How Aussie Broadband’s Fault Detector Project fixes faults before customers notice

Everyone knows how frustrating it is when your internet goes down. But something Australians know all too well is the pain of what comes after – not knowing whether the problem will be truly fixed, or whether you’ll have to reschedule for another appointment.  

But what if your ISP could fix issues so you never had to wait for an NBN technician unless it was totally necessary? What if your ISP could detect that your internet was down or losing speed – and could fix the issue before you even noticed? 

That’s the simple idea behind Aussie Broadband’s Fault Detector Project. Our customer service team always strives to fix issues on the first call. But not everyone has the time in their day to raise a fault, and sometimes folks may not even know there’s an issue.

How the Fault Detector Project began

Our internal analysis found that more than half of all internet dropouts and speed issues – and as many as 80 percent of faults for users on fibre to the node (FTTN) – could be identified before the person noticed. 

As Aussie Broadband continued to grow, we wanted to find a way that could spot all these issues at scale. That was the general idea, and the idea was given to Special Projects Officer Jackson Malley. 

“I got that and went back to my desk and was like, ‘OK, I’ve got to come up with something. Where am I going to get the data from,” Malley recalled. 

After meeting with Aussie Broadband’s network operations team and NBN Co, Malley had a starting point. The combined data provided enough information to begin working on process maps. 

A process map is like a complicated flowchart: it helps lay out the logic for how a system should work. One simple example is when a service is down: the system needs to be smart enough to identify the difference between the service being out, versus a service that’s been paused because the user’s on holiday.  

After spending a few weeks testing the logic out, the process maps were passed onto software engineer Cayde Dixon. Dixon began his career at Aussie Broadband as a level one consultant in the call centre, and that experience proved invaluable in the project’s design. 

“I think it was really good having someone who understood customer service, and how it works,” Malley said.  

The project’s work has become a key part of Aussie Broadband’s strategic cascade, which lays out a series of breakthrough objectives for the business through to FY2025. The goal is for 25 percent of all faults to be automatically identified through future iterations of the Fault Detector Project. 

The real key: understanding the data 

A key part of the project’s design wasn’t necessarily the data analysis, but the interpretation. Take dropouts, for example. NBN Co’s definition of a performance incident for FTTN or HFC connections is when you get 4 to 9 “unexpected dropouts within one calendar day”.  

If your internet only dropped out once a day, that’s not enough to meet NBN Co’s standards for investigation. And that makes sense to a certain degree – if it drops out for two seconds at 1.03am, chances are the user would never notice. But what if a user’s internet repeatedly drops out for, say, 60 minutes a day?  

“That was a frustrating thing for us – [NBN Co] treats a 90 minute dropout the same as a 15 second dropout,” Malley said. 

These were the sorts of cases we wanted the Fault Detector Project to pick up. The system also needed to know the difference between a service that had degraded, or wasn’t delivering the speeds that it should.  

“When you tell a computer is one number less than another, just 0.0001 is enough of a difference for the computer to pick up on,” Dixon explained. “But when you’re talking from a customer service standpoint, you don’t really care if you’re getting 99.99 megabits per second instead of 100 megabits. If you’re getting 80 megabits when you should be getting 100, that’s a problem. So we needed to create a buffer so we could actually say, ‘That’s actually a fault and we can solve this.’”  

But interpreting the figures was only part of the solution. If we couldn’t find a way to efficiently import that data into our system, or into actionable points that our consultants could act on – or chase up with NBN Co – then the project would be a miss.  

So Malley and Dixon designed the Fault Detector Project to show Aussie consultants exactly what the problem is – and the precise steps needed to fix it. “[Staff now] just have to go, ‘Hey, here’s an issue, I’m going to fix it for you,’ instead of having to interpret a bunch of different data to understand what’s going on, then try and work out what the solution might be.”  

And the benefits for customers have already been enormous. In its first 36 weeks, the project has saved staff from having to make around 74,000 unnecessary calls. That’s about 2,400 fewer calls every week, or around 105 hours saved every day. For the Aussie Broadband customer, that’s around 19,000 hours they haven’t had to spend on the phone with us. 

Changing the game for everyone 

One of the project’s biggest successes, according to Malley, has been the ability to surprise our customers. 

“[One example] I saw early on was a customer who was like, ‘I can’t believe you’re this proactive’,” he recalled. “When you come from another telco, you have to initiate the call, you go through all the troubleshooting, they’re absolutely hopeless and then you have that terrible experience.”  

“With Aussie Broadband, it’s like, ‘Hey, we found an issue and here’s what we’re going to do to fix it.’ And customers are blown away; it’s not the standard industry response,” Malley said. 

The project has already been able to fix users’ internet while they’re away. But perhaps one of the most important wins has been to help rectify one of Australia’s biggest frustrations with internet faults: having to reserve days to stay home for an NBN appointment.  

“This is us driving innovation with the NBN,” Malley said. “If we can identify these issues, and it’s not in the customer’s premises, we can tell NBN Co not to involve the customer in the repair.”  

The project’s been a success across the board. The system is raising over 1500 faults a month, and has saved the business around $470,000 in costs in its first 36 weeks. It’s also designed to work across all technology types, whether we’re servicing users on copper internet, users connected on Aussie’s homegrown fibre network, or Aussie customers on alternative networks like OptiComm. 

Continuing to innovate is at the heart of the next iteration of the Fault Detector Project. Our developers are already working on incorporating more data types at greater scale. Not only will it better monitor the degradation of a service – like when your internet’s suddenly worse after a few days rain – but the next version will be able to help resolve some customers’ poor Wi-Fi connections.  

The next version of the project will also become completely autonomous for certain types of faults, freeing up Aussie staffers to concentrate on the bigger problems that need a human touch. The software has also been built to scale across Aussie’s organic and white label customer base, so someone on the most basic, entry-level broadband plan will get the same benefits from the project as a company on an enterprise-grade service. 

The Fault Detector Project’s ingenuity has already been recognised by the telco industry, winning an ACOMMS award for the Best Innovation from a Large Company. But our goal here isn’t to win awards – it’s to Change the Game in telco for everyone.