Catching fraud before it lands
Sector Payments and online marketplace SME.
The situation
Chargebacks and scam losses were climbing, and the old fixed rules either missed new patterns or blocked good customers. A tiny risk team could not keep up with the volume.
What CAPIO did
We built an anomaly detection layer over their transaction data that scores payments in real time and flags the suspicious ones. Every new blocking rule is proposed by the system and approved by a human before it goes live, so the team stays in control. Models run over their existing payment signals, with a review queue and a full log of what changed and who approved it.
The outcome
- Fraud and chargeback losses down by more than a third *representative*
- Fewer good customers wrongly blocked, because scoring replaced blunt rules *representative*
- A small team able to supervise the system rather than chase every alert *representative*
It flags, we decide. That balance is the whole reason we trust it.
- Turning AI activity into a governed programme
- Compliance and competence as the growth engine
- Seeing the whole field, not five percent of it
- An expert made available at scale
- The support desk that answers itself
- An expert on tap for a small firm
- Fluency across a whole small company
- Contracts read in seconds
- More jobs per van, fewer miles
- Cutting the energy bill on what you already run
- A small dev team that ships like a bigger one
- Keeping subscribers who were quietly leaving
- Lifting the whole field, not just the top performers
- Buying at the right moment, not out of habit
Case studies are anonymised. Representative outcomes are benchmarked to independent research and illustrate typical results rather than a specific client figure.