Qld tax office avoids 'creepy' facial and voice tech on path to intelligent systems

Queensland’s Office of State Revenue (OSR) has left behind technology that might be deemed “creepy” by clients during its three-year shift to next generation tax IT systems.

Speaking at the Gartner IT Symposium/Xpo last week, deputy commissioner Simon McKee said the agency had deliberately chosen such an approach to ensure the program’s client-centric focus.

“They don’t want gold plated solutions or for us to be creepy - that is, creepy with their data,” he said.

“When we asked them, [they said] they don't want facial recognition, voice recognition at the moment until they’re more comfortable with that, so we’ve taken it off our roadmap for now.”

“All they want to do is to interact with us as easily, seamlessly and securely as possible on devices at their time or choosing, ideally through their natural business processes.”

McKee said the decision was a result of the systems redesign process the agency embarked on under its digital transformation program more than two-and-a-half years ago.

As a division of Queensland Treasury, OSR is concerned with managing approximately $15 billion of revenue raised through state tax, mining and petroleum royalties each year.

“In the past, we thought we were subject matter experts. We designed our digital solutions and processes around this and then we trained our staff [and] trained our taxpayers or clients,” he said.

“Under the program, we completely flipped that approach… We ran a series of process sprints - five sprints over eight weeks.

“We brought in taxpayers, industry staff to understand what matter to them. Not what they wanted, but what they need. A subtle, but important difference.

“We wanted their experience to be the best it could be. We then adjust our digital roadmap and redesign processes through digital ends, always with the client at the center.”

The transformation program, which has now entered the final stretch, has taken an agile approach to the delivery of 46 projects that vary in “size, complexity and risk”.

McKee said the agency had now delivered 31 of those, with another eight  “in flight” and a further seven yet to be delivered.

“Rather than potentially waiting years to realise any benefits, we can deploy these products and solutions relatively quickly. It also de-risks the program; that is the overall program,” he said.

“For example, if one project or product were to final, the others move on and of course there are exceptions where you have dependencies or co-dependencies, such as our platform as a service renewal.

"We replatformed and shifted our PaaS arrangement to a secure data center in Canberra with [Microsoft] Azure and SAP."

One of those projects saw SRO introduce machine learning, which has been a major point of focus during the shift to “next generation tax revenue management capabilities”.

Having introduced SAP Leonardo machine learning technology last year, the agency is now able to predict when taxpayers are at risk of defaulting to secure the state’s revenue base.

“We're able to predict taxpayers at risk of default with greater than 71 percent accuracy and, as we add more external data, we expect that to increase,” he said.

The technology has allowed OSR to introduce “personalised payment plans” to help clients meet their financial obligations. This, in turn, allows the agency to “deliver more timely revenue collection” to the government.

It has also allowed OSR to “implement targeted campaigns to reduce debt levels by at least 5 percent”, which McKee described as “significant” when looking at aggregate debt.

“That provides near real-time insights, the taxpayers historical journeys and predictions on the future behaviours,” he said.

“We've now expand the machine learning applications to other revenue lines to address revenue line specific challenges and opportunities.”

The agency is also looking at the factors behind why clients aren’t able to pay their tax on time using the machine learning data with SAP Qualtrics.

“These insights are being delivered by an SAP analytics cloud dashboard to be easily consumed by staff, so they can take the necessary action,” he said.

“And we're going to make that available right across your organisation, including to the Under Treasurer, who can interrogate the data in real-time across all our data holdings.”

OSR is similarly working with SAP to “eliminate bias” with the machine learning solution. 

However, McKee said that this relied on the agency’s strong data foundations through a single HANA enterprise data warehouse and SAP data hub to control the flow of data.

OSR is now looking to extend the agile approach take during the transformation across the remaining areas of the 520-strong agency, which was a benefit of being “relatively small”.

“We've embedded agile in our enterprise systems division, in our transformation program and now we're moving that right across the office,” he said.

“And we're going to make that available right across your organisation, including to the Under Treasurer, who can interrogate the data in real-time across all our data holdings.”

McKee said that although the program had cost $80 million over three years, it would pay for itself several times over thanks the efficiencies introduced. 

“We’re going to do a couple of hundred million back and other efficiencies, and the important thing is delivering those improved digital services to our taxpayers,” he said.

Justin Hendry attended Gartner IT Symposium/Xpo on the Gold Coast as a guest of Gartner