Collections compliance monitoring that checks every call, voicemail, text and letter — not a small sample
Every conversation checked, not a small sample, each finding tied to its second.
See one case, screen by screen ↓Software that checks collectors’ conversations with borrowers against the rules that govern them — Regulation F, the FDCPA, state rules and your own policy — and records what it found and who decided. Collections Compliance Monitoring checks every call, voicemail, text, email and letter overnight, not a sample, and pins each finding to the second or sentence where it happened.
How to choose AI solutions for lending →
Updated
A collections quality team hears a few percent of calls and reads even fewer voicemails, texts, emails and letters — across its own collectors and every agency it places accounts with. The sample is pulled by hand and each call is heard end to end.
What the sample misses is what hurts: a collector who tells a borrower’s sister about the loan, a dialer that calls at 21:18 because it read the time zone from an old area code, a voicemail that names the debt on a work line, a letter with the wrong past-due amount, a borrower who just said they lost their job being asked for payment anyway. Those are found when a complaint, a lawsuit or an examiner finds them first — and an agency that is scored on recoveries every month and on compliance from a handful of calls looks the same as the best one.
Estimated split for one collections compliance analyst’s week, by hand and with the solution.
Six specialist agents check every conversation against the rules and pin each finding to its moment; the compliance analyst decides.
Monday 12 October at Northmere Bank: its own collectors and three agencies had 41,286 conversations with borrowers. Here is what happens next, screen by screen, in the working solution.
17,952 calls, 9,384 voicemails, 8,516 texts, 3,941 emails and 1,493 letters, checked against 41 rules. 1,214 carry a finding: 38 red, which a person confirms one by one; 412 amber, reviewed in weekly batches; 764 vulnerability signals, routed to a team. Red by team: Northmere’s own team 8, Summit Ridge 7, Pinegate 9, Calder & Voss 14.
“A 2% sample would have heard 826 conversations — about one of these 38.”
The night’s red findings sit on the left as mini waveforms, so Dana sees where in each call the problem is before she opens it. She opens C-1012-58431: Jordan Pike of Calder & Voss calling Marcus Bell about his 2019 Tacoma auto loan, $1,284.50 past due at 63 days, 18:52 local time in Phoenix. Five flags sit on the waveform. She clicks the red one, the playhead jumps to 02:41, and she hears it.
“It’s the Tacoma loan with Northmere. He’s two payments behind, twelve eighty-four fifty.” — said to the borrower’s sister, who identified herself at 00:14.
Twenty-three seconds later the collector says the truck will get picked up if it is not paid by Friday. The checker looked at the account before calling it false: no repossession assignment on ••4471, and the bank’s policy is a right-to-cure letter first at 75 days, with the loan at 63. It is flagged amber, with the rule cited, for Dana to add to the agency letter. A violet flag at 01:02 marks the sister’s words about cut hours and a hospital stay, with a button to refer the account to the hardship team.
The checker never decides a breach — it proposes, with the quote.
Beside the finding, the run is shown agent by agent. The transcriber turned 03:56 of audio into 32 turns from 2 speakers and redacted 1 card number. The disclosure checker found the recording notice at 00:00. The time checker placed the call at 18:52 in Phoenix, with address and area code agreeing, and counted it as the 4th call on this debt in 7 days. Any step can be run again.
Dana confirms the red finding. A coaching item for Jordan is created with the moment attached, Calder & Voss’s 30-day compliance score moves from 91.4 to 91.3, and the decision lands in the review log. Had she wanted to dismiss it, she would have needed a reason and a second signature from Hal Brennan, the compliance manager.
“Confirmed by Dana Whitlock · 09:12. Coaching item for Jordan with this moment · Calder & Voss Collections score 91.4 → 91.3 · in the review log.”
23,334 messages were read, each note pinned to the sentence it is about. A Pinegate voicemail for Grace Holm went to the Hollis Freight dispatch office, her work line on file, and asked her to call back “about your Northmere auto loan balance”. A limited-content message may not mention the debt — and coworkers can hear this one.
Reg F §1006.6(d) · FDCPA §805(b)
A Calder & Voss letter to Harold Briggs says $2,316.40 past due, including a $120.00 collection fee. Northmere’s system shows $2,196.40 past due on 9 Oct, and the retail installment contract allows a late charge only, not a collection fee. Dana can confirm the finding and ask for a corrected letter.
Amounts compared with Northmere’s system of record at 02:10.
Every attempt is placed on the borrower’s own clock — local time from the address on file, not the phone’s area code. Seven contacts fell outside 08:00–21:00, and five came from Pinegate’s campaign PG-West: borrowers who moved east kept California mobile numbers, so 18:18 Pacific on the dialer was 21:18 at Ellen Ortiz’s door in Richmond, Virginia. Six accounts went over 7 calls in 7 days.
The checker never dials or blocks calls itself — it alerts. A person holds PG-West after 20:00 Eastern.
Every confirmed finding becomes a coaching item with its moment attached. Jordan Pike has three: the 02:41 disclosure, the 03:04 repossession statement and a disclosure to Leah Sorensen’s roommate at 00:57. Agency collectors’ items go to the agency’s own trainer with the clips; Northmere’s team leads mark their own collectors coached.
Calder & Voss had 14 of the 38 red findings — 2.79 per 1,000 conversations. The remediation letter to Rhonda Vale is drafted: 4 third-party disclosures and 3 threats or false statements, led by call C-1012-58431, with the moments attached and a retraining plan asked for within 10 business days. Beth reads it and sends it; nothing goes to an agency without her.
The findings writer never sends to an agency without a person.
The same night’s conversations, seen by the five people who act on them — what their week looked like, and what it looks like now.
Six specialist agents check every conversation against the rules and pin each finding to its moment; the compliance analyst decides.
Transcribes every call and voicemail, separates the collector from the other party and time-stamps every word.
Checks the recording notice, the debt-collector disclosure, right-party verification, the limited-content voicemail script and the opt-out line on texts and emails.
Finds third-party disclosures, threats, false statements and actions not on the account, and quotes the moment.
Places every contact on the borrower’s own clock from the address on file, counts attempts per debt over seven days, applies state rules, and flags contact after an attorney, a written stop or a bankruptcy.
Hears hardship, illness, bereavement, job loss, disability and wanting to complain, and routes the account to the hardship or complaints team.
Writes one finding per conversation with the quoted moment, the coaching item and the effect on the agency score, applied when a person confirms.
Confirms each red finding; a second person signs any dismissal. The agents propose; a named person decides.
The analyst can ask in plain words about any conversation from the night’s check — or have a remediation letter drafted for a person to send.
Why is 02:41 a third-party disclosure?
At 00:14 the woman who answered said “I’m his sister”. At 02:41 Jordan told her the creditor, the loan, that Marcus is two payments behind and the amount, $1,284.50. A sister is a third party: a collector may only ask her where to reach him, not discuss the debt. Jordan had held the line twice before, at 00:37 and 02:11.
Which calls were outside the allowed hours yesterday?
Seven contacts were outside 08:00–21:00 at the borrower’s address. Pinegate: 5 (4 calls, 1 text) — the dialer took the time zone from California area codes; the borrowers live in the East. Calder & Voss: 1 at 07:42, dialled by hand on Central time. Summit Ridge: 1 at 21:06 — the borrower asked for evening calls, so it was dismissed.
How does Calder & Voss compare with the other agencies?
Calder & Voss scores 91.4, down from 94.0 thirty days ago, against Pinegate 95.2, Summit Ridge 96.9 and the bank’s own team 97.8. Yesterday it had 14 of the 38 red findings from 12% of the conversations — 2.79 per 1,000, seven times the bank’s own team’s rate. Most are third-party disclosures and threats.
Draft the remediation letter to Calder & Voss
Drafted for Rhonda Vale. It names the 4 third-party disclosures and 3 threats from Monday, leads with call C-1012-58431 at 02:41, attaches the moments and asks for a retraining plan within 10 business days. Beth Okafor sends it; I do not send to agencies.
13 screens from the working solution, on its sample data. Pick one to see it large.
41 rules from Regulation F, the FDCPA, state law and the bank’s own policy, each one switchable. A change is proposed as a new rules version for the compliance manager to approve.
| Setting | Default | Choose from |
|---|---|---|
| Allowed hours, borrower’s local time | 08:00–21:00 | 08:00–20:00 · 08:00–21:00 · 09:00–20:00 |
| Calls in 7 days, per debt | 7 | 3 to 7 |
| State rules: Massachusetts 2 calls in 7 days, New York City recording notice, Washington 3 calls a week | On | On · Off |
| Who confirms a red finding | Compliance analyst | Compliance analyst · Compliance manager · Quality analyst |
| Dismissing a red finding | A reason and a second person | On · Off |
| Clean conversations heard by a person | 1% a day | 0.5% · 1% · 2% a day |
| Vulnerability signals route to | Hardship team | Hardship team · Specialist support team · Team lead |
| Card numbers in transcripts | Redacted before anyone reads them | On · Off |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
“demo” = seen in the working solution, on its sample day of conversations · “target” = the design goal, measured in the live solution · “estimated” = our estimate · “typical” = published figures (CFPB FDCPA Annual Report 2025 — 207,800 debt collection complaints in 2024, 109,900 in 2023; MaestroQA via Retell AI — 3–5% of calls reviewed by hand). Rules cited are 12 CFR 1006 (Regulation F) and 15 U.S.C. §1692 (FDCPA). People, the bank, the agencies and borrowers named on this page are fictional — characters and sample data in the working solution.
Software that checks collectors’ conversations with borrowers against the rules that govern them — Regulation F, the FDCPA, state rules and your own policy — and records what it found and who decided. Collections Compliance Monitoring checks every call, voicemail, text, email and letter overnight, not a sample, and pins each finding to the second or sentence where it happened.
All five channels. Calls and voicemails are transcribed and time-stamped; texts, emails and letters are read sentence by sentence. Voicemails are checked as limited-content messages, texts and emails for an opt-out, and letter amounts against the bank’s system of record.
It takes the borrower’s local time from the address on file, not the phone’s area code. When the address and the phone disagree, the contact must be allowed in both. It also counts calls per debt over seven days and applies state rules such as Massachusetts’ 2 calls in 7 days.
Yes. The same checks run on every conversation from your own team and each agency, so the agency compliance scores compare like for like. Confirmed findings feed the agency’s score, and the agency management lead sends remediation letters with the call moments attached.
No. The agents propose findings with the quote and the rule; the compliance analyst hears the moment and confirms or dismisses each red finding. Dismissing a red needs a reason and a second person, and the agents never contact borrowers, block calls or write to agencies.
The review log records every conversation checked, the rules version it ran under, the finding, the decision and who made it. It is append-only and can be exported as an exam package, searchable by account, borrower or conversation.
Yes. Each of the 41 rules can be switched on or off, and the allowed hours, calls per 7 days, state rules, who confirms, the calibration sample and where vulnerability signals go are settings. Every change becomes a new rules version for approval.
The Agentic Solution Engine builds and deploys it from your requirements and documents — your collections policy, letter templates, agency list and a sample of recorded conversations — and it goes live once every quality gate has passed. We will walk you through it on a day of your own conversations first.
Every contact inside the 7-in-7 rule, spent where it pays.
1-min demoBorrower Hardship ManagementSame request, same decision, every time, with the reason written down.
1-min demoCollections Agent AssistThe one call the rules allow this week, handled right first time.
1-min demoWe’ll run Collections Compliance Monitoring on a day of your own calls, messages and letters.