Johan van der Merwe, Engineering Manager with approximately 12 years at Transaction Junction, shared his insights on the company’s AI transaction monitoring initiative.
Towards the end of 2023, Transaction Junction launched an initiative to enable low-level monitoring on every datapoint in our transactional ecosystem. This meant analysing the response codes of each and every transaction, regardless of the customer or the channel. To accomplish this, we had to overcome several challenges:
1. Efficient and Optimised Monitoring
Our monitoring needed to be precise and effective, but without any negative impact on performance. Given our high transaction volume, efficiency was critical.
2. Determining Expected Trends for Each Channel
With a wide variety of channels, each with its own profile, we needed a way to identify what a “normal” pattern looked like for each one. This was essential for spotting genuine issues without compromising the specific nuances of each channel.
3. Filtering Out False Positives
With so many datapoints being monitored, alert fatigue could easily become an issue. We had to ensure alerts would only trigger for trends indicating a true potential problem.
4. Automatic Detection of New Channels
As our ecosystem grows, new customer channels are continually added. We needed a monitoring system that would automatically detect these new channels and apply the same rigorous, low-level monitoring regardless of the channel’s size or transaction volume.
Given these complexities, it quickly became clear that there was no one-size-fits-all solution. We needed something capable of analysing each datapoint every minute while adjusting dynamically. This was where AI monitoring became our best option.
How 365/24/7 AI-Assisted Monitoring Makes a Difference
Our AI-assisted monitoring system brought new capabilities that fundamentally improved how we manage and assess trends. By analysing a two-year rolling window of transaction data, AI anomaly detection helps us distinguish each channel’s typical patterns, quickly identifying any activity that deviates from the expected. This has allowed us to proactively address issues, keeping disruptions to a minimum.
One of the biggest wins has been a measurable improvement in our transaction success rate (which is already industry leading). Ensuring customers have a smooth and reliable experience. By automatically sending instant alerts only for potential issues, our teams are now better able to focus and act swiftly, reducing Mean Time to Resolution (MTTR) considerably.
Wide-Reaching Monitoring for Every Channel
Another standout feature of our AI-assisted monitoring system is its ability to automatically detect any new channel or customer endpoint, no matter how big or small. This autodetection feature allows our monitoring to scale seamlessly, with the smallest customer channel receiving the same high level of attention as the largest.
The Big Picture: A More Resilient Transactional Environment
By implementing this AI-assisted monitoring approach, we’re not just enhancing reliability but also future-proofing our transactional ecosystem. Whether it’s pinpointing subtle trends, flagging potential issues before they escalate, or consistently monitoring every channel with the same high standards, this system has proven invaluable in helping us stay a step ahead. Smarter, AI-driven monitoring isn’t just a tool; it’s a crucial part of our commitment to keeping transactions flowing smoothly, day in and day out.