling, and a merchant who is casual about authorisation records will breach it long before their card book shows any stress at all. If you have merchants billing recurring ACH, the authorisation trail is a portfolio risk, not a paperwork detail.

Retrievals are the early warning nobody reads

A retrieval, an issuer asking for documentation on a transaction without disputing it, is the cheapest signal in the whole system and the most ignored. It costs the merchant almost nothing and it usually means a cardholder did not recognise the charge on their statement.

That is a fixable problem and it is fixable before it becomes a dispute. Most of the time the cause is the descriptor: a merchant trading as one name and billing under another, or a recurring plan whose descriptor names a product the customer has forgotten subscribing to. A rising retrieval count with a flat dispute count is a merchant who is about to have a dispute problem and does not know it, and an agent who calls them about it is doing the single highest-value thing available.

Look at the distribution, not the average

A portfolio-level dispute ratio is nearly useless as a management number because it is dominated by whichever merchant processes most. What you want is the per-merchant distribution and the outliers.

• Rank merchants by dispute ratio, not by dispute count. A small merchant with six disputes on two hundred transactions is a much bigger problem than a large one with forty on forty thousand.
• Watch the trend on each, not the level. A merchant moving from a tenth of a percent to a third is a story even though both numbers look fine.
• Separate recurring from one-off. Subscription merchants accumulate disputes through failed renewals and forgotten sign-ups, and the fix is billing hygiene rather than fraud controls.
• Check whether one product or one channel is producing them. Disputes concentrated in a single SKU or a single payment link usually mean the description does not match what arrives.

When one merchant is dragging the rest

Eventually you will have an account whose ratios are a threat to the book. The instinct is to defend it, because it is revenue and because you wrote it. The arithmetic is usually against you: a merchant producing meaningful monthly residual is rarely worth the repricing of a portfolio, and almost never worth an acquirer's attention.

Before you get to that decision there is a sequence worth running, in order. Fix the descriptor. Check that refunds are being issued promptly, because a merchant who argues with customers for three weeks converts refund requests into chargebacks at a startling rate. Look at delivery timing, since a long gap between charge and fulfilment is the most reliable dispute generator there is. Then, and only then, talk about fraud tooling, because it is the expensive answer to a question the first three usually resolve.

Make the review a calendar item

The reason portfolio health deteriorates is not that agents do not care. It is that nothing prompts them to look. Disputes arrive as individual notifications and get handled individually, and the ratio is never computed by anybody until it is computed by a monitoring program.

A monthly pass over the ranked list, ten minutes, is enough to catch nearly all of this while it is still cheap. Review the outliers, call the two merchants whose trend is moving, and write down what you told them. The portfolio you keep is the one you looked at.