Precision Insights / Government IT project failure

Government programmes

Government IT project failure
follows six causes the auditors keep finding

Audit reports on Australian government technology programmes, from 2009 to 2026, keep finding the same causes of failure. The lessons are well known, and the auditors report that they are not consistently applied.

Precision Insights By Precision Consulting Corporation 8 minute read

Reviewed by a Precision senior practitioner who has held programme director or equivalent accountability, and who leads independent assurance for government programmes.

In brief

  • Ten audit reports from five Australian jurisdictions, from 2009 to 2026, find the same six causes of failure in government technology programmes.
  • The business case is the first failure: costs and timeframes underestimated, assumptions untested, and benefits stated without targets.
  • In several of the failures examined, an assurance review or an adviser had already raised the warning.
  • A board can turn each finding into a question to ask before approval and at every gate.

The checklist

These sections of the insight are printed together as a checklist to keep or share:

Download the checklist (PDF)

Why do government IT projects fail?

Australian auditors keep finding six causes: an optimistic business case, governance that cannot challenge, scope settled too late, a supplier trusted beyond the evidence, assurance warnings set aside, and benefits nobody measured. Most failures carry several of them at once.

None of this is new. Victoria's auditor reported in 2015 that weaknesses it had previously found in ICT project planning and delivery "continue unabated", and Queensland's concluded in 2020 that lessons from earlier projects are widely known and not consistently applied.

Ten Australian audit reports on government technology programmes, 2009 to 2026, and the cause each finding illustrates
Audit reportWhat the auditor foundCause
ANAO Report 21 of 2008-09, business system project approvals Business cases lacked the information for a reality check, and benefits had no targets.Business case; benefits
QAO Report 7: 2010, Queensland Health payroll The agency never formally agreed the scope, and more than 47 change requests followed.Scope
VAGO, Digital Dashboard, 2015 Agencies could not demonstrate the benefits of their ICT investments.Benefits
ANAO Report 59 of 2016-17, myGov Largely effective, with no measures developed to quantify the forecast savings.Benefits
QAO Report 10: 2019-20, State Penalties Enforcement Registry A steering committee without the skills or independence to govern the ICT work.Governance; supplier; assurance
QAO Report 7: 2020-21, Delivering successful technology projects Lessons are widely known and not consistently applied.All six
ANAO Report 1 of 2021-22, Defence enterprise resource planning Contractors involved in decisions about their own contracts.Governance; supplier
Audit Office of NSW, Driver vehicle system, 2024 A third attempt, with no approved business case.Business case
WA OAG Report 21: 2024-25, Major IT Projects Ten projects at least $1.6 billion over the cost first advised to government.Business case
Audit Office of NSW, Upgrades to core policing technology, 2026 Gateway reviews and advisers warned of delivery optimism and supplier risk.Assurance; supplier

How optimistic are government IT business cases?

Too optimistic, in small projects and large ones. The first estimate is usually the one that fails: the auditors find costs and timeframes understated, assumptions untested, and options left unexamined.

The Commonwealth auditor found in 2009 that business cases often gave decision makers too little information for a reality check of the approach, the benefits, the cost and the timetable. Half of the 13 projects it followed took at least half as long again as planned.

Western Australia's first transparency report on major IT projects, in June 2025, found ten projects estimated at least $1.6 billion more than the $2.6 billion first advised to government, with half at least doubling in cost and delays of four to 78 months. The Auditor General observed that entities can easily underestimate costs and timeframes without robust planning and a good understanding of their business and system needs.

In New South Wales, Transport for NSW had spent at least $36 million since 2019 on replacing its driver and vehicle system and was on its third attempt, still without an approved business case. The executives responsible could not clearly say why the earlier attempts had not progressed.

What governance failures do the auditors find?

Steering committees that cannot challenge. The auditors find committees without the skill, independence or information to test what they are told, relying on the advisers and suppliers whose work they are meant to oversee.

The Queensland Audit Office's 2020 report on the State Penalties Enforcement Registry is the clearest account. For most of the project the steering committee had two decision-making members, no independent member and no prior experience of a project of that size. It relied heavily on consultants and contractors, underplayed the risk of failure, and kept advising that the issues would be resolved.

In 2021 the Commonwealth auditor found Defence's administration of its enterprise resource planning work largely effective, and still reported that its governance had been undermined by conflicts of interest, with contractors involved in decisions about their own contracts.

Why do scope and supplier problems compound each other?

Because every requirement left undefined at signature becomes a variation later. The contract grows, the schedule slips, and the agency loses the ability to judge whether the product can meet its needs.

In the Queensland Health payroll implementation the scope was never formally agreed by the agency that would use the system, and the contract price rose from $6.19 million to $24 million. The auditor concluded that the certainty a fixed price contract was meant to give had been severely compromised by the absence of an agreed scope.

The State Penalties Enforcement Registry relied on its supplier to say what good practice looked like, and did not define its requirements in detail until about 15 months after signing. Three significant variations and more than 300 change requests followed. After termination the agency had no system, and the software stayed with the supplier under the service arrangement.

Were the warnings given before these programmes failed?

In several of the failures the auditors examined, yes. An assurance review or an adviser had already named the risk, and the programme went ahead with it accepted or not sufficiently mitigated.

The NSW Auditor-General reported in May 2026 that Gateway reviews and external advisers had warned the NSW Police Force about delivery optimism, resourcing constraints and the capability of its highest ranked supplier. The contract was terminated in 2022, after close to $20 million had been spent on work with no ongoing benefit. The Queensland auditor found the State Penalties Enforcement Registry chose to remain overly optimistic rather than pause when it had the chance.

Assurance works when it is acted on. Before Defence's enterprise resource planning work reached second pass, Gateway reviews and an independent assurance review identified significant deficiencies in its planning, and the Commonwealth auditor found the planning after that point largely effective. Our insight on the Gateway review process explains what each rating should trigger.

Why are the benefits of government IT projects so rarely proven?

Because the targets and the measures are rarely set when the investment is approved. Without a baseline and an owner, nobody can later show whether the benefit arrived.

The Commonwealth auditor found in 2009 that benefits were often described in general terms, without targets or a plan to measure them. In 2017 it found that myGov had attracted nearly double the accounts its business case forecast, but that no measures had been developed to quantify the savings, so it could not be determined whether they had been realised.

Victoria's auditor found in 2015 that only a quarter of 1,249 ICT projects had a benefits realisation plan, and that a little over one in ten of the 788 completed projects had had their expected benefits assessed.

Precision perspective

What should a sponsor or board do differently?

Turn each recurring audit finding into a question the board asks before approval and at every gate, and require evidence for the answer. A programme that cannot answer one of these questions is carrying the risk the auditors keep finding.

Six recurring audit findings, the question a board should ask about each, and the evidence that answers it
Recurring findingThe question for the boardEvidence that answers it
Optimistic business case Has anyone outside the team tested the estimate, the assumptions and the options?An independent review of the cost and schedule estimate, and the options set aside
Governance that cannot challenge Does the steering committee have the skill, an independent member and its own information?Its membership, decision log and the papers it actually receives
Scope settled too late Are the requirements agreed by the business that will use the system, before the contract is signed?Signed requirements and the trend in change requests
Supplier trusted beyond the evidence Has the product been seen working at comparable scale, by the team that will deliver it here?Reference sites, due diligence on the local delivery team, and who decides on contract variations
Assurance set aside What did the last review recommend, and what was done about each recommendation?The action record from every assurance review
Benefits never measured Who owns each benefit, against what baseline, and when is it measured?A benefits register with owners, baselines and dates

Ask the questions early. Among the Queensland auditor's actions for boards and executives is to take on only the number, size and nature of projects they have the capability to deliver.

When should a government IT project be stopped?

When the evidence says the path to the outcome is no longer credible, and before the money already spent becomes the reason to continue. The Queensland Audit Office says cancellation can be appropriate if it comes before significant losses accumulate.

The same report listed three cancelled Queensland projects that had together spent more than $138 million, and said senior leaders should feel empowered to consider cancellation when it is in the state's interest. A reset is the other legitimate answer: the NSW Police Force reset its core technology work in 2024, and the Auditor-General found that governance and programme management had improved significantly since. Our insight on programme recovery sets out how that decision is made.

Sources and further reading

Statements about government frameworks were checked against these sources on 16 November 2026. Frameworks are revised, so read the current guidance for your jurisdiction.

How we work

How Precision helps a government sponsor

We give sponsors and boards an independent read on a government programme before the auditors do: a health check or readiness review that tests the business case, the governance, the supplier position and the benefits against the evidence.

Where the evidence calls for a reset, we say so. A programme we help recover is one we no longer assure, and we take no readiness or recovery work on a programme one of our practitioners has reviewed.

Concerned about a government programme?

Tell us about the programme and the decision ahead of it. Initial conversations are confidential and without obligation.

Discuss independent programme assurance More Precision Insights