How a Georgia Lender Turned 43,000 Past Customers Into Transfers

A Georgia lender had 43,000 past customers it wasn't talking to. It gave them one reason to come back, a pre-approved loan of up to $1,000, and had an AI voice agent call them.

In the first 14 days, that was 13,159 calls and about 100 live transfers, across 10 branches. Each transfer rang at the customer's own local branch while it was open.

Today it's 224 live transfers and counting, and 28 branches instead of 10.

That's the short version. Here's how it happened, what it took, and the one number from it that you can hold up against your own list.

Before: 43,000 customers and nobody calling them

The lender is Sunset Finance, a network of consumer lending branches in Georgia.

Like most installment lenders, it had a list that nobody was working. 43,000 people who had already borrowed once. People who had filled out the paperwork, made the payments, and walked out the door.

Think about what that list is. These aren't strangers. Sunset already paid to find every one of them. They know the branch. The branch knows them.

And the offer was already there. A pre-approved loan of up to $1,000. Nothing to invent, nothing to test in a lab.

So why was the list sitting still?

Two reasons, and you probably know both.

Mail was too slow. You send a piece, you wait weeks, and you still don't know who read it. Nobody can answer a question on a postcard. And every round costs money before you learn anything.

The branches had no time to dial. Picture a branch on a Tuesday afternoon. Someone at the counter making a payment. A walk-in asking about a renewal. The manager on a collections list. The phone ringing. Now hand that team 43,000 numbers and ask them to start calling. It's not a question of effort. The hours don't exist.

So the list waited. The way these lists usually do.

The decision: start with a pilot, not the whole network

Sunset didn't turn this on everywhere at once. It started with a pilot on 10 branches.

That's worth pausing on, because it's where a lot of lenders get stuck. The instinct is to plan the full rollout first. Every branch, every region, every edge case. Months go by in meetings, and the list gets older.

A pilot flips that. You pick a few branches, run it, and look at real numbers. Then you decide whether to keep going.

If you're wondering how the whole thing runs once it's switched on, we walk through it step by step in how PTS AI works inside a branch network.

What PTS AI took off Sunset's plate

Here's the part most people skip when they talk about AI calling.

The voice is the easy part now. Plenty of companies can make an AI say words on a phone. What stops lenders is everything around the call. Somebody has to own it.

For Sunset, PTS AI owned it. This is what the branches didn't have to do:

• Call every number, not a sample. All 43,000 past customers, not the easy ones.
• Call only during branch hours. A transfer only works if someone is there to pick up.
• Route each customer to their own branch. The one in their town. The one they'd walk into.
• Call the no-answers again. Plenty of people don't pick up the first time. Somebody has to keep track.
• Log every call automatically. Nobody at the branch typing notes after a call.
• Report by branch. So Sunset could see which branches, which hours and which customers were producing transfers.

None of that is glamorous. It's also exactly the work a mail vendor can't do and a busy branch won't get to.

The agent itself is called Tommy. First, Tommy calls the customer and explains the offer. Then Tommy answers their questions and finds out whether they're interested. Finally, if they are, the call goes live to their own local branch, and a person there takes it from that moment.

The branch staff only picked up the phone when there was an interested borrower on the other end.

If you want the full playbook for this kind of campaign, it's in how to reactivate paid-off borrowers without a mail campaign.

The pilot numbers

Here's what the first two weeks produced.

About 100 live transfers in two weeks. Each one was a past customer who heard the offer, said yes, and got handed to a person at their branch while they still cared.

Not a callback list. Not a "we'll reach out tomorrow." A live conversation, that day.

It's worth being clear about what we're not claiming. We're not publishing funded loans, revenue or conversion rates from this program. Those numbers belong to Sunset, and they depend on what happens inside the branch after the transfer. What we can show is the part we're responsible for: calls placed and conversations delivered.

Where it is now

The pilot gave Sunset a real number to decide on. It decided to keep going.

Since then it's 224 live transfers and counting. And the program has grown from the 10 branches in the pilot to 28 branches now.

Something else changed too. Sunset now knows which branches, which hours and which customers produce transfers. That's the thing mail never gave them. Instead of guessing where to spend, they can look at a board by branch.

The one number that travels: 8 per 1,000

If you run a lender, the 43,000 doesn't matter much to you. Your list is a different size. The 28 branches don't matter either. Your network is its own thing.

The number that matters is the rate. In the pilot, about 8 live transfers for every 1,000 calls placed.

Here's why that's the only number you can carry over.

Every other figure in this story depends on Sunset's list, Sunset's offer and Sunset's branches. The rate is different. It's a ratio. You can put it next to your own list and do the arithmetic yourself.

First, count the past customers you have with a working phone number. Then think about how many calls it takes to work that list, including calling the no-answers again. Finally, apply roughly 8 transfers per 1,000 calls. That gives you a rough idea of how many borrowers could be talking to your branches.

Two honest warnings.

It's a starting point, not a promise. Your list might be older or newer. Your offer might be stronger or weaker. Your market is yours. The rate will move.

Transfers aren't loans. What a transfer is worth to you depends on your own funded-loan value and on how well your branch handles a warm call. That part is yours to measure. The case study page will run that math with your own two numbers if you send them.

There's a reason this kind of list is worth the arithmetic at all. Harvard Business Review has written about how much cheaper it is to keep an existing customer than to find a new one, citing estimates of five to 25 times. Your paid-off borrowers are about as existing as customers get.

Why waiting costs more than it looks

Here's the uncomfortable part.

Your past customers aren't only on your list.

Lenders can ask the credit bureaus for lists of people who meet their criteria, based on what's in their credit reports, like borrowing and payment history. The FTC explains how these prescreened offers work. Someone who just finished paying off a loan on time can have exactly the kind of history those criteria are built on.

So while your list waits, your paid-off borrower can be hearing from lenders they've never met.

And the list doesn't stay still. The borrower who paid off last month remembers your branch, your staff, the way you treated them. The one who paid off two years ago remembers less. Every month the list sits, it gets a little colder.

None of that shows up on a report. It just quietly moves the next loan somewhere else.

Sunset had 43,000 people like that. It decided to call them first. If you want to see what week one of that looks like for your own branch, read what the first week of an AI calling install looks like.

If you'd rather hear what your borrowers would hear, the AI will call you. Leave your number.