Which bias do you like? Delayed Defence ERP has them all
A large IT project off the rails, who would have thought?
It’s not actually news and picking on the Defence ERP program is like shooting fish in a barrel. It’s not that people aren’t trying to do a good job, it’s just that sometimes they just dig a hole they can’t get out of by themselves.
We know big projects don’t work and we know this one isn’t bucking the trend. Sunk cost, loss aversion, overconfidence and optimism, confirmation, or some other favourite bias, Defence ERP exhibits them all.
Rumelt, author of Good Strategy, Bad Strategy, observes that you don’t need to be a movie director to know a movie is rubbish. You also don’t need to be a specialist to know the difference between good and bad strategy.
After three failed attempts, and at an estimated $3.5 billion cost, you have to ask why are we persisting and what we have to do to change.
The opportunity cost is significant
The project is now estimated to cost $3.5 billion with delivery in 2030. That would be $2.5 billion over the initial approval, and nearly a decade late.
It is simply too big to ignore. It is designed to integrate 500-600 independent applications across Defence. The capability of those users will only be uplifted in accordance with the ERP schedule. Unfortunately, there is every chance that this cost will double, that it will take even longer than its current delayed date, and scope will be limited to reduce risk and claim some semblance of success.
It’s not much money really — only $240 per individual taxpayer. A billion dollars will get you 1,000 people with a clipboard and a spreadsheet for 10 years, as an alternative to a big IT project. A billion dollars will also get another 65, or more, infantry fighting vehicles. That’s another 400 capable, protected, lethal soldiers on the ground.
What’s the combat multiplier of another billion dollars on an ERP system?
A little background
Defence approached the market in 2016 for systems integrators, having selected SAP’s S/4HANA as the technical solution.
Defence settled on IBM, reportedly handing them $112 million for the first phase, and $128m for its second phase. The first rollout was scheduled for 2020-2022 with the whole project expected to take eight years. (ITNews, May 23, 2016).
Accenture has reportedly been paid $16m to review the program, Service Now $62m for portfolio and project management and $51 million to EY for change management (ITNews, September 25, 2024).
By 2019, estimated project cost was $2 billion, and today’s estimates are as high as $3.5 billion (Canberra Times, September 17, 2024).
Supposedly there has been some de-scoping to de-risk the project. Mobile access has been removed, ostensibly because few people have authorised mobiles that can access the network. It’s a connected world and we know mobility delivers productivity and performance gain, so mobile access seems like a function you would want. While connectivity is not an ERP problem, someone else’s failure shouldn’t be an excuse for reducing functionality, it should be the impetus for getting it solved.
The other de-scope was “data integration”. I don’t know what they mean, but it does sound like one of the fundamental reasons for an ERP. De-scoping, delivering minimum achievable functionality presents a Hobson’s Choice to the user: you can have nothing, or you can have what I say you can have.
Last week, Defence advised industry that the planned “go live” for Tranche 1B was delayed for additional testing and rescheduling. Reasonably, there is no target date because they are rescheduling, again.
BHP, a company with $40 billion annual revenues and 90,000 employees, delivered an SAP ERP transformation for $1 billion in three years. The argument is that Defence is significantly more complex. Maybe, but stop making it more complex. The bigger the problem, the harder to solve.
Looking into the crystal ball
Flyvberg and Gardner’s book (“How big things get done”, 2023) draws on a database of more than 16,000 mega projects, about 10% being IT projects. Their conclusions have a sound basis.
They found that the average IT project over-runs by 73%, with just 18% having an over-run greater than 50%. Unfortunately, when it goes wrong it goes very badly wrong. Those 18% of projects have a mean over-run of an extraordinary 447%.
Contrary to the common financial adviser’s disclaimer, past performance is a good indication of future performance. I’m pretty certain Defence ERP is on the path to joining the tail of the “very badly wrong”.
Why is this still a good idea?
The foundations for a single ERP were laid more than a decade ago when the department decided on SAP as a single solution. Twenty years later, assuming all goes to plan this time, Defence will have an enterprise-wide resource management solution. Ask:
- Why, 20 years later, will SAP be the answer to everything? From HR and finance to aircraft maintenance, and 34 different functions in between. Wouldn’t fuel management, for example, be best done by a specialist fuel system?
- Does that mean that critical capability functions such as preparedness reporting, workforce planning and analytics, or explosive ordnance supply chain management, and a host of others, are delayed every time there is a delay in the uber-project? Why is this good?
- Where does this approach fit with the “conflict is imminent” posture?
- Is this empowering “business owners” in Defence to be innovative and adaptive or is it exacerbating centralised control and bureaucracy? Imposing decision rights rather than encouraging the right decision? Are we enabling or hobbling the enterprise?
Finally, one must ask where has technology gone since this concept of a single ERP was crafted.
We need specific functionality to deal with workflow, compliance and reporting. ERPs deliver this. Maybe it makes sense to extend HR, finance and supply chain management into one system, but the focus is data integration and visibility not application and system homogeneity.
In the 1990’s the answer to data integration was a single solution. By the 2000’s it was less important as integrated reporting tools became available, just not accepted by Defence as internal groups exercised stovepipe control of their data.
By 2010 cloud computing, data lakes and effectively unlimited compute power became available. Today we aren’t as reliant on structure and control, we don’t need sameness, we need data access and empowerment. AI will extend our opportunity even further.
While there is truth in Defence’s claim that they are complex, you don’t solve complexity by conformity, compliance, and centralisation.
What it takes to change?
The people who dig holes defend them.
It’s an unusual person who readily acknowledges the choices of the past might not be the best choices for the future. It’s a significant loss of face for an organisation to say to a minister that their previous request for billion-dollar project approval, along with the multiple contract approvals, project reports, and rescheduling proposals may not be the way of the future.
The sunk cost fallacy, which we have invested so much we need to keep going, is a strong influence. So too is the fear of loss, which carries twice the weight as a gain. Loss isn’t just money, it’s also loss of reputation, influence, and control.
If the Minister were to ask, “Are we doing the right thing with the ERP program” the most likely response would unsurprisingly be “We have reviewed our work and with a few tweaks (and another billion dollars or two) this is the right approach”.
Departments defend the holes they have dug — just look at the last two royal commissions.
Ministers rarely have the domain expertise to properly challenge a department. We don’t vote for parliamentarians based on their technical expertise, and they aren’t made Ministers by their parties because of their domain knowledge.
It’s not just a Defence issue, though they do have a reputation for being masterful at obfuscation.
What to do
While I’m open to the possibility that Defence is on the right path, I have low confidence that their advice is wise. This is not just an issue for Defence, it is a public sector challenge, though the size and complexity of Defence, does make it stand out.
There are simply too many questions, not enough evidence, and inadequate past performance to give a higher level of confidence and trust. It’s not an issue of integrity, it’s the confidence that wise decisions in the public good are being made, and programs are being well implemented, delivering best value from the public purse.
Public servants and the public service are not inherently bad, but there is evidence suggesting a loss of direction. I’m sure there are a variety of views on politicians and Ministers, but I would allow them to want good delivery outcomes for their policies.
The way to better performance and higher confidence is to open the decisions to scrutiny and independent review. If you can’t defend it, you shouldn’t do it.
Long term, our governance of public sector delivery needs a nudge. In the short term, however, we need to do something with this poorly performing, repeat offender, mega project. I don’t think we can leave it to Defence.
The minister needs independent advice — not from another department or from people who owe their allegiance and future to the public service or a politician.
In this case, create a “tiger team” of unusual suspects. Choose a few from non-defence industries, some independent IT expertise bringing knowledge of both big solutions and the potential of future solutions, and one or two who understand the complexities of Defence but aren’t wedded to the past. We might find we don’t need many people, just different ones.
A blended group of commercial, technical and defence experience with a dose of impatience. Pull apart the Defence ERP strategy — its purpose, potential and contracts. Set the standard to be clarity of outcome, commercially sound, achievable in a public sector environment, delivering value to the public.
Defence has nothing to fear if its strategy is sound, and it should change if it isn’t.
Across the public sector, we could do with a little less of the cosiness of the public sector and the embedded industry, and a bit more open and independent advice to the minister. Then we can judge ministers on their choices, and departments on their performance.
What have we got to lose by trying something different?
