AWS Builder Center
Building a Fuel Fraud Audit System with n8n and Real GPS Data

Building a Fuel Fraud Audit System with n8n and Real GPS Data

How I built an automated system that catches fuel fraud in logistics fleets by comparing GPS tracking data against driver trip claims, using n8n, Google Sheets, and Slack alerts.

COMPUTER SCIENTIST
Logistics companies lose real money every month to fuel fraud. Drivers overstate distances, pad trip claims, or report routes that never happened, and by the time anyone notices in the monthly fuel bill, the damage is already done. During my internship at skillIT Agency I got the chance to build a system that catches this in near real time instead of after the fact.
The core idea was simple. GPS devices already track where a vehicle actually goes. Drivers separately submit trip claims through a form. If those two numbers do not match within a reasonable margin, something is wrong. The hard part was building something that could do this comparison automatically, every day, without someone manually cross checking spreadsheets.
How the system works
The system pulls GPS trip data from Traccar and compares it against driver submitted claims collected through Google Forms. Everything gets reconciled inside Google Sheets, which acts as the central database. An automation platform called n8n orchestrates the whole flow, and any trip that comes back with a large variance triggers a Slack alert so the operations team can look into it the same day instead of finding out weeks later.
Each trip gets flagged with a color based on how far the claimed distance is from the GPS recorded distance. Small variances get a green flag and move on. Bigger gaps get amber. Anything that looks like a clear mismatch gets flagged red and goes straight to the team.
A lesson that cost me real debugging time
One bug taught me more than the rest of the project combined. The workflow sends alerts to Slack using an HTTP request. If you do not explicitly set the request method to POST, some automation tools will silently default to GET instead. There is no error message. The request goes through, Slack just does not receive the message the way you expect, and you end up staring at a workflow that looks perfectly fine while nothing actually happens on the other end.
It took a frustrating amount of time to catch because everything else in the workflow looked correct. The fix was one line, explicitly setting the method to POST, but the lesson stuck. When an automation platform lets you skip a setting and quietly picks a default for you, that default is not always the one you want.
Why this matters beyond one project
This was not a demo built to look good in a portfolio. It is a working system built for a real internship deliverable, meant to run against actual fleet data. That changes how you think about edge cases. A demo can ignore messy data. A real audit tool cannot, because the messy data is exactly what you are trying to catch.
If you are a student building automation projects, my advice is to pick something with a real consequence attached to it, even a small one. Bugs behave differently when something is actually depending on the output being correct.
What is next
The reconciliation and alerting pieces are working end to end. Next up is a proper dashboard so the operations team can see trends over time instead of only reacting to individual flagged trips, plus a demo dataset that shows the fraud patterns clearly for training and presentation purposes.
If anyone here has worked with n8n, Traccar, or built similar reconciliation systems, I would like to hear how you approached the edge cases. This is the kind of project where the interesting problems only show up once real data starts flowing through it.
Any opinions in this article are those of the individual author and may not reflect the opinions of AWS.
Enjoyed reading this content? Let the author know!

Your likes, comments, shares, and saves help creators reach more builders.

Loading recommendations

Loading article