Share:

Boston Recurring Payments — Why They Fail

3 August 2026 · Studio 1 · Data pulled directly from Square (Boston Recurring, location L3TZS37RCY0QY)

The headline

Over a two-week window, more than half of every automatic recurring charge we attempt in Boston fails. This is not a collections effort problem. The money is being asked for and the card is saying no.

54%
of recurring charge attempts FAILED
(309 of 568, 20 Jul – 3 Aug)
41
missed payments in just
3 days (1–3 Aug)
$6,914
due across those
41 missed payments
$45,879
actually owed by those same
41 people once arrears count
The $6,914 figure is the trap. Chasing "what's due today" understates the problem by a factor of six. The people who missed a payment this week are, on average, already nine payments behind. The real exposure sitting in those 41 names is $45,879.

These are not people having a bad week

We checked every one of the 41 against their full invoice history in Square — every payment they have ever been asked for, and whether they paid it.

Who they actually arePeopleDue nowTotal owed
Chronic — 3 or more misses behind25$4,408.78$36,156.81
Never paid a single invoice, and long-running7$1,238.50$8,237.50
Brand new — first invoice ever, missed it4$774.15$774.15
Second miss2$217.80$435.60
Good payer, genuine first slip3$274.50$274.50
TOTAL41$6,913.73$45,878.56
Only 3 of 41 are good payers who slipped once. 27% (11 people) have never paid us a single invoice — not one, ever — and they owe $9,012 between them. A "friendly reminder, these things happen" message is the wrong tool for 38 of these 41 conversations.

Square tells us why the card said no

Every failed attempt comes back with a decline reason code from the issuing bank. That means we can sort people into the right conversation before anyone replies to us — we know whether it's an admin problem or a money problem on the day it fails.

Reason the card was declinedPeopleValueWhat it actually means
GENERIC_DECLINE11$1,709.57Bank won't say why. Genuine unknown.
TRANSACTION_LIMIT11$1,565.30Amount too high / credit limit reached. Money.
INVALID_ACCOUNT8$1,462.03Bank can't find the account. Card is dead or closed.
No attempt made at all7$1,391.83We hold no card for them. Nothing was ever charged.
PAN_FAILURE2$315.00Card number invalid.
CARDHOLDER_INSUFFICIENT_PERMISSIONS1$250.00Card restricted by the issuer.
INSUFFICIENT_FUNDS1$220.00No money in the account. Money.
TOTAL41$6,913.73
Not a single expired card in the batch. The instinct is that most failures are an out-of-date card and a 30-second fix. In these 41, CARD_EXPIRED appears zero times — and only 4 times in all 309 failures across the fortnight. Building the process around "your card expired" would address 0 of 41 cases.

The four conversations this creates

The decline code routes each person to a different opening message. Same debt, four completely different asks:

BucketPeopleThe opening message
MONEY12Limit reached or no funds. Skip the card question entirely — open with the bad-month framing and offer a smaller instalment.
DEAD CARD8The account is closed. They need to add a new card — not correct an expiry date. Low friction, no discount conversation needed.
UNKNOWN11Bank gave no reason. Use the either/or message — "is this a card problem or a timing problem?" — and let them self-sort.
NO CARD ON FILE7We never had a card. Never tell these people their payment was declined — it didn't happen, and they'll know. The ask is to set up payment for the first time.
ADMIN3Bad card number or a restricted card. Quick correction, no money conversation.
Why this is worth doing. 23 of the 41 (56%) get a message that never mentions discounts, hardship or payment plans — because their problem isn't money, it's a broken card or no card at all. Today all 41 get the same chase. Fixing that is free.

Two things to fix before we build on this

The existing daily missed-payments report can't support this process yet. Two defects, both found while pulling this data:

ProblemImpact
"Miss number" is not the miss count. The report reads the sequence number off the invoice (invoice 12 of a plan) and calls it "12th miss". Wrong on 29 of 41 rows — and wrong in both directions. One customer shown as a 9th miss was on her first; another shown as 12th was on her 15th. The entire escalation ladder is keyed off a number that is close to random.
The report only reads 6,000 invoices. Boston Recurring alone holds 32,343. It doesn't just drop rows — it corrupts each customer's history. At a partial read the arrears totalled $36,696 and 15 people looked like they'd never paid; at a full read it was $38,747 and 11.

Both are a single change to one script. Once fixed, the daily report can carry the true miss number and the decline reason side by side, and the collections team gets a routed call list every morning instead of a flat chase list.