Skip to content
← All insights

Attrition in Merchant Portfolios

A 20 percent annual volume leak means you board just to stand still. Underwrite for stay, not just for split.

Attrition is the share of a merchant book that stops producing residual in a period: closed doors, switched processors, thinner volume, or a re-priced margin. It is not last month's split, and it is not "how many deals I boarded." A book that leaks 20 percent of volume a year (illustrative, not an industry average) has to board that much net-new just to stand still. Underwrite for a fat split on accounts that leave and you fund a treadmill. Underwrite for stay and the split you already have compounds. This is the leak named in how ISO residuals work.

Three numbers people mix up

People say "attrition" and mean three different drops. Write down which one you are using.

MID count. How many merchant IDs processed last year and do not process now. Clean, and incomplete. A $8,000 residual whale and a $40 residual side account both count as one.

Volume. Card sales of the starting cohort versus the same cohort a year later. A MID that stays open and falls from $80,000 a month to $40,000 is volume attrition with no closed account. Compare like months on seasonal files.

Residual dollars. The check. Volume can hold while mix, buy-rate, or excluded lines eat the pool. That is the same reason a 70 percent split on a fat buy rate can look stable in headcount and still shrink the deposit.

Net "book growth" after new boarding is not retention. It is production minus the leak. Always compute a continuing cohort first: the MIDs that were already paying you at the start of the window. Then, separately, add what you boarded. Two curves. The gap between "if the book froze" and "what actually paid" is the work.

Worked example: 20 percent volume leak (illustrative)

All figures in this section are illustrative. They are not a book average or a quote.

Start of year:

  • 40 merchants at $50,000 a month each
  • Book volume: $2,000,000 a month
  • Net margin after cost: 40 BPS ($8,000 pool)
  • Your split: 60 percent
  • Monthly residual: $4,800 ($57,600 annualized if the book froze)

Now leak 20 percent of volume over twelve months and board nothing. Volume falls to $1,600,000. The pool is $6,400. Your residual is $3,840. You lost $960 a month, $11,520 a year, with no change to the split and no change to the buy rate.

Replacement, just to print the same $4,800: $400,000 a month of similar-margin volume, about eight $50,000 merchants. Those eight do not grow you. They fill the hole. If the eight are new shops, they will leak too, so gross boarding has to exceed 20 percent for the net to land at flat.

Key figure
20%
Illustrative annual volume leak. On a $2 million monthly book at 40 BPS and a 60% split, that leak is $960 a month, $11,520 a year, with no new boarding.
Illustrative arithmetic, not an industry average. Confirm the continuing-MID cohort on your own residual report.

If the remaining book keeps leaking 20 percent a year and you never replace:

YearMonthly residualAnnualized
0 (frozen)$4,800$57,600
1$3,840$46,080
2$3,072$36,864
3$2,458$29,496

Three years, no new deals, same split: the check is about half. A recruiter who only quotes the split is showing you a replacement quota, not a career.

Volume leak is not always headcount leak. Two $200,000 accounts or eight $50,000 accounts are the same 20 percent hole. Measure dollars first, then count.

Closings are only one leak

Some attrition is the economy. In the first quarter of 2025, 284,000 private-sector establishments died, a 3.0 percent quarterly death rate (U.S. Bureau of Labor Statistics, Business Employment Dynamics, 29 Jul 2026). That is doors dark, not processors shopped. You cannot save a closed store with a cleaner statement.

The rest of the leak is on you, the ISO, or the file you sold:

  1. Death. The merchant ceases. BLS is the backdrop, not your scorecard. High-churn verticals and brand-new shops sit closer to this bucket than a ten-year retailer on the same street.
  2. Switch. The merchant is still taking cards, somewhere else. Rate shock after a teaser, funding delays, a broken terminal, a landlord's POS, or a software suite that bundles processing will move a live file. This is the leak an honest model and a reachable help desk actually touch.
  3. Shrink. Same MID, less volume. Lost a location, lost a season, moved cash-heavy, or the card mix did not change and the sales did. Count it. Do not wait for a closed date.
  4. Reprice. Same volume, thinner pool. Mix shifts toward rewards credit, the desk raises the buy, monthly extras get excluded from the split, or you "saved" the account by cutting markup. Headcount looks loyal. The residual is not.

A fifth case is contractual, not merchant-side: the residual was never vested, a production minimum zeroed the stream, or a clawback ate a closed MID. That is not attrition of the merchant. That is attrition of your claim. Read the events that stop payment before you model the ones that start it.

How to measure it on your book

  1. Export every MID that paid you in month 0: volume, residual dollars, status, open date, last processed date.
  2. Freeze that list. That is the cohort. Write the month-0 totals at the top of the sheet.
  3. Twelve months later, classify each row: still processing, closed, or open with zero residual.
  4. Sum volume of the still-processing rows and compare to their month-0 volume. That is shrink plus death plus switch, in dollars.
  5. Sum residual dollars of the still-processing rows and compare to their month-0 residual. The extra gap versus volume is mix and reprice.
  6. Keep new MIDs on a second sheet. They are production. They are not retention.

Ask the residual desk for closed-MID reason codes if they have them: merchant requested, inactive, risk, equipment return, ISO transfer. A stack of "inactive" with no outreach log is a service failure dressed as a natural death. If the desk will not give you a continuing-cohort view, you cannot diligence the book, and a buyer will not either.

Underwrite for stay, not for the split

The boarding question is not "will this file approve." It is "will this file still be here in month 18, on a model that still fits."

Age and proof of business. A new shop can be a good merchant. It is a worse residual bet than a three-year operator with statements you can read. Closings cluster in young establishments. That is directionally a BLS survival fact, not a license to invent a percentage for your vertical.

Model fit. Interchange-plus, tiered, flat-rate, surcharge, dual pricing, and interchange optimization are not moral grades. They are fits. A debit-heavy store sold a fat blended teaser will shop the first time they see the real mix. Walk the pricing-model choice as advice, not as a template. Accounts on a model that matches ticket size and card mix leak slower because the statement does not feel like a bait.

Switching cost. Software-connected merchants, true MIDs with their own gateway, and equipment the merchant owns are stickier than a standalone terminal dropped on a counter. A bundled software pitch can also trap you if the partner owns the merchant relationship. Know who the merchant calls when something breaks.

Concentration. One account at 20 percent of residual is the 20 percent leak, waiting. Diversify by merchant, by vertical, and by ticket. A "great book" with three files is a job, not a portfolio.

The sale you made. If you won on a qualified rate the merchant will never see, you scheduled the switch. If you won on as-billed effective rate and a model they can explain to a partner, you scheduled a renewal. Stay is cheaper than replacement. Replacement consumes the same hours as growth and produces a flat line.

What a buyer does with the number

A residual book trades on a multiple of monthly residual, then gets haircut for leak, concentration, and whether the rights actually transfer. How to value a residual book is that math. The input the multiple is most sensitive to is not the headline split. It is whether next year's $4,800 is still $4,800.

Buyers pay up for a continuing-cohort rate on a spreadsheet, with closed MIDs listed, not narrated. They discount a recruiting story. The file that pays you reliably is the file you can sell.

What to demand, in one breath

Get a residual report that lists closed MIDs, last-processed dates, and the basis the split is calculated on. Compute continuing-cohort volume and residual yourself. Do not accept a recruiter's "our attrition is low" without the cohort, the window, and whether new boarding is inside the percentage. Then read vesting, portability, and successor protection so the merchants who do stay are actually yours.

Chase the split and you hire a replacement engine. Chase stay, on a disclosed buy, on residuals that vest, and the 20 percent treadmill is a number you run, not a number that runs you.

FAQ

What is attrition in a merchant portfolio? Attrition is the share of a book that stops producing residual in a period. Merchants close, switch processors, shrink volume, or get re-priced onto a thinner margin. Count it on a frozen starting cohort. Do not mix new boarding into the retention percentage.

Is 20 percent the industry average attrition rate? No. The 20 percent figure in this article is illustrative arithmetic, not a published ISO average. Measure your own continuing MIDs for volume and residual dollars. Recruiter "book averages" are not a rate until they name the cohort, the window, and whether new accounts are in the denominator.

How do I calculate attrition on my residual book? Freeze every MID that paid you in month 0. Twelve months later, split that same list into still processing, closed, and zero residual. Compare their volume and residual dollars to month 0. The drop is attrition. Anything boarded after month 0 is production, not retention.

Why can volume attrition differ from residual-dollar attrition? Volume tracks card sales. Residual tracks the pool after buy rate, mix, and excluded lines. A book can hold volume and still shrink the check if mix shifts to rewards credit or the desk raises the buy. That is why you compute both, and why the split headline is not the offer.

Are business closings the same as processor switches? No. Closings are an economy-wide leak. In the first quarter of 2025, 284,000 private-sector establishments died, a 3.0 percent quarterly death rate (BLS, 29 Jul 2026). You cannot save a dark store with a better statement. You can save a live merchant who is shopping the rate you sold them.

How does attrition change what a buyer will pay for my book? Buyers start from a multiple of monthly residual, then haircut for leak, concentration, and contract quality. A high-split book that needs constant replacement is a treadmill, not an asset. Durable rights plus a measured continuing-cohort rate are what survive diligence.

Sources

Questions about how this applies to your business?

Talk it through

Your next move

Clarity looks good
on your business.

Analyze a statement