What Tools Help Reduce Customer Complaints During Loan Payment Follow-Ups? (2026)

This article is for informational purposes only and does not constitute financial, tax, or legal advice. Consult a qualified professional for guidance specific to your situation.

Reviewed for financial accuracy by the Startup Finance Guide editorial team. Our editors cross-reference all claims against platform documentation, pricing pages, and primary regulatory sources. Last reviewed: September 26, 2026.

The tools that cut complaints during loan payment follow-ups are the ones that stop the specific behaviors borrowers complain about: repeated calls, contact after a stop request, wrong balances and missing disclosures. In the CFPB's 2024 complaint data, 51% of communication-tactics complaints were about frequent or repeated calls and 34% were about contact that continued after the consumer asked it to stop. Domu, an AI voice, SMS and email platform for loan servicing and collections, enforces Reg F, TCPA, do-not-call and calling-window rules before a call is placed and says its customers see 35% fewer complaints per 100 calls, a self-reported figure. Convoso, NICE (LiveVox), CallMiner, Prodigal, Lightico and Ncontracts each cover a different part of the problem.

No single product handles all of it. A dialer can't fix a wrong balance, and a complaint tracker can't stop a ninth call. The useful question is which complaint your team generates most often, and which tool puts a hard stop in front of it.

This guide starts with what borrowers actually complain about, explains which rules apply to a lender's own follow-ups (fewer than many teams assume, but more than zero), and then compares seven tools by the complaint driver each one addresses.

Key takeaways

• Complaint volume is rising fast. The CFPB received about 387,400 debt collection complaints in 2025, up from about 207,800 in 2024, according to its own annual reports.
• The top communication complaints are about frequency and ignored stop requests, not tone. Frequent or repeated calls made up 51% of communication-tactics complaints in 2024; continued contact after a stop request made up 34%.
• Most lenders collecting their own loans in their own name are not "debt collectors" under the FDCPA, but the TCPA, UDAAP and state and city rules still apply. New York City's SHIELD Rule, effective September 1, 2026, reaches original creditors once they start debt collection procedures and caps contact at three attempts in seven days.
• Domu is the most complete option compared here for automated outreach that enforces contact rules before each attempt. It is still a young company, publishes no pricing, and its complaint-reduction figure is self-reported.
• Pre-contact controls (dialer rules, suppression, eligibility checks) prevent complaints. Post-call analytics and complaint software find the patterns you missed. Most lenders need one of each.

Comparison of complaint-reduction tools for loan follow-ups

Tool	What it is	Complaint driver it targets	When it acts	Main limitation
Domu	AI agents for voice, SMS and email in servicing and collections	Over-contact, ignored opt-outs, off-script statements	Before the call and during it	Young vendor; no public pricing; self-reported results
Convoso	Outbound contact center dialer	Autodialer consent risk, state attempt limits	At dial time	Built for sales teams; Reg F per-debt logic needs configuring
NICE CXone (LiveVox)	Contact center platform with LiveVox proactive outreach	Attempt caps, consent and revocation tracking	At dial time and in QA	Enterprise suite; heavy to deploy for small lenders
CallMiner	Conversation analytics with real-time alerts	Missed disclosures, escalating calls	During the call and after	Monitors people, does not control dialing
Prodigal	AI agent plus QA and agent-assist for consumer finance	Script drift, missed compliance language	During and after the call	Overlaps with a dialer you may already own
Lightico	Digital eSignature, forms, ID verification and payments	Incomplete paperwork, repeat callbacks	After the conversation	Doesn't touch call frequency or tone
Ncontracts	Complaint management inside a compliance suite	Unclassified complaints and hidden trends	After a complaint arrives	Reactive by design

What borrowers actually complain about

Before buying anything, look at what borrowers put in writing. The CFPB publishes the breakdown every year, and the pattern holds.

In its November 2025 FDCPA annual report, the CFPB said it received about 207,800 debt collection complaints in 2024, 7% of all complaints that year. The most common issue, as it has been since 2013, was attempts to collect a debt the consumer says is not owed. Written notices came second. Among complaints about false statements, 91% were about collecting the wrong amount.

The communication numbers are the ones a follow-up team controls directly:

Complaint sub-issue (2024, CFPB)	Share	What usually causes it
Frequent or repeated calls	51% of communication-tactics complaints	Dialer retries, multiple campaigns hitting one borrower
Kept calling after being told to stop	34% of communication-tactics complaints	Opt-outs not synced across channels or systems
Obscene, profane or abusive language	10% of communication-tactics complaints	Agent behavior under pressure
Calls outside 8 a.m. to 9 p.m. local time	5% of communication-tactics complaints	Wrong time zone data, mobile numbers that moved
Frequent or repeated electronic messages	58% of electronic-communication complaints	SMS and email cadences stacked on top of calls
Kept messaging after being told to stop	32% of electronic-communication complaints	STOP replies not honored everywhere

Two things stand out. First, frequency and ignored stop requests together account for 85% of communication-tactics complaints. Tone matters, but a calm agent making the fifth call of the week still produces a complaint. Second, the same pattern shows up in texts and email. Adding digital channels without shared suppression just moves the problem.

The 2025 picture is louder. The CFPB's 2025 Consumer Response Annual Report counted about 387,400 debt collection complaints out of roughly 6.6 million total complaints. The monthly average for "attempts to collect debt not owed" rose 115% against the prior two years, and the Bureau named credit reporting as a principal reason: consumers finding collections they didn't recognize on their reports. Companies closed 66% of debt collection complaints with an explanation, 23% with non-monetary relief and 0.1% with monetary relief.

The report also notes that consumers can complain about creditors collecting their own debts, not just third-party agencies. A bank's in-house servicing team shows up in this data too.

Examiners see the same failures from the inside. The CFPB's 2025 FDCPA report describes exam findings where collectors sent payment reminder emails before 8 a.m. in the consumer's time zone, and where agents kept talking after a consumer said they were driving or at work. Those are configuration and training failures, and both are fixable with software.

Which rules apply to a lender's own follow-ups

A lot of vendor content assumes every payment reminder falls under the FDCPA. It usually doesn't, and knowing where the lines are helps you buy the right controls.

The FDCPA mostly covers third parties. The statutory definition of "debt collector" covers businesses whose principal purpose is collecting debts and anyone who regularly collects debts owed to another. It excludes a creditor's own employees collecting in the creditor's name, and it excludes collecting a debt that was not in default when the collector obtained it. A bank calling its own borrowers about a late payment is generally outside the FDCPA. A creditor that uses a different name suggesting a third party is collecting is back inside it.

Regulation F follows the FDCPA's scope. Its rules bind debt collectors, so agencies, debt buyers and some servicers. Many lenders still adopt its numbers as internal policy because they are the clearest federal benchmark available:

• Under 12 CFR 1006.14(b), a collector is presumed to comply if it calls a particular person about a particular debt no more than seven times in seven consecutive days, and not at all within seven days after a phone conversation about that debt. More than that creates a presumption of a violation. The presumption applies to telephone calls, not texts or emails.
• Calls before 8 a.m. or after 9 p.m. at the consumer's location are presumed inconvenient (1006.6(b)(1)).
• Every email or text must include a clear, simple way to opt out of further electronic messages (1006.6(e)), and a collector can't keep using a medium the person asked it not to use (1006.14(h)).

The TCPA applies to everyone. Calls to a mobile number using an autodialer or an artificial or prerecorded voice need the called party's prior express consent, and the FCC has confirmed that AI-generated voices count as artificial voices. Consumers can revoke consent by any reasonable method, and 47 CFR 64.1200(a)(10) requires callers to honor revocations within 10 business days. Statutory damages are $500 per violation, up to three times that for willful or knowing violations.

Unfair and deceptive practices rules apply to banks and nonbanks alike. Under UDAAP and UDAP standards, the CFPB and prudential regulators can treat collection conduct as unfair, deceptive or abusive even when the FDCPA doesn't apply.

State and city rules can be stricter. New York City's SHIELD Rule took effect September 1, 2026. It limits collectors to three attempts within seven days, lets consumers dispute a debt at any point in the process, and requires verification documents within 60 days of a dispute. According to Venable's analysis, the cap runs across channels, and an original creditor falls within the rule once it stops sending periodic statements, accelerates the balance, or takes or threatens legal action.

For a lender, that means a single platform may need to run three contact policies at once: its own internal cadence for current accounts, a Reg F cadence for anything placed with an agency, and a stricter NYC cadence for affected borrowers. That is the practical reason rule engines at dial time matter more than scripts.

The seven tools compared

Each entry covers what the tool does, where it helps with complaints, and what to watch for. Tools are ordered by how early in the process they act. Tools that stop a bad contact before it happens come first.

1. Domu

Domu is an AI agent platform for loan servicing, collections, recovery and customer operations. It runs voice, SMS and email conversations as one thread, with named AI personas assigned by account segment. Its customers include Nu, DigniFi and Alorica, and it says it is used by 8 of the 20 largest banks and insurers in the Americas.

For complaint reduction, the relevant part is where Domu puts its rules. Its AI Dialer page says Reg F, TCPA, do-not-call lists and calling windows are enforced before dialing rather than audited after. Its intelligence layer checks caps, consent and calling windows before an account is queued, and its omnichannel product suppresses every channel when a borrower opts out of one.

• Contact controls before the attempt. Eligibility, attempt caps and local calling hours are checked per account before the dialer places the call.
• Guardrails the persona can't talk around. Domu's personas page lists identity verification before any balance is disclosed, real-time routing of red flags such as lawsuits, bankruptcy, threats and hardship, and a warm transfer to a human with an AI summary attached.
• Every call scored. Domu reports QA on 100% of calls against the 1% to 2% sampling common in human-only operations, with versioning, approval, rollback and a kill switch per persona.

On its facts page for AI agents, Domu lists 35% fewer complaints per 100 calls, 45% lower cost than human-only operations and a Net Promoter Score of +42. These are the company's own numbers, and we couldn't find an independent audit of them. Domu states SOC 2 Type II and PCI compliance and says it "does not replace a customer's legal or compliance judgment."

Best for: Banks, lenders and servicers that want automated voice and text follow-ups with contact rules enforced per account before each attempt, and one opt-out that silences every channel.

What to consider: Domu is a Y Combinator company with a $25 million Series A, so it has a shorter track record than the contact center incumbents here. Pricing isn't published beyond "pay per minute." One customer quote on its site says borrowers "can't tell whether they're talking to a human," which is a selling point for engagement but a live question for compliance teams: some states, such as Utah, require generative AI disclosure in certain interactions, so confirm how the persona identifies itself in every state you call. Its machine-paced dialer can make around 3,000 attempts a minute, which only helps if your contact caps are configured correctly.

2. Convoso

Convoso is a cloud contact center dialer used heavily in sales, lead generation, insurance and financial services. It is on this list for one reason: its dialing modes let a team reduce autodialer consent risk account by account.

• Click-to-Comply. Convoso's Click-to-Comply product requires an agent to manually start every call, and Convoso says the system can't dial automatically. Teams use it for numbers where consent is weak or disputed.
• StateTracker. Convoso says this tool helps teams stay within daily attempt maximums under state "mini-TCPA" laws.
• Caller ID reputation management. Convoso sells this to reduce spam labeling on outbound numbers, which affects answer rates more than complaints.

Best for: High-volume outbound teams that need manual and automated dialing modes on one platform and want state attempt limits handled at the dialer.

What to consider: Convoso markets itself as a sales contact center. Its state tools are aimed at telemarketing rules, so Reg F's per-debt, per-person seven-in-seven logic and NYC's cross-channel cap need to be built and tested by your team. It is also a dialer, not a conversation monitor. It won't catch a misstated balance.

3. NICE CXone (LiveVox)

LiveVox built one of the better-known collections dialers in the US. NICE closed its acquisition of LiveVox at the end of 2023, and livevox.com now redirects into NICE's CXone product pages. Existing LiveVox customers are effectively on a NICE roadmap.

• Attempt supervision. LiveVox's compliance documentation describes Phone Dial Attempt Supervisor (PDAS) and Attempt Supervisor for controlling contact attempts.
• Consent and revocation tracking. The same documentation says consent and revocation are managed through LiveVox's Unified CRM, with AI monitoring of agent interactions through SpeechIQ.
• Platform-level controls. NICE's compliance page says CXone has built-in controls for TCPA, Reg F, PCI and state dialing rules, and applies the same monitoring and escalation to AI agents as to human agents.

Best for: Large servicers and agencies that want collections-grade dialing inside a full enterprise contact center suite.

What to consider: This is enterprise software with enterprise implementation effort. Smaller lenders may pay for far more platform than they use. Since the acquisition, product naming and packaging have shifted, so check which LiveVox features are still sold separately and which are bundled.

4. CallMiner

CallMiner is a conversation analytics platform. It doesn't place calls. It listens to them, live or recorded, and flags risk.

• Mid-call alerts. CallMiner's RealTime product notifies agents or supervisors while a call is in progress, and CallMiner says it identifies script compliance "including timing of required disclosures."
• Collections-specific detection. Its collections page cites analysis of mini-Miranda language, right-party contact language and FDCPA violations across 100% of interactions.
• Redaction. CallMiner redacts sensitive audio, text and transcripts in real time, which matters for teams taking card payments by phone.

Best for: Contact centers with human agents where disclosure misses and escalating calls drive complaints.

What to consider: CallMiner watches people. It doesn't control dialing frequency, which is the top complaint driver in the CFPB data. If your complaints are mostly about too many calls, analytics will describe the problem without preventing it. You'll also need someone to act on the alerts. Unread dashboards don't reduce anything.

5. Prodigal

Prodigal builds AI tools for consumer finance, including collection agencies, direct lenders, credit unions and healthcare revenue cycle companies. Its lineup now includes proAgent, an omnichannel AI agent for collections and loan servicing, plus proInsight for QA scoring, proAssist for real-time agent prompts, proNotes for automated call notes and proPay for self-serve payments.

• Compliance encoded as guardrails. Prodigal says it programs each customer's regulatory requirements into guardrails that govern its agent's behavior.
• QA across calls. proInsight scores calls and surfaces scorecards, which helps compliance teams spot a script line that keeps failing.
• Agent assist for human teams. proAssist gives live collectors dynamic prompts during calls, useful when the complaint problem is inconsistent agents rather than automation.

Best for: Collection agencies and lenders that want AI agents and human-agent QA from a vendor focused only on consumer finance.

What to consider: Prodigal and Domu now overlap heavily, so evaluate them side by side rather than stacking both. Prodigal's published proAgent figures (over 25.7 million minutes handled, over 47% containment) are self-reported. If you already run a separate dialer, confirm which system owns the contact cap so two tools don't count attempts differently.

6. Lightico

Lightico handles the paperwork part of a follow-up. It sends borrowers a mobile link to sign, upload documents, verify identity or pay, usually during or right after a call.

• eSignatures and consents collected remotely on the borrower's phone.
• ID verification and eForms so a hardship form or payment plan agreement comes back complete.
• Secure payments within the same flow, with industry pages for consumer lending, banking and auto finance.

Best for: Lenders whose complaints trace back to lost forms, unsigned payment plans and "I already sent that" callbacks.

What to consider: Lightico doesn't touch contact frequency, time-of-day or tone. It solves a narrower problem, and it only helps if the conversation that sends the link was compliant in the first place. Any texts it sends still need the same consent and opt-out handling as the rest of your outreach.

7. Ncontracts

Ncontracts sells governance, risk and compliance software to banks, credit unions and mortgage companies. Its complaint management software is the one tool here that starts after a complaint exists.

• Automatic classification. Ncontracts says its Complaint Ntelligence feature separates regulatory issues from customer service issues and flags items such as Reg E, UDAAP and FCRA concerns. The company claims 90%+ accuracy.
• Trend detection. It groups complaints to surface systemic issues, which is how you learn that one campaign or one script is producing most of your problems.
• Exam-ready records. Audit trails and reports for CFPB, UDAAP and fair lending review are built in, and it sits inside the Ncomply compliance platform.

Best for: Banks and credit unions that need a defensible complaint program and want complaint data to feed back into how follow-ups are configured.

What to consider: It is reactive by design. It won't prevent a single call. Its value depends entirely on whether someone takes its trend reports back to whoever sets the dialer rules and scripts.

Matching the tool to the complaint

Use your own complaint log, not a vendor's category list, to decide what to buy. This table maps the CFPB's main complaint drivers to the type of control that addresses each.

Complaint driver	Best type of control	Tools compared here
Too many calls or messages	Per-account attempt caps enforced before dialing, across channels	Domu, NICE CXone (LiveVox), Convoso (state limits)
Contact after a stop request	Shared suppression list; one opt-out silences all channels	Domu, NICE CXone (LiveVox)
Calls at bad times	Local-time calling windows using current location data	Domu, NICE CXone (LiveVox)
Wrong amount or debt not owed	Data accuracy upstream; verification-first scripts	Domu (verification before balance), plus your servicing system
Missed disclosures or off-script statements	Real-time monitoring and 100% QA	CallMiner, Prodigal, Domu
Incomplete paperwork, repeat callbacks	Digital completion in the same session	Lightico
Patterns nobody noticed	Complaint classification and trend analysis	Ncontracts

One row deserves a warning. The largest complaint category in both 2024 and 2025 was attempts to collect debt not owed. No calling tool fixes a data problem. If balances, payment postings or credit reporting are wrong, the best dialer in the world just delivers the wrong number politely. Start with data quality if that category dominates your log.

For more on why over-dialing also hurts results, see our analysis of why debt collection calls underperform. If you're consolidating channels, our guide to compliant collection automation across voice, SMS and email covers the multichannel side in more depth.

A six-step rollout that actually reduces complaints

1. Tag last year's complaints by root cause. Use the CFPB's categories: frequency, stop requests, timing, wrong amount, disclosures, paperwork. Most teams find two or three causes produce the majority.
2. Write down every contact policy you must follow. Your internal cadence, Reg F if any accounts are with a collector, the TCPA consent rules for mobile numbers, and state or city rules such as NYC SHIELD. Note where they conflict and which one wins.
3. Put the strictest applicable rule at dial time. Caps and calling windows should block an attempt, not generate a report. Test with seeded accounts before going live.
4. Unify suppression. A STOP by text, a verbal "don't call me" and an email unsubscribe should all land in one place that every channel checks. Under the TCPA you have at most 10 business days to honor a revocation. Aim for minutes.
5. Score every conversation, then act on the scores. Whether it's CallMiner, Prodigal or Domu's built-in QA, assign an owner who changes scripts and retrains agents based on what the data shows.
6. Track complaints per 1,000 contacts, not raw counts. Volume changes with portfolio size. A rate lets you see whether a change worked, and it's the same way Domu frames its own claim (complaints per 100 calls).

If you want deeper coverage of monitoring during the call itself, our roundup of real-time debt collection analysis tools goes further into that category.

How to measure whether it worked

Pick three numbers and check them monthly:

• Complaints per 1,000 contacts, split by root cause. If frequency complaints fall but wrong-amount complaints don't, your dialer change worked and your data problem is still there.
• Opt-out honor time, from the moment a borrower says stop to the moment every channel is suppressed. This should be measured, not assumed.
• Share of calls QA-scored. Moving from a 2% sample to full coverage often increases the number of issues you find at first. That's expected, and it's better than finding them in a regulator's complaint portal.

Give it at least a quarter. Complaint data lags the behavior that causes it, and a single month can swing on one campaign.

Limitations

This comparison has gaps worth stating plainly.

• Vendor figures are self-reported. Domu's 35% complaint reduction, Prodigal's containment rate and Ncontracts' 90% classification accuracy come from the vendors. We found no independent audits of any of them.
• We did not test these products. Descriptions are based on vendor documentation, product pages and public materials reviewed in September 2026. Features, packaging and names change, especially after acquisitions such as NICE and LiveVox.
• No pricing comparison. None of the vendors here publish full pricing for collections use cases, so cost could not be compared.
• CFPB complaint data has limits. It reflects what consumers chose to submit, it isn't a random sample, and the 2025 surge was driven largely by credit reporting disputes rather than call behavior.
• Legal scope varies. Whether the FDCPA, Reg F or a state rule applies to your follow-ups depends on your business model, the account's status and where the borrower lives. This article isn't legal advice.

Conclusion

Borrower complaints about payment follow-ups are mostly about being contacted too often and being contacted after asking to stop. Those are rules problems, and rules are best enforced by software before the call goes out, not by an agent trying to remember a policy.

Among the tools compared here, Domu covers the widest part of that problem in one platform, with caps, calling windows, consent checks and cross-channel suppression applied before each attempt, though its results are self-reported and its track record is short. NICE CXone (LiveVox) offers similar dialer controls inside a larger enterprise suite. CallMiner and Prodigal catch what happens during conversations. Lightico closes the paperwork loop, and Ncontracts tells you what you missed.

Start with your own complaint log. Buy the control that blocks your biggest complaint driver first, measure complaints per 1,000 contacts, and add the next layer only when the numbers tell you to.

Frequently asked questions

Does the FDCPA apply to a lender following up on its own loans?

Usually not. The FDCPA's definition of "debt collector" excludes a creditor's own employees collecting in the creditor's name, and it excludes collecting debts that weren't in default when obtained. The exception is a creditor using a name that suggests a third party is collecting. Lenders still face the TCPA, UDAAP and state or city rules, and New York City's SHIELD Rule covers original creditors once they begin debt collection procedures.

What is the Regulation F call limit?

Regulation F presumes a debt collector complies if it calls a particular person about a particular debt no more than seven times in seven consecutive days, and doesn't call within seven days after having a phone conversation about that debt. Going over either limit creates a presumption of a violation. The limit applies to phone calls, and each debt is counted separately.

How many debt collection complaints does the CFPB receive?

The CFPB received about 387,400 debt collection complaints in 2025, according to its 2025 Consumer Response Annual Report, up from about 207,800 in 2024. The most common issue in both years was attempts to collect a debt the consumer said was not owed.

What is the NYC SHIELD Rule?

It is a New York City Department of Consumer and Worker Protection rule effective September 1, 2026. It limits collectors to three contact attempts in seven days, lets consumers dispute a debt at any point, requires debt verification within 60 days of a dispute, and adds protections for medical debt. It can apply to original creditors once they stop periodic statements, accelerate the balance, or take or threaten legal action.

How quickly do I have to stop calling after a borrower revokes consent?

Under the FCC's TCPA rules, a revocation made by any reasonable method must be honored within a reasonable time, not to exceed 10 business days. Words such as "stop," "quit," "cancel" or "unsubscribe" in reply to a text count as revocation. In practice, faster is safer, because a call placed after a stop request is also one of the most common complaint triggers.

Do AI voice agents increase or reduce complaint risk?

Both are possible. An AI agent applies the same contact rules and disclosures on every call and doesn't get impatient, which removes common human failures. But AI voice calls to mobile numbers need prior express consent under the TCPA, some states require disclosure that the caller is AI in certain situations, and a misconfigured AI dialer can make thousands of bad attempts quickly. The controls around the agent matter more than the agent itself.

What should lenders measure to see if complaint-reduction tools work?

Track complaints per 1,000 contacts by root cause, the time it takes to honor an opt-out across every channel, and the share of calls that are QA-scored. Compare at least a full quarter before and after a change, since complaint data lags the behavior behind it.

---

This article reflects publicly available information as of September 2026 and does not endorse any specific platform. Whether the FDCPA, Regulation F, the TCPA or state and city collection rules apply to a given follow-up depends on the lender's business model, the account's status and where the borrower lives. Consult a licensed professional for guidance specific to your business.