How to avoid another Canberra tech wreck

Modernising how government works

The real benefit was the standardised data and benchmarking Finance and agencies would get, enabling a far more strategic approach to procurement, contract management, remuneration, and expense management. Modern enterprise management that is considered best practice in most bigger corporates.

Instead of flying blind on its remuneration, consulting, contractor and procurement spend, the Commonwealth would finally get a grip on its annual operating costs of about $230 billion.

It would also help revolutionise the treacle-like administrative culture that dominates so much of the federal system, stymying innovation and transformation.

Electronic invoicing, travel management, and remote services for at-home staff were early wins, as about 40 agencies were moved onto the central platform, which was developed by Services Australia.

But the costs of the so-called GovERP program began to blow out as the bigger agencies and service hubs demanded customisation to meet their different operating worlds.

In an echo of the failed registry modernisation project, yet again the project began to be run as a tech upgrade by juniors. With no forceful business ownership and ministerial champion to drive the top-line business gains.

Naive thinking

This week, Gallagher ditched Finance’s vision of a single enterprise system, arguing it was “based on naive ideas about the ability to standardise complex corporate systems”.

Services Australia will now take over the SAP platform. Other agencies will be able to use enterprise software the Digital Transformation Agency sources – a return to the let the flowers bloom approach that is not exactly the One APS model David Thodey called for in his 2018 review.

This is expected to open the market to smaller local players, most notably TechnologyOne. Many local councils and state agencies use TechOne ERP software, arguing the Australian software better suits their requirements.

The winding back of the large scheme is the latest in big tech projects which have been closed. Each saw their costs blow out, and the business value get lost amid a haze of customisation and workarounds to suit particular processes and methodologies.

The $530 million registry modernisation system the Tax Office was building was junked and sent back to the Australian Securities and Investments Commission when its new cost was estimated to be an eye-watering $2.8 billion. 

Similarly, a $191 million welfare payment calculation engine was also closed down after it was found to take minutes to make a calculation, instead of the two seconds the current platform delivers.

And a $1 billion visa processing plan was aborted, costing taxpayers $100 million in development.

Ending dumb digital change

In each case, the costs began to blow out under the weight of numerous requirement changes.

The federal budgeting system demands these projects be funded upfront, forcing agencies into the absurd situation of having to predict costs while requirements are still not being developed. A truly dumb way to fund digital transformation.

At a recent showcase to vendors, NSW chief information and digital officer Laura Christie revealed the benefits of using a central dedicated fund to avoid the kind of tech wrecks Canberra seems to specialise in.

That fund is meagrely metered out based on strict criteria, with three dollars of benefit for every dollar invested. More than 270 projects with a median cost of $4.2 million were funded at a third lower cost than average project costs. On average, projects were delivered 56 per cent faster.

The whole program was estimated to be delivering economic returns of almost $5 billion. Not a bad payback for a $1.2 billion investment.

Government Services Minister Bill Shorten has seen the light and ordered his CIOs to rethink their procurement approach to develop projects in small, bite-size amounts.

A different approach

Boston Consulting Group global public sector lead Miguel Carrasco has been saying for more than 12 months that agencies need to radically change their technology development strategies.

A new BCG papers says the days of high cost, bespoke software are gone. They argue 90 per cent of business value of new implementations comes from streamlining legacy ways of working.

Where once off-the-shelf software too often failed to meet its promise, these days solutions are faster to deploy and require smaller teams, thanks to their modularity, configurability, cloud-based deployment and no-code approach, he says.

Throw in generative artificial intelligence and development is now being turbocharged.

The key is for agencies to let go of their “specialness” and rework their needs to adapt to the software, rather than the costly and slow exercise of customisation.

Carrasco and his team also advise agencies to separate the build from the ongoing maintenance, and insist vendors transfer that knowledge to break the dependency on vendors.

Most government functions are similar, and the key is to bend agencies’ processes to the software and not the other way around. Case management is an obvious example of where, if agencies are less precious about their special requirements, new systems can be turned live in weeks rather than years.

This requires senior executives to take ownership and for the default to be no special code, unless there is a proven need.

The losers in this important strategic shift are the big system integrators (Accenture et al) whose business model is framed around customisation and the ongoing incumbency this ensures.

Dealing directly with software vendors to extract benefits through configuration, rather than new code, becomes much more important in this approach.

And that requires secretaries confident in technology and digital transformation. Work in progress.