Every redundancy starts before the redundancy starts. Long before an employer sits an affected employee down with a proposal document, someone has already been modelling org charts, running numbers, and thinking about who stays and who goes. That is not misconduct. It is how businesses plan. The question the Employment Relations Authority keeps coming back to is not whether that thinking happened before consultation opened, but whether it had already closed the door by the time consultation began.
Two 2026 determinations, decided seven weeks apart, draw that line with unusual clarity — and reward reading together.
The Miro board that wasn't quite enough
In Murgatroyd v Xero (NZ) Ltd [2026] NZERA 305, Anna Murgatroyd's redundancy from Xero's education team came with a striking piece of evidence: a Miro board, created by Vikki Bean, Xero's general manager of education and content delivery, in July 2024, months before the October 2024 restructure was announced, showing her name in a box marked "propose to disestablish." On its face, it looked like the smoking gun every predetermination argument will jump on.
The Authority took it seriously and still rejected the sham argument. The audit logs that could have proved who wrote the box, and when, no longer existed — Xero had not asked Miro for them until May 2025, by which time only records from November 2024 onward remained. The final restructure did not match the July board. The Authority accepted that planning documents evolve; a board used for early brainstorming is not the same as a locked decision. Predetermination requires proof, not just an uncomfortable document.
That did not save the dismissal. It failed for a narrower, more specific reason: the four-person selection panel included her direct manager, who had already raised performance concerns about her during an earlier roadshow, and those concerns were never put to Murgatroyd for comment before the panel made its decision. The Authority found that omission made the process unjustified, and set a global compensation figure — reflecting both that omission and Xero's failure to do more to help her find another role within the business — of $20,000, reduced to $16,000 once Murgatroyd was found to have contributed to the problem herself by staying silent about a conflict she knew about, rather than raising it during the process. She was also awarded two weeks' lost remuneration.
The scores that were never shared
Duvaux v Mega Ltd [2026] NZERA 182 is a cleaner failure. Mega had genuine commercial reasons to restructure its technology department, and the Authority accepted that without hesitation. The problem was how selection actually worked: the criteria and weightings set out in the consultation proposal had already been applied to score employees before the proposal was issued. Gaetan Duvaux was never shown his own scores. When he gave feedback on how the criteria should apply to his role, Mega did not engage with it.
Mega argued the outcome would likely have been the same regardless — Duvaux's own scoring, on the numbers, put him in the disestablished group either way. The Authority rejected that as a defence. The point of consultation is the chance to change the outcome, not a formality to be satisfied once a decision already exists. Because the scoring was locked in and undisclosed before the proposal went out, consultation about selection was, in substance, hollow. Unjustified dismissal, $8,000 compensation, three months' lost remuneration.
The line, stated plainly
Put the two side by side and the test becomes obvious in a way that reading either case alone does not quite deliver: the question is not whether an employer thought about outcomes before consultation opened. It always has. The question is whether, once consultation opened, the employee still had a genuine chance to change the employer's mind.
Xero survived the sham argument because the evidence could not show the outcome had actually been fixed in July — the final structure moved, the audit trail was gone, and the Authority was not willing to infer bad faith from an incomplete record. Mega failed because there was no such gap: the scores existed, they were never shared, and the feedback that followed had nowhere left to land.
This matters because it cuts through a piece of folk wisdom that circulates in restructures: that any planning document created before a formal proposal is dangerous, and any employer who has "already decided" is automatically exposed. Neither case supports that. What both cases support is a more precise and, frankly, more useful test for any employer running a restructure: can you show, concretely, that feedback given during consultation was capable of changing something? If scores, criteria, or a shortlist existed before consultation opened, were they disclosed, and was there real room left to move?
That question exposes a pattern familiar to anyone who has sat across the table from an employer mid-restructure: the decision is often already made, and what follows is an attempt to build a process that looks straight on paper. Murgatroyd and Duvaux show that the Authority is not fooled by tidiness. A clean-looking timeline of proposal, feedback, and decision does not establish that consultation was genuine if the substance underneath it — who scored what, who knew what, who could still be moved — tells a different story. Untidiness, on the other hand, is not automatically fatal either, as Xero discovered. What matters is the substance, not the paperwork.
