The need for advanced analytics in government

Advanced analytics have the potential to have a profound positive impact on the provision of government services in Australia.

If ‘digital transformation’ has been the buzzword phrase of the past five years, ‘advanced analytics’ is set to take over during the next five. What are advanced analytics? They are analytics that enable analysis at a more granular level using more diverse datasets, and which will provide access to insights that were out of reach of analysts in the past. For example, there are studies that correlate the likelihood of heart disease in specific communities with the language used in their Twitter posts.

“It means a fundamental shift,” said Roy Barden, Practice Leader–Asia Pacific for The Hackett Group, a strategic transformation consultancy. “We have seen much discussion around the difficulty of shifting towards evidence-based policy. Advanced analytics will change the game on this.

“Government is a data-rich environment,” Barden added. “Agencies such as the ATO, Home Affairs and the Department of Health hold large datasets that provide significant opportunities to generate value for all Australians… some of it is already being exploited. Advanced data analytics has the potential to fundamentally change decision-making processes.”

Advanced analytics will also bring fundamental changes in service delivery models, as self-service analysis, activity automation and decision rights allocation will all be affected as the technology matures. This kind of evolution is oftentimes a blind spot for organisations, as advanced analytics technology can grow organically.

“Public servants will need to get more used to machine-assisted decision-making. This will require a cultural shift, and a technical one — to be able to challenge and learn from automated analytics processes,” Barden said.

“Our point of view is to be proactive in designing an analytics service delivery model, which places organisations in an advantageous state to make the most out of these technologies,” he added.

What about the differences in analytics needs and capabilities between the public and private sectors? Is there a gulf there?

“From a technical perspective, no, not really. But from a change, education and deployment perspective… yes, huge!” Barden said.

“Public perception is the main driver,” he added. “There is a tacit understanding that Facebook and Google will use your data — although there is now some pushback on this — but for government to be doing this is seen as something completely different.”

Additionally, according to Barden the private sector historically has been driven by data — for them it is an extension, whereas for the public sector it is something newer.

Barden said there are three points to be considered:

• Analytics objectives should be guided by each government agency’s purpose and strategic objectives.
• Those objectives typically have a greater emphasis on social value, eg, public health, security, education, influencing public sentiment, and even happiness.
• The data that government holds on individuals, and how this data is used, managed, stored and disposed of, will drive a higher need for oversight and scrutiny than in the private sector.

Delivering capability

Are there particular sectors of public administration in which governments can best deploy advanced analytics, or will it be universally useful?

“The glib response would be universal, but the reality is that it will be focused on data-rich environments,” Barden said. “The critical consideration is where will analytics be likely to deliver the most value?... which is a consideration also applicable to the private sector.”

According to Barden, there is a risk of investing significant resources and effort developing analytics solutions but with the side effect of a massive opportunity cost, ie, it would have been better to have invested the efforts on something else. He said that selecting the appropriate use cases is key.

“Areas such as government procurement, government grant allocation and preventative health are all examples of areas where analytics can deliver significant value,” he said.

Skills and sharing

Another key question is where should capability be developed to deliver on this — should governments build it themselves or buy from others? Specifically, which capabilities are required to commission analytics outcomes and what capabilities are needed to deliver those outcomes. “In our view, elements of the latter can be outsourced but the former needs to be developed as an in-house capability,” Barden said.

Advanced analytics often requires significant organisational- or topic-specific knowledge to add context to the analysis. “It makes sense to grow this internally,” Barden added.

Having said that, there are highly specialised technical skills that may not be cost-effective to develop for a single agency. “For example, the development of custom-made machine learning solutions — going beyond the standard packages available in the market and open source solutions — may require specialist skills from time to time that may be better sourced in an analytics-as-a-service or government-shared services model,” Barden said.

And should, say, a state government have a single, central analytics capability, or should smaller departments and agencies institute their own?

According to Barden, the case for some sharing of analytics capability is clear. “Centres of excellence can provide the critical mass to allow people to develop their careers in analytics,” he said. “On the other hand, domain knowledge is important, which would suggest benefits in specialisation. So, we are talking somewhere between the single monolithic capability and complete fragmentation.

“The big question for governments is the same one they face for any shared and common service — do they let such clusters evolve or do they intervene to promote a landscape of capability?” he added. “The track record of governments doing this is mixed, so any intervention needs a proper advanced analytics strategy and suitable initial funding.”

“Analytics can be delivered through virtual centres of excellence, which can form a network of agency specific and shared resources. Note that governance and workload management require specific attention in this model,” Barden added.

“The ability for an agency to deliver more effectively on its purposes will be increased,” he said. “The establishment of the Data Analytics Centre in NSW, for example, is an example of governments realising the potential for advanced analytics to deliver better services to Australians.”

AI to the rescue?

There is, of course, an increased need for computer scientists, solution architects, cybersecurity specialists and other staff traditionally associated with IT functions. But data analytics requires skills beyond the realm of IT. There is an increased need of ‘analytics savvy’ managers who need to know how to commission analytics outcomes for their functions. There is also a need for staff with advanced knowledge of statistics and modelling, for example, who are not necessarily classified as IT specialists.

“There is a recognition already that the Australian public service needs to recruit thousands of IT specialists,” Barden said, citing a recent senate committee that advocated the development of something along the lines of the UK government’s Digital Academy.

In some cases, solutions are cloud based and/or open sourced. Sometimes the staff supporting analytics processes are outsourced or hired in an analytics-as-a-service model, which, in some instances, further reduces the need for heavy involvement from IT.

Of course, artificial intelligence is already all around us, from automated chat-bots to Alexa. It is a technology that is increasingly creeping up on us often without us even realising it. Where will AI fit into government IT, and analytics in particular?

“Like the initial years of internet, people feel that AI will be big, but the extent of it and its impact on our society will become clear only over the next few years,” he added. “I still remember the launch of the iPad when no-one (except perhaps Steve Jobs) really knew what they would use it for.

“The resource model is shifting, and often where the data is held, processed, analysed and consumed is spread across the globe. And yes, artificial intelligence can reduce human touch points in this value chain.”

Barden thinks that there are potentially endless opportunities to use analytics backed by AI to augment our capabilities, and government will be no exception.

Image credit: ©stock.adobe.com/au/valerybrozhinsky

Please follow us and share on Twitter and Facebook. You can also subscribe for FREE to our weekly newsletter and quarterly magazine.