Every industry that extends credit, banks, hospitals, utilities, telecom carriers, and retailers, is fighting the same losing battle behind the scenes. Their collections operations, the teams responsible for recovering unpaid balances, are still running largely on phone trees, spreadsheets, and agents dialling down a list. And the cracks are showing in the numbers.

This isn’t a niche back-office problem. It’s a cash-flow, compliance, and customer-retention problem that touches nearly every consumer-facing business. Here’s what the current data reveals about the severity of the issue, and what a more effective, automated approach actually achieves.

A Mountain of Debt, and a System Not Built to Handle It

The scale of the problem starts with the sheer size of outstanding consumer debt.

• Australia: record $3.33 trillion household debt (ABS, June 2025), ~114% debt-to-GDP, plus the regulatory angle, ASIC naming debt collection misconduct a 2025 enforcement priority and the credit-licence/AFCA membership requirement since 2021, which strengthens the “manual = compliance risk” argument.

• Europe: EU/euro area debt-to-GDP trending down (49.4%/50.7%, Eurostat/ECB) and NPLs falling industry-wide, but I flagged the fragmented, country-by-country collections rules across the EU, since that’s actually a strong point in NovaCollect’s favour (a multilingual, standardised automated approach is more valuable specifically because Europe’s manual, jurisdiction-by-jurisdiction processes are hard to scale).

• United States Total U.S. household debt reached approximately $18.4 trillion by mid-2025, comprising a range of debt types, including credit card balances, auto loans, and student debt. Credit card balances alone have become the fastest-growing category, expanding at close to 15% annually. Delinquency is rising alongside it: New York Fed data shows that around 4.4% of all outstanding debt is now in some stage of delinquency, with consumer loan delinquency at its highest point since 2012.

More accounts falling behind means more volume flowing into collections queues that, in most organisations, are still worked one call, one email, and one spreadsheet update at a time.

Manual Collections Is Quietly Breaking Under the Weight

Three things are converging to make traditional, manual collection operations far less effective than they used to be:

1. Customers have stopped picking up. Answer rates for calls from unknown numbers have fallen to below 15% in many studies, some as low as 8-11%, down from roughly 60% just a few years ago. Decades of robocall spam (an estimated 50+ billion unwanted calls hit U.S. phones annually) have trained consumers to screen and ignore anything that isn’t a saved contact. Analysts have started calling this the “digital trust deficit” and it guts the effectiveness of a phone-first, manual outreach model.
2. The old playbook is generating complaints, not payments. When call-heavy, one-size-fits-all outreach doesn’t get through, agencies often respond by dialling harder and more often, which backfires. Complaints about aggressive debt collection practices reportedly more than tripled year-over-year in a recent quarter, climbing past 140,000 from roughly 44,000 the year before. That’s not just reputational damage; escalating friction is frequently the on-ramp to the most expensive resolution path of all: litigation.
3. It’s operationally expensive by design. Manual collections lean on large call centres, high headcount, and agents repeating the same routine tasks, payment reminders, balance lookups, status checks, over and over. Industry research on contact centre operations puts the achievable savings from automating this kind of repetitive work at 25-50% of operational cost, alongside a comparable drop in average call handling time. Quality assurance is similarly stuck in the past: most manual QA programs can only review 1-3% of total customer interactions, meaning the vast majority of collector conversations, good or bad, go completely unchecked.

The Customer Satisfaction Cost Nobody Puts on the P&L

Collections is the one department where a bad experience does double damage: it doesn’t just lose the relationship, it also makes the money harder to collect.

• Slow, inconsistent manual outreach means longer resolution times, which directly correlates with lower first-call resolution, a metric tightly linked to both recovery rates and debtor satisfaction.
• Generic, poorly timed, or repetitive contact attempts are a leading driver of the complaint surge noted above, and complaints translate into churn, negative reviews, and regulatory scrutiny.
• Nearly all customer service leaders, 97% in one recent industry survey, now say that AI-driven, conversational engagement has a measurably positive effect on customer satisfaction scores, precisely because it replaces generic scripts with faster, more relevant, self-service-first interactions.

In other words: the businesses still relying on manual, high-friction collections aren’t just paying more to run the operation;  they’re actively damaging the customer relationships they’re trying to preserve.

Where the Industry Is Actually Headed

The market has already started voting with its budget. The global debt collection software market was valued at roughly $6 billion in 2025 and is projected to more than double by the early 2030s, growing at nearly 10% a year, as finance, healthcare, utility, and telecom organisations shift spend from headcount to automation. AI adoption among mid-to-large collection operations reportedly jumped from about a third in 2022 to roughly two-thirds by 2025, but most of those early adopters are still only automating one or two narrow tasks, not the whole workflow.

That gap, between organisations that have automated a task or two and organisations that have automated the entire collections motion, is exactly where the next competitive advantage lives.

This Is the Problem NovaCollectAI Was Built to Solve

NovaCollectAI, part of the Nova platform from DigiU, takes a different starting point than most collections tools on the market: instead of digitising the manual process, it replaces the manual process.

Where legacy collections still depend on agents making outbound calls one at a time, working from static lists, and re-explaining the same account details in every language a customer happens to speak, NovaCollectAI is built for autonomous, multilingual debt collection, engaging customers directly, at scale, without a human having to manually initiate or manage each conversation.

That matters because it addresses the three failure points above head-on:

• It meets customers where they’ll actually respond. Instead of relying on a single outbound phone call, the channel with the lowest answer rates in the industry right now, an autonomous system can engage across channels and follow up automatically, rather than losing the account the moment one call goes unanswered.
• It removes language as a barrier to resolution. Multilingual, automated engagement means a customer can resolve an account in the language they’re most comfortable in, without waiting for a specific agent to be available, a common bottleneck in manual operations with limited multilingual staffing.
• It scales without scaling headcount. Because outreach, reminders, and routine account questions are handled autonomously, collections teams can work larger portfolios dramatically without proportionally growing the call centre, directly addressing the cost structure that makes manual collections so expensive.
• Every interaction is consistent and auditable. Unlike manual QA, which industry-wide only reviews a small fraction of interactions, an automated system generates a complete, reviewable record of every customer engagement by default, turning compliance from a sampling exercise into full coverage.

Combined with NovaContact, DigiU’s AI-powered voice and contact centre product, the Nova platform lets organisations run collections and broader customer contact through the same intelligent infrastructure, rather than stitching together a dialler, a CRM, a compliance tool, and a reporting spreadsheet.

What Changes When Collections Stop Being Manual

Put the industry data next to what automation is already delivering in adjacent contact centre environments, and the shift NovaCollectAI represents becomes clear:

The Bottom Line

The state of collections in 2026 is a story of two widening gaps: the gap between rising delinquency and shrinking answer rates, and the gap between organisations still running manual, complaint-generating processes and those that have moved to autonomous, customer-first engagement.

Manual collections aren’t just slower and more expensive, it’s actively working against the customer relationships and recovery rates it’s supposed to protect. NovaCollectAI was built to close that gap: replacing single-channel, headcount-limited outreach with autonomous, multilingual engagement that scales, stays compliant by default, and treats the customer relationship as worth protecting, not just chasing.