Case Study · Consumer finance

This Lender Already Had 43,000 Past Customers. It Just Wasn't Talking to Them.

A network of lending branches in Georgia, a proven offer — a pre-approved loan up to $1,000 — and 43,000 customers who had already borrowed once. The problem was never finding more people. It was getting those customers back into a conversation without burning money on mail.

The offer
You already paid for these customers once. We call every one of them and hand you the ones who want to talk.
Nobody on your team dials a number. The branch only picks up when there's already an interested customer on the line.
13,159calls in 14 days
224live transfers & counting
10 → 28branches (pilot → now)

Real numbers from a pilot that's still running — not a projection.

See How It Works See the Results
The challenge
The problem

A List of Past Customers Isn't Revenue — Until Someone Calls.

Mail is slow and expensive. Branch staff can't dial 43,000 numbers. So the database just sat there — full of customers who already trusted them, going cold.

Mail campaigns Branch staff Excel Customer database CRM
How it works
How it works

Three Moves Between an Idle List and a Borrower Inside the Branch.

The system does the dialing; the branch closes the loan. Every move is there for a reason.

1
Reach every customer
We call and text every past customer with the pre-approved offer — all 43,000, not a sample. We bring the infrastructure to dial at that volume.
So that no customer goes uncalled because the branch ran out of hours.
2
Find the interested
Tommy, our voice agent, has the whole conversation: explains the offer, answers questions, and tells apart who's actually interested from who just answered the phone.
So that branch staff only ever pick up someone who already said they're interested.
3
Transfer to the branch
We hand the call live to that customer's own local branch, while it's open and someone's there to answer. The call gets logged on its own.
So that the conversation lands while the customer still has the phone in their hand.
The stack Twilio Tommy n8nn8n Supabase Written in JSJavaScript JSNode.js PostgreSQL 5HTML5 3CSS3
Built, not bought

There's no CRM on that strip, because the one this program needed didn't exist.

A generic CRM won't hand each branch only its own leads, or record what a call produced the moment it ends. So we built the interface around the program. It shipped with AI assisted development: it gets rebuilt in an afternoon when the cadence changes, not in a quarter.

  • One screen per branch. Each one sees its own and nothing from anyone else's.
  • Outcomes write themselves. Recorded as the call ends, not typed in later.
  • The week, already added up. By day and by branch, with nobody assembling a report.
  • A way to answer back. A branch comments on a transfer, and that changes the next week's list.

The screens below are the real thing. Branch names, staff names and customer data are pixelated: the numbers are ours to show, the identities are not.

The system creates the conversation. The branch creates the loan.

The reporting

Every Number on This Page Comes From a Screen We Built.

This is what the branches and the head office actually open. Not a mockup and not a slide: the product that runs the program.

The pixelated blocks are client data we masked on purpose.
The week at a glanceTransfers, interested follow-ups and voicemails, updated while the cadence runs.
The program, written downOffer, cohort design and the result of every week, so nobody argues about what was agreed.
The work nobody countsTen thousand voicemails in a month, and the hours they would have cost a person.
Outcomes by dayLive transfers, email requests and interested follow-ups, one bar per day.
Branch by branch, day by dayWhere the program is working and where it stalled, without anyone building a report.
What actually ranCalls executed per day, so a gap shows up the same week it happens.
Every call has an endingTransferred, callback, no answer, wrong number: one taxonomy across every branch.

Branch names and customer data are pixelated on purpose. The numbers are real. The identities are not ours to publish.

What we absorb
What we absorb

The Part Nobody Wants to Do.

Their branches were never going to dial 43,000 numbers, and no mail vendor was going to have the conversation. This is the work we took off the table — and the reason the list finally moved.

All 43,000 numbers, not a sample. The whole list enters the dialing queue.
At the right hour, not any hour. Calls go out when the branch is open and someone can actually pick up — not at 7pm.
Per-branch routing, not a shared queue. Each interested customer goes to their own local branch.
A no-answer gets called again. A number that didn't pick up isn't a number that's done.
Every call gets written down. What happened and what's next records itself — nothing depends on anyone's memory.
Each branch sees its own activity. Branch-level reporting, not one total nobody can act on.

None of this is glamorous. That's exactly why it was available — it's the work a client doesn't want to do and a mail vendor can't.

The results
The proof — in the numbers

It Started as a 14-Day Pilot. It's Now 224 Transfers and Counting.

Not a projection. It started with 10 branches — ~100 live transfers in the first two weeks — and now runs across 28 branches, dialing customers the lender already had.

43,000past customers
13,159calls in 14 days
224live transfers & counting
10 → 28branches (pilot → now)
Every transfer is an interested customer Tommy handed live to their branch — not a cold lead. That's 224 live transfers so far across 28 branches, with measurable activity every day.
In the pilot: about 8 live transfers for every 1,000 calls placed.
That's the number that matters, because it's the only one you can apply to your own list. 13,159 calls in 14 days, ~100 live transfers, 10 branches. It's what happened here — not a promise of what happens for you.
What it's worth
What it's worth

Every Transfer Is a Warm Borrower Walking Into a Branch.

These aren't cold leads. They're customers the lender already had — reactivated at a fraction of the cost of mail. Each transfer rebuilds the path to revenue:

Revenue ←Loan funded ←Branch visit ←Live transfer ←Interested ←Tommy's call ←Customer list

From there the lender knows which branches, which hours and which customers produce transfers — and puts the effort there, instead of spending on mail it can't measure.

Run it with your own numbers
1The size of your list. How many past customers you have with a phone number.
2The rate observed here. About 8 live transfers per 1,000 calls placed.
3What a funded loan is worth to you. That number is the one only you have.
Multiply the three and you have what your idle list is likely worth. That's exactly the number we'll compute for you, free, before you commit to anything.
The client
A real operation

A Real Lender. Since 2001.

PTS AI Client
A consumer finance company
Multi-branch consumer lender in the US Southeast · PTS AI client

The client understands multi-branch consumer lending because it has operated in the industry since 2001.

Across dozens of locations in the US Southeast it provides personal installment loans, managing customer relationships at scale.

Its authority comes from real operations — not theory. As a PTS AI client, they provided the operating environment and branch-level expertise to validate AI customer reactivation — expanding the program from an initial pilot to 28 locations.

Your turn
Your turn

Got an Inactive Customer Database?

Tell us how many past customers you have and what a funded loan is worth to you. We run your list against this program's real numbers and hand you back a figure — before you talk rollout, contract or price.

What you get, and what it costs you
✓A figure, not a proposal. What your idle list is likely worth, with the math shown.
✓The assumptions, written down. Which rate we used and where it came from, so you can argue with it.
✓No cost and no commitment. If the figure doesn't justify moving, there's nothing to discuss and we'll say so.
The only thing you risk is the time it takes to send us two numbers. We carry the work of finding out whether it's worth anything.

You already have the customers. We put them back on the phone.

Show Me What It's Worth See the Engine
PTS AI