Contents
    1. 1Return on calling 100,000 past borrowers
    2. 2Why past borrowers beat any new lead
    3. 3Your phone team has always a more urgent call to make
    4. 4From your list to new loans in five steps
    5. 5Next steps

    Use case · November 2026 · 4 min read

    Your past borrowers are your best source of new loans

    Emīls Vāvere

    Emīls Vāvereon LinkedIn

    CEO & co-founder · 9 November 2026

    Every lender keeps a list it hardly uses: customers who took a loan, paid it back and said you may contact them again. After a few years it runs to hundreds of thousands of names. These people know your brand, have been through your application and have shown that they repay. They are the people most likely to borrow again, and you already have their number.

    Almost nobody calls them. Once the loan closes, the customer gets a marketing text at best, and the next time they need money, some of them borrow from whichever lender reaches them first. An AI voice agent can call every name on the list and turn it into new loans and extra revenue.

    Return on calling 100,000 past borrowers

    Calling 100,000 past borrowers: about €200,000 of revenue a year against about €15,000 of calls, a first-year ROI of about 1,200% 100,000 past borrowers are each called once. About 30% pick up, so about 30,000 conversations. Revenue chain, left: about 2.2% of conversations end in a new loan, the past-borrower rate in internal data, so about 660 new loans; times an average loan of about €1,500 gives about €990,000 lent; times an 20% revenue a year gives about €200,000 of revenue a year. Cost chain, right: the 30,000 conversations times €0.50 each give about €15,000 for the calls. At the bottom, two bars on one euro scale: revenue, a long orange bar, and the cost of calls, a short grey bar. First-year ROI on the calls is about 1,200%: €200,000 of revenue minus €15,000 of calls, divided by €15,000. 100,000past borrowers, each called once ~30% pick up ~2.2% end in a new loan × ~€1,500 average loan × ~20% revenue a year × €0.50 per conversation ~30,000conversations ~660new loans ~€990,000lent ~€200,000revenue a year ~€15,000for the calls revenue a year cost of the calls First-year ROI: ~1,200%(€200,000 − €15,000) ÷ €15,000
    1. Customers who borrowed from you know your name, so more of them pick up than on a list of strangers.
    2. They have borrowed from you before and know how it works, so they are more likely to say yes.
    3. The agent is billed by the minute, so the seven in ten who don't pick up cost nothing.

    Why past borrowers beat any new lead

    You already know these people, and they know you:

    Past borrowers have already shown they repay, and their record with you says more than any bureau file on a stranger. They also gave you permission to call, so the call comes from a lender they chose and a name they recognise. And they are cheap to win again: you paid for the ad or the lead once, when they first applied, and the next loan costs a few minutes of calls. Winning a new customer costs five to 25 times as much as keeping one (Harvard Business Review, 2014).

    New loans per picked-up call, by list: past and current borrowers take a new loan far more often than refused or rejected applicants Horizontal bar chart of the share of picked-up calls that ended in a new loan, by who was called, in Convershake internal data. Past borrowers: about 2.2%. Current borrowers: about 1.4%. Refused applicants: about 0.24%. Rejected applicants: about 0.13%. The two borrower bars are orange, the two applicant bars grey. Past borrowers ~2.2% Current borrowers ~1.4% Refused applicants ~0.24% Rejected applicants ~0.13% Conversion rate per picked-up call · Convershake internal data

    Your phone team has always a more urgent call to make

    Your phone team's day is already full: customers behind on payments, applicants waiting for a decision, customers calling with questions. Each of those calls protects money you have lent or closes a loan someone wants today, so they come first. Every call an agent makes is a call they can't make somewhere else. An hour on past borrowers is an hour of late payments left unchased and applicants left waiting, so the list never reaches the top.

    An AI voice agent is extra capacity on top of your team. It calls all 100,000 past borrowers while your agents spend the same days on late payments, applications and customer questions. Nobody has to decide which calls to drop.

    From your list to new loans in five steps

    1. Pick the list. Customers who repaid and agreed to marketing calls.
    2. Choose the offer. It is optional. I'd add one, short and concrete: a limit that is already approved, a lower fee than last time, money the same day.
    3. Agree the script. What the agent says, when it makes the offer and when a call goes to your team.
    4. Send the list.
    5. The agent calls. It asks about the customer's plans first, then makes the offer and is as persistent as you want it to be. A yes gets a link to apply.
    A sample returning-customer call, from the thank-you on: a question about plans, a ready limit with a lower fee, a question about the fee, and a link by text The middle of a call from the lender's AI agent to Laura, a returning customer, 0:08 to 0:45. Drawn as a timeline: a vertical time axis runs down the middle with a waveform that leans toward whoever is speaking; the agent's turns sit on the left, Laura's on the right, each at its moment in the call. Agent (0:08): Thanks for paying off your last loan with us, Laura. Is anything bigger coming up in the next few months? Laura (0:14): Actually, yes. The car needs winter tyres, and this month is already tight. Agent (0:20): As a returning customer, you can borrow up to €1,500, with a lower fee than last time. Apply today and the money arrives today. Laura (0:31): How much lower is the fee? Agent (0:34): You’ll see your exact fee in the application before you sign. It takes about two minutes. Shall I send you the link? Laura (0:43): Yes, please. Agent (0:45): You’ll get it by text in a moment. Good luck with the tyres, Laura. The offer, up to €1,500, is highlighted. Lender AI Agent Agent Laura Returning customer 0:08 Thanks for paying off your last loan with us, Laura. Is anything bigger coming up in the next few months? 0:14 Actually, yes. The car needs winter tyres, and this month is already tight. 0:20 As a returning customer, you can borrow up to €1,500, with a lower fee than last time. Apply today and the money arrives today. 0:31 How much lower is the fee? 0:34 You’ll see your exact fee in the application before you sign. It takes about two minutes. Shall I send you the link? 0:43 Yes, please. 0:45 You’ll get it by text in a moment. Good luck with the tyres, Laura.

    Questions get answered on the call, and anyone who asks for a person goes straight to your team.

    When the call comes matters as much as what it says. Three moments work best:

    • Right after the last repayment, while a good experience with you is fresh.
    • A few weeks before Christmas, the summer holidays and the start of the school year, when households plan bigger spending.
    • Just before payday, when the month's money runs short.

    An AI voice agent runs as many calls at once as the list needs, so every name gets a call within a few days of each moment, while people are most likely to borrow.

    Every number on the list comes back tagged with its result:

    The five results a call can end in One row of five results: Interested (highlighted), Not interested, Wrong number, Call back later, Wants a person. Interested Not interested Wrong number Call back later Wants a person

    Interested customers apply on their own through the link. You count the new loans against the list.

    Next steps

    You already hold a list of customers who borrowed from you, repaid and agreed to hear from you. An AI voice agent calls every name on it, and the yeses become new loans and revenue you are not earning today.

    Frequently asked questions

    What if our list converts worse than that?

    The margin is wide enough that a weaker list most likely still pays off, even after funding costs and credit losses. A pilot on part of your list shows your own numbers before you commit to more.

    How do we know the calls brought the loans?

    Leave a random part of the list uncalled. Compare the share of each group that takes a new loan. The difference, times the number of people you called, is what the calls added.