At a glance
By Ben Falkenmire
BlueScope Australia’s new finance operations and transformation team has one core goal – to deliver autonomous finance solutions to the business.
The team, helmed by Stephannie Jonovska FCPA, forecasts that it will save BlueScope more than 6500 labour hours in the 2023-2024 financial year alone.
“We haven’t found a way of saying those hours equal this much value – yet,” says Jonovska, who chairs CPA Australia’s Digital Transformation Centre of Excellence.
The team does receive anecdotal return on investment (ROI) feedback. “
For example, we freed up 700 hours in fringe benefits tax analysis, so now the tax team can work with the business on other significant projects,” she explains.
Most projects in the pipeline involve bots or robotic process automation (RPA), but the team also increasingly works in the machine learning space. This may have even greater potential, Jonovska says.
“You get a different level of value from machine learning. The RPA frees up your time. Machine learning gives you that next level of transparency and enhanced business partnering, so why wouldn’t you add the two together?”
Fix mistakes and detect fraud
Research firm McKinsey estimates that up to half the world’s daily activities could be automated by 2055.
Despite recent advances in automation, the accounting profession still has a long way to go.
In 2020, The Institute of Management Accountants and Deloitte surveyed 800 accounting professionals about their processes. The data reveals that 76 per cent of respondents use manual accounting processes, or processes that involve considerable manual effort.
The more manual a process is, the more it is prone to error or vulnerable to fraud. This is why Bryant Richards, associate professor of accounting at Nichols College in the US, believes smart automation is vital for the accounting profession.
“There is significant waste within the financial recording and closing process associated with moving, coding and validating data,” Richards says.
“The financial close process is subject to error. I have been involved in numerous restatements and frauds that were caused by human error or intention. These events could be prevented by autonomous accounting.”
Mitigating fraud and corruption are the main drivers for businesses taking up autonomous accounting projects, says Mohit Sharma, CEO of Sydney-based Mindfields. Sharma has noticed a recent trend toward automation in the businesses to which he consults.
“Internal audit and fraud management become much easier the more autonomous your accounting systems are. Everybody’s doing autonomous accounting for different reasons,” Sharma says.
Address headcount concerns
Some employees may be resistant to automation due to concerns it will be used to reduce headcounts.
“AI is not going to take over your job. People using AI are going to take over your job,” Jonovska counters.
According to Richards, as accountants free up their time by automating manual tasks, they will have greater space to focus on value-added projects that, ordinarily, they would not have time to complete. Within organisations, this could see accountants taking on more of a consultative role.
“Non-financial departments tend to benefit in important ways from receiving counsel and leadership from financial experts. Increasing the service level of the accounting department will add value throughout the organisation while creating an environment that will attract and retain top financial talent,” Richards explains.
Automation best practice
Best practice views automation as a journey, not an implementation step. The first step is to determine readiness.
Jonovska and her team start by focusing on the challenges, which include timeliness, accuracy, “grind” work, lack of transparency, analysis and forecasting needs.
Working with other business units, Jonovska then assesses if a given solution is feasible, in terms of both data and people. She also looks at the solution’s viability, in terms of the hours needed for implementation versus the potential hours it will save.
Automation is not a “plug-and-forget” solution, Jonovska says. Managers and staff need to own the numbers the tools produce.
“This is not a black box, but a glass box. People need to understand how it works and why – and they don’t need to be a data scientist.
“CFOs need to show that it’s human plus machine that gives you a better outcome. They need to be able to talk about it and be role models. If you’re putting people through training, do the training,” she explains.
Having the right data and understanding the data journey is also critical before automating, says Professor Michael Davern FCPA, chair of accounting and business information systems at the University of Melbourne.
“If you’re not at scale, you don’t have enough data to do anything,” Davern says.
“The other problem is understanding what the data means. Even a simple transaction or forecast isn’t just about the result – it’s also about the business process that you’ve gone through to get to the outcome.”
“The real risk with autonomous accounting is we lose insight into the path the data went, and end up further distancing ourselves from the reality of what’s happening in the business.”
Understand, then optimise
Knowing the data’s journey is a small step to a bigger one – knowing the business process that could benefit from automation. Davern is adamant management should know a process inside and out before optimising it through automation.
“Automation is an opportunity to redesign the process and leverage the intellectual capital you’ve got for the future. It’s about getting staff familiar with the technology, the process and what the data means. That’s a mindset change for some people.
“From a management perspective, you need to spend more time helping staff to learn and understand the business context.”
Sharma advises his clients to start with a process where success is likely, and risk is low. His mantra is to “automate the gain areas first, not the pain areas”.
These include well-documented processes that are unlikely to require changing in the short to medium term. The act of moving data from one database to another is another low-risk area ripe for automation.
Have realistic expectations
Given the concerns around job loss and change, management plays a pivotal role in introducing and implementing autonomous projects.
Communicating transparently with staff about the process and managing their expectations are key to success, according to an Auckland University of Technology study in 2022. The study looked at staff responses to the introduction of RPAs at a New Zealand financial institution.
One of the study’s authors, Professor Angsana Techatassanasoontorn, says staff perceptions are instrumental to the success of the bot.
“If they were really worried about their job, they were not as open with the business analyst and sharing the information of what tasks they were doing,” she says.
“Staff who were unhappy were those downstream in the automated process, because they were now getting output from software robots, rather than a human they had been working with for some time and could trust.”
Consistent communication with employees is crucial, from the top down, so that all staff members hear the same narrative.
Jonovska also suggests managers treat the machine the same way as they would treat a human and not expect it to deliver straight away. Bots and machine learning require regular finessing by humans, she says.
“It’s actually the human plus the machine that works better.”
The judgement call on automation
The ROI of automation projects can be tough to estimate, and it requires management to think broadly.
BlueScope uses the “hours saved” metric to measure the success of its automation projects, alongside the less straightforward “cost of not acting” metric.
“If we don’t invest in autonomous finance, our people will move to somewhere else, because the work and the impact they can have are more interesting. In the next five years, teams like mine are going to be very normal,” Jonovska says.
Davern says the value question requires open mindedness from management and an acceptance that not everything will be measurable.
“It’s got to be a judgement call – that’s why management get paid the big bucks,” Davern says.
“But don’t look for a single KPI, because you’re not going to get everything down to a quantified number.”
Automation tools to try
Mohit Sharma advises businesses to take an iterative approach to adopting automation tools.
“Don’t try to be perfect with each tool. A lot of tools allow you to purchase add-ons instead of buying from another source,” he says.
Tools for RPA
- UI Path
- Automation Anywhere
- Blue Prism
Tools for machine learning and AI