AI in Debt Recovery: How Machine Learning Finds Debtors and Assets

Debt recovery used to depend on persistence: letters, phone calls and investigators knocking on doors. Persistence still matters, but the firms that recover the most today are the ones that know where to knock first. That is where artificial intelligence has changed the industry.

At Tenaza Asset Recovery & Intelligence, where I serve as CEO and President, we built our own AI model for skip tracing and asset tracing because off-the-shelf tools were not designed for African data, and because our work spans Nigeria and a correspondent network in more than 40 countries.

Skip tracing: finding the person

A debtor who has moved, changed numbers or closed accounts leaves fragments behind: company registrations, court filings, property records, social and professional profiles, phone and address histories. Individually each fragment is weak. Machine learning is good at linking weak signals, matching name variants and spellings, scoring the probability that two records describe the same person and ranking the leads most likely to succeed. Investigators then spend their time on the strongest leads rather than the longest list.

Asset tracing: finding what can be recovered

Locating a debtor is only half the work. Recovery depends on identifying assets that can be attached: property, vehicles, shareholdings, receivables and funds moved to related parties. AI helps map ownership networks, flag transfers that look designed to put assets out of reach and connect entities across jurisdictions. My training as a forensic accountant and Certified Fraud Examiner taught me that money usually tells the truth even when people do not; AI simply reads the trail faster.

Prioritising a portfolio

Banks and lenders rarely hand over one account. They hand over portfolios of thousands of non-performing loans. Predictive scoring lets a recovery firm segment that portfolio: accounts likely to settle with a structured conversation, accounts that need legal action, and accounts where tracing is required before anything else. That changes recovery rates and costs at the same time.

Guardrails matter

Power without discipline creates risk. Responsible recovery means lawful data sources, compliance with data protection law, human review of every decision that affects a person, and professional, respectful engagement with debtors. Tenaza is registered with Nigeria's data protection regulator for that reason.

How a modern recovery workflow runs

1. Portfolio intake. The lender shares account data. The first step is cleaning it: matching identities, correcting addresses and removing duplicates.
2. Enrichment. Models compare the data against permitted public and commercial sources to find current contact details, employment and business interests.
3. Scoring. Each account receives a recoverability score based on contactability, assets, ability to pay and legal position.
4. Strategy by segment. High-score accounts receive negotiation and settlement offers. Accounts with hidden assets go to investigation. Disputed or complex accounts go to legal review.
5. Field and legal action. Investigators verify assets on the ground, and lawyers act where enforcement is justified.
6. Feedback. Outcomes are fed back into the models, so the next portfolio is prioritised better.

What lenders should ask a recovery partner

• Which data sources do you use, and are they lawful in each jurisdiction?
• How do you verify AI findings before acting on them?
• How do you treat vulnerable debtors and handle disputes?
• What reporting will we receive, and how often?
• Can you trace assets across borders, and through which network?

A credible partner answers these questions clearly and in writing.

Why this matters for the wider economy

Non-performing loans tie up capital that banks could lend to businesses and households. Faster, fairer recovery frees that capital, lowers the cost of credit and encourages lenders to serve more customers. Technology does not replace the human judgment involved in recovery, but it makes the process faster, more accurate and more consistent.

The result

AI does not replace investigators, lawyers or negotiators. It gives them better maps. For lenders carrying bad debt, that means faster recoveries, clearer evidence and fewer dead ends. In Tenaza's own words: we grip, we hold, we recover.

Common questions

What is skip tracing?

Skip tracing is the process of locating a person who cannot be reached at their last known address, using public records, commercial data and investigation.

Can AI recover debts on its own?

No. AI helps find people and assets and prioritise accounts, but negotiation, verification and legal action still require experienced professionals.

Is asset tracing legal?

Yes, when it uses lawful data sources and follows data protection, privacy and debt collection rules in each jurisdiction.