01
High volume
Many accounts per collector, with little time for each conversation.
Success story · Sercom
Sercom added Saldia to a retail card portfolio, across both early- and late-stage buckets. In two weeks the agent closed 295 payment promises for $49.5M: first across the whole portfolio, ahead of every human collector.
Who they are
Sercom is a collections agency with more than ten years in the market. They didn't come to Saldia looking for new technology: they came with a volume problem and the operational judgment to tell whether the tool was any good.
That experience ended up shaping the product. Sercom helped design the solution and was the first case in production: the negotiating policy, the delinquency buckets and the hand-off criteria came out of how a real operation works, not out of a hypothesis of ours.
The challenge
The portfolio is retail card (anonymized) and covers two buckets: early- and late-stage delinquency. It's the kind of portfolio where human collections gets expensive — many low-balance accounts, where each contact costs the same as on a large account but recovers far less.
01
Many accounts per collector, with little time for each conversation.
02
The cost per contact doesn't drop, but the amount to recover does.
03
Human collections ends when the shift ends; the debtor answers when they can.
What they deployed
The agent negotiates over voice, chat and WhatsApp 24 hours a day, on 30 dedicated channels. It proposes plans within the portfolio's policy, closes the payment promise and logs it — and hands off to a human collector when the case calls for it.
The same agent on all three channels, with the same policy.
No gaps for shifts or weekends.
Covers both buckets of the portfolio, with different rules in each.
Results
Pilot from September 1st to 14th. The numbers compare Saldia against the contact center's average collector, on the same portfolio.
295
bot payment promises
#1 by count across the whole portfolio.
$49.5M
in promises
#1 by amount across the whole portfolio.
4.6×
more promises
295 vs. 64 for the average collector.
3.2×
more value
$49.5M vs. $15.7M for the average collector.
Amount promised per collector · collectors anonymized.
How to read these numbers. The bot's average ticket is $167,734 per promise, lower than a collector's. That isn't a weakness: the bot works the smaller-balance buckets, precisely the ones a human operation gets the least out of. It wins on volume, which is exactly what it was put there to do.
What the case shows
The bot didn't top the ranking because it negotiates better than a good collector. It topped it because it works 24 hours, doesn't tire out on account number two hundred, and takes without complaint the part of the portfolio that eats a human's day to recover very little.
That leaves collectors where they do make a difference: the large accounts, the cases with a complaint, the situations that call for judgment. Same team, workload split a different way.
Let's talk
A 30-minute conversation is enough to see whether your operation has a similar case.
Book a call