The nine-day gap that says everything
On 19 March 2024, officials handed Immigration Minister Erica Stanford a written briefing telling her the Biometric Capability Upgrade (BCU) was tracking well, that “the project approach was sound and robust, the build is achievable, and the risk management practice is effective.” Nine days later, on 28 March, her office finally got hold of the independent quality assurance reports she had asked for five separate times. Those reports told a different story: “we have doubts as to whether the project will in fact deliver at all, and we question its continuation.”
That gap, between the reassurance officials chose to give and the evidence they were sitting on, is the whole story. The project was terminated in November 2025 with at least $35 million spent and nothing delivered. The 2026 Budget confirmed a $31.2 million write-off. For any business owner who has ever run a large capital project, the failure modes here are painfully familiar.
A budget that quietly kept climbing
The BCU, contracted with NEC and sitting inside MBIE, was meant to fix security vulnerabilities and modernise biometric processing. In 2019 its whole-of-life cost was set at $19.5 million. Over the next four years it took four increases, reaching $35 million by 2023 with, as Stanford put it, “almost as far as I can tell no questions asked.”
By autumn 2024 costs had hit $39.9 million before mysteriously decreasing with no paperwork explaining why, while $4.44 million was transferred out between 2022 and 2025, further muddying the true figure. Independent reviewer Greg James found that “the underlying rationale for several cost shifts, particularly those in 2024 that triggered ministerial concern, remains unclear.” Millions moved with no trail. In a private company that alone would trigger a forensic audit.
Gaming the threshold
Here is the part every director should read twice. Cabinet rules require full Cabinet sign-off once a project’s whole-of-life cost tops $35 million. It is a control, exactly like a board approval limit. As Stanford explained, “it’s supposed to be signed out by Cabinet every time the whole of life cost goes up… And that seemingly did not happen.”
When costs pushed past that line, officials tried to slip a rise to $40 million into an unrelated fees and levies Cabinet paper without the minister’s knowledge. When she refused, they restructured the project to keep reported costs under the threshold, and later admitted to Stanford this was intentional. A 5 March 2024 email from MBIE deputy secretary Greg Patchell described the project as “well in flight” and “likely to need to exceed the current spending limit of an all-of-life total of $35m”, treating a governance control as a cash-flow nuisance. Former Labour minister Andrew Little had already declined two requests to lift the budget above that line. Officials tried again under the new government without disclosing the refusals.
The failure modes any business shares
Strip away the politics and this is a procurement case study with five textbook failures. A visible approval limit became a target to engineer around. The people running the project controlled what the decision-maker saw, so IQA reports sat unread while verbal briefings said all was well. Seven years and $35 million in, sunk-cost momentum kept the thing alive with no external forcing function until ministerial advisors demanded the raw reports. People who raised concerns were moved off the project, removing the internal check. And cost shifts of millions happened with no documentation at all.
Stanford’s advisor Kara Isaac put it bluntly in a 28 March 2024 aide memoire: “it appears a number of people have deliberately mislead you about the status of a $32m (and counting) project.” Stanford called it “almost as bad as it gets.”
Why sacking a few people won’t fix it
Public Service Commissioner Sir Brian Roche has appointed Michael Heron KC to investigate the integrity concerns. That matters, but the deeper lesson is structural. As one analyst observed, “a threshold that everyone can see is a threshold someone will engineer around”, and the culture that ran the building rewarded keeping bad news off the minister’s desk. Notably, three successive immigration ministers across two governments appear to have been managed rather than informed. This is not a partisan failure; it survived a change of government.
The kicker is the bill isn’t done. The old biometric system the BCU was meant to replace remains ‘unstable’, so more money will be spent to fix the original problem. The $35 million didn’t buy a solution. It bought a return trip to the starting line, and a case study every business with spending authorities should staple to its next procurement charter.
Sources
- From ‘achievable’ to ‘unlikely’ in nine days: Immigration NZ’s botched $35m project (2026-07-22)
- ‘Furious’ minister reveals officials misled her, knowingly avoided Cabinet over ‘doomed’ immigration project (2026-06-16)
- Seven years, $33m, no results: failed govt IT project ‘misled’ minister (2026-06-16)
- Immigration officials ‘deliberately withheld’ information on failed technology upgrade (2026-06-17)
- Immigration project wasted millions, finances remain unclear (2026-06-17)
- Inside the Immigration NZ $33.4m IT biometric upgrade with NEC that became a case study in failure (2026-06-17)
- What is the right question about the Immigration NZ saga?
- Democracy Briefing: What’s wrong with the public service?