Autofine
Recovering unpaid fines, starting from those most likely to pay
Autofine is a system for recovering uncollected public transport fines. A machine learning model identifies the offenders most likely to pay; the system verifies their identity and immediately sends a reminder letter.
+80%
more fines recovered than a random campaign (average goal)
4
steps, from selection to the reminder letter
ML
a machine learning model decides where to start

The challenge
A significant share of the fines issued on board is never paid. Traditional reminder campaigns write to everyone, or to a random sample: they cost a lot and recover little.
- Scope
- Recovery of unpaid fines
- Technology
- Machine learning, automated letter generation
- Partners
- Postel and local public transport companies
- Goal
- On average over 80% more fines recovered than a random campaign
How it works
- 01
Selection
A machine learning model picks out the fines most likely to be paid, so the work starts with those.
- 02
Identity check
For each selected fine we confirm who the offender is before any letter is sent.
- 03
Reminder letter
Letters are generated automatically and sent through the agreement with Postel.
- 04
Agreements with operators
The process is shared with the local public transport companies that issued the fines.
What it offers
Data-driven priorities
Time and money go to the fines most likely to be collected.
Automated process
From selection to dispatch, letters go out with no manual handling.
An integrated chain
Model, checks, mailing and operator agreements in a single flow.
The result
The goal is to recover on average over 80% more fines than a randomly run campaign, focusing cost and effort where the return is highest.
Other projects in this area
Give your data real value
Tell us where you're starting from: we'll reply within one working day.
