AI-powered revenue recovery turns late payments into loyal customers

Scott Nadzan is Chief Revenue Officer at KredosAi, where he leads go-to-market strategy across sales and marketing to drive growth for the company's agentic AI platform for collections and revenue recovery. He previously led global go-to-market strategy and customer success at Panopto, following his tenure as CEO of Ensemble Video, which Panopto acquired. Throughout his career, he has focused on building digital solutions that engage and educate audiences, converting that engagement into lasting relationships. Outside of his executive roles, he has spent over 20 years mentoring undergraduates at Syracuse University.

Almost everyone has a subscription, a loan, or a monthly bill sitting somewhere with an expired card attached to it. A card gets replaced, a bank account changes, autopay quietly breaks in the background, and none of that means someone stopped wanting the service.

How a business handles that moment shapes whether the customer sticks around. A quick, considerate fix and the customer barely notices. A clumsy or aggressive one, and a small gap can turn into a canceled account.

Household debt and delinquency are climbing to record levels

U.S. household debt hit $18.8 trillion in the first quarter of 2026, according to the Federal Reserve Bank of New York. Auto loan delinquency reached a series record of 5.6 percent that quarter, surpassing the peak of the 2008 financial crisis. The share of consumers with a third-party collection account on their credit file rose to 5.0 percent over the same period.

Translated into plain terms: more customers will fall behind on payments this year than last year, and the accounts that fall behind are, on average, harder to bring current. That pressure is already showing up on collections books across the industry.

The Consumer Financial Protection Bureau received about 387,400 debt collection complaints in 2025, an 86 percent increase over the year before. Debt collection is now the second most complained-about financial product in the country, behind only credit reporting. Complaints at that scale tend to come from customers who felt harassed, confused, or dismissed by the process meant to help them get current. Customers who feel that way are less likely to renew or return the following month, a cost that rarely shows up on a recovery-rate report.

AI-powered revenue recovery is the better term for this work

This category has long been called collections. That's what it was originally built to do. But the term carries associations that no longer match the work: real-time personalization, channel matching, and outreach that adapts as a customer's situation changes.

AI-powered revenue recovery is a more accurate name. It isn't yet an official industry category. No analyst report has that exact phrase on the cover. But it describes the goal more precisely than "AI-powered collections" does. Where the industry lands on the name and what belongs inside it are still open questions.

An AI collections platform that automates an existing dunning schedule is still automation layered on an old process. AI-powered revenue recovery works differently: it reads a customer’s payment history, channel preferences, and behavior, then adjusts which message goes out, when, and through which channel, continuously updating as new signals arrive. Debt collection automation software built this way can respond to changes in real time rather than waiting for the next scheduled review.

Pay and stay means recovering revenue without losing the customer

A simple way to summarize this approach: pay and stay.

Getting someone to pay is the easier half of the equation. Keeping them as a customer afterward is what actually protects revenue over time. Acquiring a subscriber costs real money, in sales time, marketing spend, onboarding, and support. That investment is at risk whenever a collections experience damages the relationship enough to cause churn. Software built for AI-powered revenue recovery is designed to protect that relationship, not only the individual payment.

Digital-first, experience-focused collections strategies tend to produce stronger retention outcomes than traditional dunning schedules. A late notice that feels considerate is far less likely to end a subscription than one that feels punitive, and that difference compounds across a customer base over time.

Consider a streaming service, a gym membership, or a phone plan. A late notice that feels considerate tends to get forgiven. One that feels like pressure tends to send people looking elsewhere. Technology can help identify the right channel and approach for a given customer. It cannot set a company’s risk tolerance or recognize when a hardship case calls for a person instead of a message. That judgment stays with the team managing the account. The technology’s role is to ensure that time and the customer’s patience are directed toward the right approach.

Reduced delinquency compounds into long-term revenue growth

Exact client figures are not shared publicly, but the math is illustrative enough to walk through.

A reduction in delinquency rarely stays a one-time gain. It tends to compound, since fewer accounts fall behind each cycle, fewer relationships get strained, and more customers remain active rather than churning entirely. Over time, the greater opportunity comes from the customer relationship that continues to generate revenue, rather than from a single recovered payment.

These effects tend to build on each other: fewer wasted contact attempts, faster self-service resolution, and outreach that matches how someone actually wants to be reached, rather than one approach applied broadly. Finance and customer experience leaders both have a stake in this, not only the teams running the collections queue. AI debt collection software functions as much as a growth lever as an operational one.

AI agents bring scale to revenue recovery

The next stage of this shift is already taking shape as AI agents. Where earlier automation decided when and how to reach someone, an agent can carry the conversation once contact is made, maintaining context across the interaction instead of resetting with every message. That means fewer dead ends where a customer's question falls outside a scripted flow and gets pushed into a queue, and more moments where a payment plan, a balance question, or a simple mix-up gets resolved on the spot. Agents still recognize when a conversation calls for a person instead of a bot, but the range of what can be handled without that handoff keeps expanding.

Voice agents can now carry hundreds of collections calls at once

Voice is where that scale becomes most visible. Instead of a call center working through a list, one number at a time, a voice agent can handle hundreds of live conversations at once, trained on the kinds of replies customers actually give rather than on a rigid script. 

In financial services, an insurance carrier's voice agent can walk a policyholder through a lapsed premium and reinstate coverage without a hold queue. 

In telecom, a voice agent can catch a subscriber before an autopay failure turns into a service suspension, working through the same objections a live rep would field on any given day. 

In automotive lending, a voice agent can handle a missed installment call with the same patience on the hundredth conversation as on the first, and still recognize when a hardship case needs to be escalated to a person rather than continuing the script.

The long-term goal is recovering revenue without losing the relationship

Defining what AI-powered revenue recovery means as an industry is still early-stage work. Ongoing research is examining the future of customer engagement in collections, the business impact of rising consumer debt and delinquency, and how enterprises can recover revenue without damaging the relationships they have invested in building.

Household debt is not shrinking, and complaint volumes are not slowing down. The businesses that handle this well will likely be the ones whose customers barely notice the process, get back on track, and keep paying month after month.

Pay and stay is the strategy, in three words.

FAQ

How can AI help the revenue cycle?

AI shortens the time revenue remains at risk before it's recovered or written off. It identifies which accounts are likely to respond, selects the channel and timing most likely to work, and adjusts as behavior changes rather than following a fixed schedule. That reduces manual review time and cuts the number of payments that end in write-off.

If a business has no lending exposure at all, why should it still care about rising consumer debt?

Every subscription or SaaS payment competes with rent, auto loans, and credit cards for the same household budget. U.S. household debt reached $18.8 trillion in the first quarter of 2026, according to the Federal Reserve Bank of New York. When that budget tightens, subscription businesses feel it through late payments and churn, even without extending credit themselves.

Is a spike in CFPB complaints a customer service problem or a revenue problem?

Both. A rise in complaints signals that the collections process feels harassing or confusing, which is a service issue. But each complaint also tends to correlate with a canceled account, which turns that service issue into lost revenue. Reading it as only a compliance metric misses the financial impact behind it.

What makes AI-powered revenue recovery different from AI collections software that only automates reminders?

Fixed-schedule automation sends the same reminder to every account regardless of behavior. AI-powered revenue recovery adjusts message content, timing, and channel based on how each customer responds, updating continuously. That distinction decides whether the software reduces friction or just speeds it up.

Does adopting AI-powered revenue recovery replace the people running collections?

It shifts what people spend their time on rather than removing them. Judgment calls like setting risk tolerance or recognizing a hardship case still belong to the team. AI-powered revenue recovery routes the repetitive, high-volume decisions to software so people can focus on those calls.

How does personalized timing and channel selection reduce customer churn after a missed payment?

Customers respond differently to a text than a call, and differently at nine in the morning than at nine at night. Personalized timing and channel selection get outreach to land when someone is likely to engage, rather than whenever it's next on a schedule. Fewer intrusive contacts often decide whether a customer stays subscribed.