New Zealand's central digital agency has been told to stop delivering projects and start setting standards. It has also been told to open a direct channel to Australia's Digital Transformation Agency and import models that already work. Trans-Tasman sourcing and security collaboration are named as joint opportunities to explore.
The recommendation sits in the Delivering Digital Government Reset Plan, released by Public Service Commissioner Sir Brian Roche Three reviewers conducted the work over three weeks. They were Adrian Littlewood, former chief executive of Auckland Airport, Justin Gray, former managing director of Datacom, and Matt Crockett, who has led large-scale organisational transformation and performance improvement. It assessed what the Government Digital Delivery Agency must do, how it should be organised and what capability it needs.
The GDDA costs $NZ42 million a year to operate with 170 full-time staff. The reviewers found it hard to get fundamental metrics to judge technology investment performance across agencies. They also found limited understanding of baseline technology spend funded through appropriations against project funding.
Reports and surveys on investment intentions were thinly populated and focused on raw spend and activity rather than value or impact. Governance groups lacked decision rights and concentrated on process and information sharing rather than managing trade-offs across the system.
The review lists disguised examples of failure. Multiple agencies pursued digital identity initiatives with limited coordination, confusing potential private sector participants. Two large related agencies in the same building tendered separately for service delivery modules without planning for integration, before the process was stopped. One agency chose to build its own payments system rather than reuse a mature system already run by another large agency.
Another example carries a data sovereignty edge. A large agency with a mature technology stack had to rely on AI tools hosted in Australian data centres. The vendor advised that single-agency demand did not justify investment in AI compute inside New Zealand.
Agency leaders were blunt
The review quotes agency leads directly. "GDDA did provide us some independent views, but in reality they just helped us to admire the problem," one said. "It didn't make our lives any easier in delivering our project."
Another described the funding path. "The funding model is slow/bureaucratic and focused on the wrong level of detail to the point where it is not fit for the pace of change and the deployment models needed for digital projects."
A third pointed at the gap the centre was meant to fill. "There are core and common functions like data exchange that would make our business much more effective, but we aren't getting a strong direction from the centre."
The plan recommends three parallel tracks. The first is to reset prioritisation, running an urgent diagnostic of all digital projects and baseline technology spend with Treasury. Resources would then move to the highest-value work, with low-value or conflicting projects paused or retired.
The second is to reposition the GDDA around strategy, standards and architecture, assurance, strategic procurement and system capability. Project delivery and ongoing service management would move to lead agencies. The review also recommends renaming and relaunching the central team to signal the change.
The third is to reform the wider system. That means fixing the business case approval process and funding model to suit technology projects, and standardising technology leadership and accountability. It also means enforcing consistent reporting of technology cost and performance, and removing overlapping digital mandates from ministers down.
The reviewers nominate a small number of foundational common-capability projects, with digital identity, data exchange and AI capability given as their hypothesis. Each would get a single accountable agency and an iterative delivery strategy.
The Australian comparison
The review examined Australia, Singapore and the United Kingdom, and concluded the DTA model is the most relevant fit. It draws several lessons from Canberra.
Strict scope discipline heads the list, with the reviewers reporting that combining central policy and delivery functions will "guarantee failure". They note the DTA has no statutory powers, but derives authority from being embedded in spending approvals and from expert advice flowing into investment committees.
The review notes that roughly 80 per cent of Australian federal technology spend sits in agency baselines rather than going through project approvals. It says that spend needs equal attention. It also records that DTA influence grew after the agency shifted from the Department of the Prime Minister and Cabinet to Finance. Concentrating commercial leverage through panels has let the DTA cover more than 75 per cent of technology spend.
The reviewers are not uncritical. They note the DTA identity project still has thin penetration and higher operating costs than commercial alternatives. Interoperability with accredited private systems beats displacement, they observe. They also record that Australia went from a 13-agency AI task force to a fully funded AI team in 12 months. Policy, accountable officials and procurement agreements with global AI firms were in place.
Long-running gap
The plan reproduces a 2013 Government Service Innovation Programme concept proposal that identified payments, data exchange and digital identity as foundational to public service delivery. Almost 13 years later, the reviewers write, they have still not been delivered.
Recommended timings in the plan are aggressive, with prioritisation actions marked for July 2026 and system reform actions for August 2026. The reviewers caution that the actions are a combination rather than a menu. Execution discipline matters as much as the design of the reforms. Sir Brian said he will consider the findings before making decisions on a way forward.