AI Debt Collection Statistics 2026

Source: NY Fed Household Debt and Credit Report

24/7. No signup required.

Section 1

Macro debt landscape

US household debt sets the ceiling for the entire collections industry. When mortgages, credit cards, autos, and student loans expand, downstream delinquencies and collection placements expand with them. Every figure below traces back to the Federal Reserve Bank of New York Household Debt and Credit Report or the Federal Reserve G.19 Consumer Credit release.

$18.04T

Total US household debt, Q4 2024

$12.61T

US mortgage balances, Q4 2024

$1.21T

Credit card debt outstanding, Q4 2024

$1.64T

Auto loan balances, Q4 2024

$1.62T

Student loan balances, Q4 2024

Source: Federal Reserve G.19 Consumer Credit Release

21.22%

Average credit card APR (Q4 2024)

$5,116.8B

Total US consumer credit outstanding (Feb 2026)

Source: Federal Reserve G.19

7.18%

Credit card 90+ day delinquency rate, Q4 2024

Source: NY Fed Quarterly Report on Household Debt and Credit

Why this matters for collections: The NY Fed shows credit card balances at $1.21T with the highest 90+ DPD rate of any major consumer category. Auto subprime delinquency is at a 15-year high. Student loan reporting resumed in late 2024 after the COVID forbearance pause, and the resulting delinquency spike will flow into collections placement volumes through 2026 and 2027.

Section 2

CFPB complaints + enforcement

The Consumer Financial Protection Bureau is the primary federal regulator of the consumer debt collection industry. Its annual FDCPA report to Congress is the highest-fidelity public source on consumer-facing collection problems and Bureau priorities.

207,800+

Debt collection complaints CFPB received in 2023

1st

Debt collection ranks first among complaint categories CFPB tracks

Source: CFPB Newsroom, Sept 2024

Medical & rental

Two CFPB priority focus areas in 2024 FDCPA report

Attempts to collect debt not owed

Most common consumer complaint type in debt collection

Source: Bridgeforce, 2026

Operator implication: The Bureau has flagged medical and rental debt as the two priority enforcement areas in its 2024 FDCPA report. The plurality of consumer complaints involve "attempts to collect debt not owed", which is a data-quality and identity-verification problem before it is a tactics problem. Cleaner placement files and identity-first call flows reduce CFPB risk surface.

Section 3

Cost economics

Collections is a unit-economics business. The fully-loaded cost per outbound contact, multiplied by attempts-per-account, divided by dollars collected, is what determines whether a portfolio works. AI changes the cost-per-contact line.

Channel	Cost per contact	Notes
Voice (live agent)	Your loaded agent cost per contact	Most expensive per contact by a wide margin, and still the strongest channel for negotiating a payment arrangement
Voice (AI agent)	Quoted per deployment	Vendors quote per deployment rather than publishing a rate card; the comparison that matters is against the row above, priced from your own payroll
SMS	Your carrier rate per message	Cheap per message and cheap to scale; consent and Reg F frequency rules bind before cost does
Email	Effectively free at volume	Cheapest reach; conversion lags voice and SMS
Self-serve portal	Fixed platform cost, marginal per use	No per-contact cost once built; TrueAccord publicly reports strong satisfaction on this channel
Letter / IVR	Your print and postage rate	Compliance-driven; required disclosure delivery

No figure

Cost per live collections contact is quoted everywhere and sourced nowhere. We could not trace a single published range to a stated sample and method, so this page declines to print one. Cost it from your own payroll, dialer and quality-monitoring lines.

Stated absence, checked September 2026. A number without a method is not a benchmark.

CPDC

Cost per dollar collected. Bridgeforce identifies this as the foundational unit-economics KPI: total collection expense divided by total dollars collected, tracked monthly per portfolio.

Section 4

Recovery benchmarks

Bridgeforce's 2026 credit-union KPI framework defines the canonical metric set: right-party contact (RPC), promise-to-pay (PTP), kept-promise rate, roll/cure rate, cost per dollar collected, legal recovery rate, and net charge-off rate. Vendor public material adds AI uplift figures on top.

RPC rate

Successful contacts with the responsible party / total contact attempts

PTP rate

Calls resulting in a payment commitment / total RPCs

Kept-promise rate

Dollars received on or before promised date / dollars promised

Roll / cure rate

Accounts moving to next delinquency bucket vs accounts curing monthly

Cost per dollar collected

Total collections expense / total dollars collected

Legal recovery rate

Dollars recovered through legal action / placed balance

Net charge-off rate

Net charge-offs / average loans, by product

Recovery rate

Cash collected on charge-offs / total charged-off dollars

Member experience

Post-resolution satisfaction + complaints per 1,000 contacts

Source: CFPB Medical Debt Burden in the United States

Public AI uplift figures

Symend: +10% recovery rate, -85% agent interactions, -50% OpEx, 10x ROI across enterprise clients (source)
Vodex.ai: 3x debt recovery improvement, 7x connect rate improvement in featured case study (source)
Skit.ai: $1B+ accounts resolved, ~1B conversations across deployments, 53,000+ creditors served (source)

Note: These are vendor self-reported claims. Independent third-party validation across portfolios remains limited.

Section 5

Medical debt

Medical debt is the largest single category of collection tradelines on US credit reports and the most politically active subsegment in 2025-2026. The CFPB Medical Debt Burden in the United States report (March 2022) remains the canonical federal data source.

$88B

Medical bills on US consumer credit reports (CFPB 2022)

58%

Share of all collection tradelines that are medical

Source: CFPB Medical Debt Burden Report 2022

43M

US credit reports carrying medical-collection tradelines

Source: CFPB Medical Debt Burden Report 2022

20%

US households reporting medical debt

Source: CFPB Medical Debt Burden Report 2022

July 11, 2025

Date US Eastern District of Texas vacated CFPB medical-debt rule

Source: Cornerstone Credit Union League v. CFPB

2025 regulatory reversal

The CFPB rule finalized in early 2025 to remove medical debt from consumer credit reports was vacated on July 11, 2025, when the US District Court for the Eastern District of Texas ruled in Cornerstone Credit Union League v. CFPB that the Bureau had exceeded its statutory authority and contradicted the Fair Credit Reporting Act. The court action means medical debt may again appear on consumer credit reports under standard FCRA rules in 2026.

Source: CFPB Newsroom, updated 24 Feb 2026

Section 6

Compliance landscape

US debt collection sits inside a dense compliance perimeter: FDCPA, TCPA, Reg F (effective November 30, 2021), state licensing, and parallel CFPB and FTC enforcement. AI voice agents must clear every layer.

FDCPA

The Fair Debt Collection Practices Act has been the bedrock federal statute since 1977. The CFPB became the primary federal regulator in 2010. The Bureau's 2024 FDCPA Annual Report to Congress is the canonical view of consumer-facing problems.

Source: CFPB Reg F Final Rule

Reg F

Regulation F took effect 30 November 2021 and modernized FDCPA implementation - codifying the 7-in-7 call cap, governing electronic communications (email, SMS, social), and requiring the Validation Notice with itemization at or near the first communication.

TCPA

The Telephone Consumer Protection Act of 1991 governs autodialed calls and pre-recorded messages. Violations carry $500-$1,500 per call statutory damages, making TCPA risk a primary driver of dialer architecture choices.

Source: FCC TCPA Rules

State licensing

Most US states require collection-agency licensing with surety-bond and registration filings. ACA International tracks state-by-state licensing requirements and compliance updates.

Source: ACA International

Section 7

AI voice agent performance

Public claims from the largest AI debt-collection vendors. We pull every figure directly off vendor homepages and case-study pages, and link the source so you can verify against the live page.

$50B+

Total debt Symend reports recovered for clients

250M+

Delinquency events Symend has treated

+10%

Recovery-rate uplift Symend claims vs traditional methods

-85%

Reduction in agent interactions Symend reports

-50%

OpEx reduction Symend claims with its platform

53,000+

Creditors served by Skit.ai

19+

Debt types Skit.ai supports

$1B+

Accounts resolved by Skit.ai

Conversations Skit.ai claims across deployments

8 years

Skit.ai live deployment history in collections + AI

Recovery rate improvement Vodex.ai cites in featured case study

Connect-rate improvement Vodex.ai reports

SOC 2 + ISO 27001

Vodex.ai security certifications

HIPAA, FDCPA, Reg F, TCPA

Vodex.ai compliance attestations

Source: NY Fed Household Debt and Credit Report

Caveat: These are vendor-reported figures from marketing pages. They are real claims that vendors stand behind publicly, but independent peer-reviewed validation across portfolios remains limited. Treat them as directional - the cost-economics direction is robust, the exact uplift percentages are deal-by-deal.

Section 8

Vertical breakdowns

Recovery economics differ wildly by vertical - the same dialer strategy can produce very different recovery rates on credit-card balances versus auto deficiency balances. NY Fed Q4 2024 data plus public industry sources for averages and delinquency.

Vertical	Average balance	Delinquency	Recovery note
Credit Card	$6,580 avg per cardholder (TransUnion 2024)	7.18% 90+ DPD (NY Fed Q4 2024)	Highest dialer activity, peak charge-off bucket
Auto Loan	$24,297 avg new loan (Experian Q4 2024)	4.78% 90+ DPD (NY Fed Q4 2024)	Subprime delinquency at 15-year high
Mortgage	$252,505 avg per mortgage (NY Fed)	0.95% 90+ DPD (NY Fed Q4 2024)	Lowest delinquency tier; highest exposure
Student Loan	$38,883 avg per borrower (Education Data Initiative)	~15.6% (NY Fed, post-resumption Q4 2024)	Reporting resumed late 2024; spike expected
Medical	Varies (median under $500 per tradeline; CFPB)	~58% of all collection tradelines (CFPB 2022)	July 2025 court vacated CFPB removal rule
BNPL / Fintech	$135 avg loan (CFPB BNPL Report 2023)	~7% 30+ DPD (Federal Reserve Bank research)	Heavily skewed to digital and SMS reminders

Sources for this table, named per cell rather than linked per cell: delinquency rates from the Federal Reserve Bank of New York Household Debt and Credit Report, Q4 2024; medical tradeline share and BNPL average loan size from CFPB; average balances from TransUnion, Experian and the Education Data Initiative as labelled in each cell. The BNPL delinquency figure is labelled "Federal Reserve Bank research" in the cell without naming a specific paper, and should be read as indicative rather than as a cited rate.

Auto

Subprime auto delinquency at a 15-year high per NY Fed Q4 2024. Dealer-financed and subprime portfolios drive the bulk of placement volume.

BNPL

Average loan size is small ($135 per CFPB BNPL Report 2023) which tilts unit economics aggressively toward digital and AI voice over staffed dialers.

Healthcare

$88B in collections per CFPB. HIPAA layered on top of FDCPA. Vodex publishes HIPAA attestation; few competitors do.

Section 9

Geography

US collections concentrate in states with the largest credit footprints. EU late-payment data tells a parallel story for cross-border operators.

US distribution

Top states by collection volume track total household debt: California, Texas, New York, Florida, Pennsylvania.
Subprime auto delinquency concentrates in the South and Midwest (NY Fed regional breakdowns).
Medical debt concentrates in states without Medicaid expansion - Black households at 28% prevalence vs 17% for White households per CFPB 2022.
State licensing requirements vary; ACA International maintains a state-by-state tracker.

EU late payment

EU B2B late payment averages roughly 50% of invoiced value paid late per Intrum European Payment Report public series.
Average days-late on EU B2B invoices typically falls 8-15 days past terms.
EU AI Act took effect August 2024 with phased obligations; Annex III high-risk systems face full compliance by 2 December 2027 after Regulation (EU) 2026/1744.
GDPR remains the primary data-protection layer; consent and legitimate-interest balancing govern collection outreach.

Source: Intrum European Payment Report

Section 10

Channel mix

Modern collections is omnichannel. Reg F formally codified email, SMS, and social electronic communications in November 2021. The right channel depends on debtor age, balance size, and prior engagement history.

Voice (live + AI)

Highest RPC and PTP capture; AI voice removes the collector time from each attempt, opening up tiers of debt that previously did not work economically. Symend reports an 85% reduction in agent interactions when AI handles the digital and reminder layer.

SMS

Highest open rate in the channel mix; preferred opener for digital-native borrowers. Reg F requires unsubscribe mechanism on every text. Often paired with self-serve payment links.

Sub-cent cost-per-send. Conversion lags voice and SMS but volume scales. Reg F validation notice requirements apply when emailed.

Self-serve portal

Highest CSAT post-resolution per public TrueAccord material. Best at recovering smaller balances where a debtor wants to settle quickly without a call.

OpenAIAnthropicAWSGoogleElevenLabs●TelnyxOpenAIAnthropicAWSGoogleElevenLabs●TelnyxOpenAIAnthropicAWSGoogleElevenLabs●Telnyx

Jessica·English

Click to start a conversation

Call our debt collection demo

24/7 live AI - call anytime

Or call directly. 24/7 demo line:

Justas Butkus

Founder & CEO, AInora

Building AI digital administrators that replace front-desk overhead for service businesses across Europe. Previously built voice AI systems for dental clinics, hotels, and restaurants.

View all articles

Deep dives into AI debt collection: software, costs, compliance, industry benchmarks.

AI Debt Collection Overview

How AI is transforming debt recovery - technology, compliance, real-world results.

Best AI Debt Collection Software

Full comparison of 10 AI debt collection platforms - features, strengths, fit.

AI Debt Collection Costs

Per-minute rates, setup fees, ROI calculations, total cost of ownership.

Skit.ai Alternatives

10 platforms compared, including European and developer-first options.

FDCPA + TCPA Compliance Guide

How AI voice agents stay compliant with US consumer-protection law.

Industry Benchmarks

Documented AI debt collection results from public vendor case studies.

Frequently asked questions

12 questions every collections operator asks about industry data.

Bridgeforce defines RPC as successful contacts with the responsible member divided by total contact attempts, but does not publish a hard benchmark - because RPC swings dramatically by vintage, channel, and dialer strategy. Public vendor material (Vodex, Symend) cites RPC uplift of 3x to 7x against manual baselines, which means typical manual operations sit in the single-digit to low-teen percent range, and AI-augmented operations push toward 20-30%+ on early-stage portfolios. Always treat RPC as a relative metric against your own baseline, not an absolute industry constant.

We do not publish a figure for this. The per-contact cost of a live collections call is widely quoted in ranges, and we could not trace any of them to a source with a stated sample and method, so this page does not supply one. Cost the contact from your own payroll, dialer and quality-monitoring lines instead. AI voice agent pricing is quoted per deployment rather than published. The cost-per-dollar-collected (CPDC) ratio - which Bridgeforce identifies as a core credit-union KPI - is where the economics actually show up, and it is measurable from your own book.

Per the CFPB Medical Debt Burden in the United States report (March 2022), 58% of bills in collections on consumer credit records were medical bills, appearing on roughly 43 million US credit reports and totaling about $88 billion. The CFPB estimated 20% of US households reported having medical debt. These remain the most-cited federal numbers; the CFPB has not published a refreshed total since the 2022 report.

No. On July 11, 2025, the US District Court for the Eastern District of Texas vacated the CFPB rule in Cornerstone Credit Union League v. CFPB, ruling the agency had exceeded its statutory authority and contradicted the Fair Credit Reporting Act. As of 2026, medical debt may again appear on consumer credit reports under standard FCRA rules. Source: CFPB Newsroom, updated February 24, 2026.

The CFPB FDCPA Annual Report covering 2023 documented over 207,800 debt collection complaints, with debt collection ranking among the top consumer-complaint categories the Bureau tracks. The two priority focus areas highlighted in the Bureau's 2024 annual FDCPA report were medical debt collection and rental debt collection. The most common complaint type is "attempts to collect debt not owed".

The Federal Reserve Bank of New York Household Debt and Credit Report for Q4 2024 placed total US household debt at $18.04 trillion, comprising $12.61T in mortgage balances, $1.21T in credit card balances, $1.64T in auto loan balances, and $1.62T in student loan balances. Federal Reserve G.19 Consumer Credit data showed total non-mortgage consumer credit at $5.12 trillion as of February 2026.

Public vendor claims, taken as the vendor reports them: Symend cites a 10% recovery-rate uplift, an 85% reduction in agent interactions, a 50% reduction in OpEx costs, and 10x ROI across its enterprise client base. Vodex.ai cites a 3x debt recovery improvement and 7x connect rate improvement in its featured case study. These are vendor figures - independent third-party validation across collection portfolios remains limited.

Skit.ai's public homepage in 2026 lists 53,000+ creditors served across 19+ debt types, $1B+ in accounts resolved, and approximately 1 billion conversations processed across deployments. Skit.ai also reports 8 years of live deployments in collections and AI. The company has raised $47.6M to date according to its press materials.

Bridgeforce identifies these as the core credit-union collection KPIs (and they generalize): right-party contact (RPC) rate, promise-to-pay (PTP) rate, kept-promise rate, roll rate / cure rate, cost per dollar collected (CPDC), legal recovery rate, net charge-off rate, recovery rate on charged-off balances, and member experience (post-resolution satisfaction plus complaints per 1,000 contacts). For AI deployments, layer in voice-AI-specific metrics: containment rate, escalation rate, average handle time, and compliance violation rate per 1,000 conversations.

Per NY Fed Q4 2024 data: credit card delinquency (90+ DPD) sits around 7.18%, the highest among major consumer categories. Auto loan 90+ DPD is roughly 4.78%, with subprime auto delinquency at a 15-year high. Mortgage 90+ DPD remains low at about 0.95%. Student loan delinquency reporting resumed in late 2024 after the COVID pause and reflects a sharp spike. BNPL delinquency hovers around 7% at 30+ days per Federal Reserve research.

Federally regulated and FDCPA-compliant outreach mixes voice, SMS, email, letter, and self-serve portals. Public material from TrueAccord and digital-first collectors consistently shows that opening with SMS or email yields the highest open rate, voice handles the negotiation step where dollars are agreed, and self-serve portals deliver the highest customer satisfaction post-resolution. AI voice agents sit between the digital opener and the human negotiator - they do volume reminder work that previously required headcount.

Markets and Markets places the AI debt collection market at $5.8B in 2025, projected to reach $15.9B by 2034 at a 17% CAGR. The broader US debt collection industry generates roughly $21.4B in annual revenue across approximately 7,000 collection agencies (IBIS World, ACA International). The global debt collection market is valued at $29.4B with a projected $38.5B by 2030 (Grand View Research, 4.6% CAGR).

Hear AI Debt Collection Live

Stats are nice. A live demo call is better.

Call our 24/7 demo line and have an AI debt collection conversation in under 60 seconds. No signup, no form, no friction.