Failed first payments — the week of 2026-09-07 to 2026-09-13
Every agreement whose opening instalment failed in the last seven complete days, the reason the
issuer gave Square, and whether the debt-collection GHL then actually texted them. 138 agreements took
their first charge in this window; 60 of them failed — 43.5%.
The per-person list is a key-gated CSV; this page carries counts only.
Failed first payments
60
of 138 agreements opened · 43.5% failed
Not collected
$17,054
$17053.91 of opening instalments
In the debt GHL
48 80.0%
12 are not there at all
Texted after the failure
45 75.0%
15 were never texted
Texted us back
21
46.7% of those texted · 35.0% of all 60
Money in since
15 25.0%
$7,381 banked — read the caveat on the reply tab
By day
| Day | Failed openings | Missed | Texted | Replied | Money in since |
|---|
| Mon, 2026-09-07 | 11 | $4206.73 | 9 81.8% | 6 66.7% | 2 18.2% |
| Tue, 2026-09-08 | 4 | $891.65 | 3 75.0% | 0 0.0% | 1 25.0% |
| Wed, 2026-09-09 | 7 | $2686.66 | 6 85.7% | 3 50.0% | 1 14.3% |
| Thu, 2026-09-10 | 6 | $1235.83 | 5 83.3% | 5 100.0% | 3 50.0% |
| Fri, 2026-09-11 | 21 | $6035.14 | 18 85.7% | 6 33.3% | 6 28.6% |
| Sat, 2026-09-12 | 7 | $1297.91 | 0 0.0% | — | 1 14.3% |
| Sun, 2026-09-13 | 4 | $699.99 | 4 100.0% | 1 25.0% | 1 25.0% |
By city
| City | Failed openings | Missed | In debt GHL | Texted | Replied |
|---|
| Texas | 12 | $2244.65 | 10 83.3% | 9 75.0% | 4 44.4% |
| Boston | 6 | $1049.66 | 5 83.3% | 5 83.3% | 4 80.0% |
| New York | 15 | $3423.72 | 14 93.3% | 14 93.3% | 6 42.9% |
| Florida | 20 | $7217.89 | 13 65.0% | 11 55.0% | 4 36.4% |
| Chicago | 2 | $2229.67 | 1 50.0% | 1 50.0% | 0 0.0% |
| Nashville | 5 | $888.32 | 5 100.0% | 5 100.0% | 3 60.0% |
Why the count jumps around the week. Charge dates cluster — the 1st and the 15th are the
two biggest days on the recurring book, so a week containing one of them carries a spike that is a calendar
effect, not a deterioration. Compare like-for-like weeks, not adjacent ones.
What the issuer actually said
Straight off the Square payment record — card_details.errors on the failed opening charge.
One row per code, with what it means on this book rather than what the Square docs call it.
| Reason | Failures | Missed | Money in since | Texted | Replied | What it means here |
|---|
Transaction limit
TRANSACTION_LIMIT | 33 55.0% | $8548.05 | 8 24.2% | 27 81.8% | 12 44.4% | The issuer refused it against a limit on the account. On this book it behaves as a MONEY code, not a technical one — it recovers at the same rate as INSUFFICIENT_FUNDS, and of the customers who answered our text and picked an option, almost all picked "money is a bit tight". |
Generic decline
GENERIC_DECLINE | 14 23.3% | $3028.72 | 3 21.4% | 10 71.4% | 5 50.0% | The issuer refused and told us nothing. No script follows from this code on its own — the text is the only way to find out why. |
Insufficient funds
INSUFFICIENT_FUNDS | 6 10.0% | $3137.49 | 2 33.3% | 4 66.7% | 3 75.0% | Not enough money in the account on the day. |
Cardholder permissions
CARDHOLDER_INSUFFICIENT_PERMISSIONS | 4 6.7% | $806.33 | 1 25.0% | 3 75.0% | 0 0.0% | The issuer has fenced this card off from recurring card-not-present charges. ONLY THE CARDHOLDER CAN LIFT IT, usually a toggle in their banking app. This is the one code on the list where "check your bank app" is the correct script — and it is heavily concentrated in the early-access neobanks. |
Issuer instalment error
ISSUER_INSTALLMENT_ERROR | 1 1.7% | $241.66 | 0 0.0% | 0 0.0% | — | The issuer rejected the charge specifically as an INSTALMENT. It is not a balance problem and not a dead card — the bank would not take this card on a payment plan. The card has to be replaced with one that will; asking them to "try again" with the same card does nothing. |
CVV failure
CVV_FAILURE | 1 1.7% | $1150.00 | 1 100.0% | 0 0.0% | — | The security code did not match. |
Card expired
CARD_EXPIRED | 1 1.7% | $141.66 | 0 0.0% | 1 100.0% | 1 100.0% | The stored card is past its expiry date. Needs new card details, nothing else. |
The two codes worth separating. Cardholder permissions is the only one on this list a
customer can fix in thirty seconds — the issuer has fenced the card off from recurring charges and a toggle in
their banking app lifts it. Nobody is told that, because the decline reason never reaches GHL at all: the
outbound text is identical whatever the issuer said. Transaction limit is the opposite trap — it reads like
a technical block and behaves like no money.
By card brand
| Brand | Failures | Missed | Money in since |
|---|
| VISA | 40 66.7% | $11067.19 | 9 22.5% |
| MASTERCARD | 16 26.7% | $5136.73 | 4 25.0% |
| DISCOVER | 3 5.0% | $562.49 | 2 66.7% |
| AMERICAN_EXPRESS | 1 1.7% | $287.50 | 0 0.0% |
Still not paid
Of the 60 agreements whose opening instalment failed last week,
45 have had no money banked since — $9672.68. "Since" means any completed charge
on that customer after the failed opening, not just the opening amount. The per-person list is the key-gated
CSV: /first-miss-week-unpaid.csv.
Still owing
45
of 60 failed openings · 75.0%
Still not collected
$9,673
$9672.68 of opening instalments
Never asked at all
7 15.6%
no text went out after the failure
Asked and silent
22 48.9%
texted, never answered
Answered, still owing
16 35.6%
engaged with us and no money followed
Who could still be worked
| Where they stand | People | Owing | What it means |
|---|
| Never reached the GHL | 5 11.1% | $852.08 | Nobody can chase them — they are not in the sub-account. Getting them in is the fix. |
| In the GHL, never texted | 2 4.4% | $462.49 | The contact exists and nothing went out. Check whether the feed dropped them or the workflow did. |
| Texted, no reply | 22 48.9% | $4866.06 | Asked once and silent. This is the calling list — a text has already failed on them. |
| Texted, replied, still no money | 16 35.6% | $3492.05 | They ENGAGED and still have not paid. Warmest of the four and the one nobody is working. |
The four groups need four different actions, and only one of them is a calling list.
The people who replied and still have not paid are the warmest names on the page — they answered a
text about a missed payment and then nothing happened. The people who were never asked at all are not a
performance problem, they are a plumbing problem: nobody has spoken to them yet.
How long they have been owing
| Days since the charge failed | People | Owing |
|---|
| 3 days or less | 3 6.7% | $654.16 |
| 4–5 days | 21 46.7% | $4174.72 |
| 6–7 days | 9 20.0% | $2101.66 |
| 8 days or more | 12 26.7% | $2742.14 |
By reason
| Square's reason | Still owing | Owing | Share of that reason | Texted |
|---|
| Transaction limit | 25 55.6% | $4973.89 | 25 75.8% | 23 92.0% |
| Generic decline | 11 24.4% | $2564.15 | 11 78.6% | 8 72.7% |
| Insufficient funds | 4 8.9% | $1057.49 | 4 66.7% | 3 75.0% |
| Cardholder permissions | 3 6.7% | $693.83 | 3 75.0% | 3 100.0% |
| Issuer instalment error | 1 2.2% | $241.66 | 1 100.0% | 0 0.0% |
| Card expired | 1 2.2% | $141.66 | 1 100.0% | 1 100.0% |
By city
| City | Still owing | Owing | Share of that city | Texted |
|---|
| Texas | 8 17.8% | $1468.82 | 8 66.7% | 6 75.0% |
| Boston | 6 13.3% | $1049.66 | 6 100.0% | 5 83.3% |
| New York | 12 26.7% | $2959.15 | 12 80.0% | 12 100.0% |
| Florida | 15 33.3% | $3297.89 | 15 75.0% | 11 73.3% |
| Chicago | 1 2.2% | $279.67 | 1 50.0% | 1 100.0% |
| Nashville | 3 6.7% | $617.49 | 3 60.0% | 3 100.0% |
This list will shrink on its own, and not because anything is automated. Money still
arrives on these accounts — a collector gets hold of someone, or a customer acts on a text. Nothing on this
book charges a card a second time by itself, so every name that drops off did so because a person made it happen.
Re-run the report to see the current position rather than working from a saved copy.
Did we text them, and did they answer
Cross-checked live against the US Debt Collection GHL sub-account
(pcTyOadbzBqZO7v3cbwn) — matched on phone number, then email, then name. Read-only: this page sends
nothing and enrols nobody.
Failed first payments
60
the whole week
Reached the debt GHL
48 80.0%
12 never arrived
A text actually went out
45 75.0%
3 are in the GHL and were still not texted
Replied
21
46.7% of those texted
Fastest text out
23.2h
median 24.4h · slowest 72.9h
Undelivered texts
2
of 130 sent · 98 by the workflow, 32 typed by a person
How long after the failure the first text went out
| Time from failure to first text | People |
|---|
| Same day (under 12h) | 0 0.0% |
| 12–24h | 6 13.3% |
| 1–2 days | 21 46.7% |
| 2–4 days | 18 40.0% |
| 4 days+ | 0 0.0% |
Nobody is contacted on the day. The fastest text in this week left
23.2 hours after the charge failed and the median was
24.4 hours — the feed into this sub-account is a daily
Google-Sheet drop, so day 0 does not exist. That is the expensive part: measured over four months,
day 1 is 48% of everything a missed opening ever returns inside ten days, and the first text lands at
the end of it.
What they said back
The workflow's opening text asks for a reason and offers three numbered options:
1 card issue · 2 money is a bit tight · 3 something else. This is the only part of a reply this report
reads, and no message body is ever stored.
| Reply | People |
|---|
| 1 — Card issue | 2 9.5% |
| 2 — Money is a bit tight | 7 33.3% |
| 3 — Something else | 2 9.5% |
| Wrote something instead of picking | 10 47.6% |
The customers agree with the money reading of TRANSACTION_LIMIT. Of the people who
picked an option, the large majority picked "money is a bit tight", and nearly all of those had a
TRANSACTION_LIMIT decline — not a card-issue one. 2 said card issue.
Whatever the code looks like, the people answering are telling us it is affordability.
Do not read the recovery column as an effect of texting. 15 of the
60 have money in since, but the people who pay straight back are exactly the people who never needed
chasing — and some of them are out of the texting list precisely because they had already paid. The comparison is
selected at both ends. Nothing on this book charges a card a second time on its own; every later payment is
a person acting — a collector on the phone, or the customer following up a text.
Where the chase leaks
Two separate holes, and they need different fixes. One group never reaches the debt sub-account
at all; the other is sitting in it with no text sent.
Never reached the GHL
12
$6,095 · no phone-number match in the sub-account
In the GHL, no text sent
3
$1,612 · contact exists, nothing went out
Total not contacted
15 25.0%
$7,707 of missed openings with no text
Emails sent
0
SMS is the only channel this sub-account uses
Calls logged in GHL
0
the team dials from GoTo, which does not write back here — absence is not proof nobody rang
The gap, by day
| Day | Failed openings | Never reached GHL | In GHL, not texted | Missed, uncontacted |
|---|
| Mon, 2026-09-07 | 11 | 2 18.2% | 0 | $2247.91 |
| Tue, 2026-09-08 | 4 | 0 | 1 25.0% | $241.66 |
| Wed, 2026-09-09 | 7 | 0 | 1 14.3% | $1150.00 |
| Thu, 2026-09-10 | 6 | 1 16.7% | 0 | $320.00 |
| Fri, 2026-09-11 | 21 | 3 14.3% | 0 | $2450.00 |
| Sat, 2026-09-12 | 7 | 6 85.7% | 1 14.3% | $1297.91 |
| Sun, 2026-09-13 | 4 | 0 | 0 | $0.00 |
Never reached the GHL, by city
| City | Failed openings | Never reached GHL | Missed |
|---|
| Texas | 12 | 2 16.7% | $316.67 |
| Boston | 6 | 1 16.7% | $125.00 |
| New York | 15 | 1 6.7% | $297.91 |
| Florida | 20 | 7 35.0% | $3405.41 |
| Chicago | 2 | 1 50.0% | $1950.00 |
| Nashville | 5 | 0 | $0.00 |
Checked, not assumed. Every one of the 12 was probed on the phone
number, the email address and the name. The phone probe returned zero contacts for each of them — they
are genuinely absent from the sub-account, not a formatting miss.
In the GHL but not texted — when the contact was created
| Contact created | People |
|---|
| 2025-11-18 | 1 |
| 2026-06-15 | 2 |
A contact created today has not had time to be texted — the feed is a daily drop, so the tail of
this table is lag rather than a miss. Anything created before yesterday and still untexted is a real gap.
How every number here is made
The population
A failed first payment = a Square customer whose first-ever charge attempt in the
recurring book lands between 2026-09-07 and 2026-09-13 and did not complete. The attempt history searched runs from
June 2023 to now across both attempt files, so a customer with any earlier charge anywhere in it is not
counted as an opening. Only the six Recurring* locations are in scope — deposits, studio sales and office
takings are excluded, because they are not instalments.
"Last week" is the last complete MONDAY-to-SUNDAY week — here 2026-09-07 to 2026-09-13 —
not the last seven days. A rolling seven-day window always contains part of the week we are currently in, which
is a different question and makes the name wrong. The current week is never included, and today is a part-day
and would read as a collapse. Run with --from / --to for any other period.
"Not paid" means no completed charge of any amount on that customer after the failed
opening — not merely that the opening amount is outstanding. Someone who paid a different figure, or paid
part of it, counts as money in and is off the unpaid list.
The decline reason is the issuer's own, read off card_details.errors on the
failed payment. Where Square recorded no reason the row shows "no reason given" rather than being assigned one.
The cross-check
Matched on phone first, then email, then exact name, against
pcTyOadbzBqZO7v3cbwn — the live US Debt Collection sub-account. Not the location in
ghl_config.json, which is a near-empty account that 403s with this token.
"Texted" means at least one outbound SMS dated on or after the moment the charge
failed, on any conversation belonging to that contact. An older text from a previous cycle does not count.
"Replied" is an inbound SMS after the same moment.
No message body is stored. The only thing read out of a reply is whether it starts with
1, 2 or 3 — the options the workflow's own opening text offers.
What this page cannot tell you
Phone calls. The debt team dials through GoTo and those calls do not write back into
GHL, so a zero in the calls column is a limit of the source, not evidence that nobody rang.
Cause and effect. Whether a person was texted is not random — the ones who pay
immediately drop out of the chase, so any comparison of "texted" against "not texted" is selected at both ends
and cannot be read as the effect of texting.
Nothing here is automated recovery. No system on this book charges a card a second time
on its own. Every later payment is a person acting.
Sources
Reports/finance_cohort/square_attempts.jsonl + square_attempts_pre2025.jsonl — the charge
attempts, refreshed by scripts/reports/pull_square_declines_history.js.
Square /v2/customers/{id} — name, phone, email, reference id.
GHL /contacts/, /conversations/search, /conversations/{id}/messages — all GET.
Built by scripts/reports/report_first_miss_week.js at 2026-09-17T11:26:24.183Z.