AI Insights · AI Strategy & ROI · 7 min read

How to Choose an AI Consultant in New Zealand: 7 Questions That Separate the Real Ones

Last updated 22 July 2026

To choose an AI consultant in New Zealand, test for four things: do they build and operate systems or only produce recommendations; will they tell you not to proceed when the case isn’t there; can they answer precisely where your data will be processed; and is their pricing staged so you can exit after each step. A consultant who passes all four is rare — and worth more than the biggest name that fails them.

The seven questions to ask every candidate

  • “Who delivers the implementation?” If the people scoping your work won’t be the people building it, you’re buying a handover risk with a consulting fee attached.
  • “Tell me about a client you turned away.” Everyone claims honesty; this question tests whether their business model allows it.
  • “Where, physically, will our data be processed?” Vague reassurance is a fail. The right answer names infrastructure, jurisdiction and who controls access.
  • “What happens to our costs if usage triples?” Distinguishes fixed-cost architecture from usage-metered pricing that punishes success.
  • “Show me something you’ve published.” A consultant’s public thinking is the only free sample of their judgement you’ll get.
  • “What does the first exit point cost?” Good engagements are staged — a small fixed step that produces a decision, not a contract you can only leave by finishing.
  • “Which of our obligations have you worked with?” Privacy Act 2020, sector codes, public-sector frameworks, Māori data sovereignty — NZ specifics, not imported generalities.

The red flags that should end the conversation

  • A flat quote before anyone has looked at your data, use case or governance requirements.
  • Guaranteed ROI percentages before discovery — real numbers come from your volumes, not their brochure.
  • “Fully compliant” claims — compliance attaches to how a system is deployed and governed, not to a product label.
  • No willingness to say “a public tool is enough for you” — a consultant who can never recommend the cheap option isn’t advising, they’re selling.
  • Strategy engagements with no path to a working system — the deck is a means, not the deliverable.

What should AI consulting cost in New Zealand?

The NZ market spans free discovery conversations, fixed-fee workshops in the low thousands, scoped implementations from tens of thousands, and six-figure enterprise programmes. The shape matters more than the number: a trustworthy engagement starts small and fixed — enough to make a confident decision — and only scales after the decision is made with real information. Be suspicious of both extremes: open-ended time-and-materials from day one, and suspiciously cheap flat packages that must cut corners somewhere you can’t see.

Big firm, boutique, or freelancer — does size matter?

Less than fit does. Big firms bring process and bench depth but often separate the selling partner from the delivery team. Boutiques live or die on reputation, so incentives usually align — if they have genuine delivery capability. Freelancers can be excellent for narrow, well-defined problems but struggle to operate what they build long-term. Whatever the size, apply the same seven questions; the answers matter more than the letterhead.

Frequently asked questions

For work involving NZ data obligations — Privacy Act, public-sector expectations, Māori data sovereignty — local context has real value, and same-time-zone delivery matters more than it sounds during a deployment. For purely generic advice, geography matters less. The test isn’t the address; it’s whether they can answer the NZ-specific questions above without improvising.

Normalise them to three numbers: cost to reach a confident go/no-go decision, cost to a first working system, and annual running cost at your expected usage. Vendors structure proposals to hide one of the three — making all three explicit is the fastest way to see which proposal is actually cheapest.

The people who scope the work deliver it; discovery is free precisely so we can say “don’t proceed” without a conflict; deployments run inside infrastructure you control, in a jurisdiction you choose; and the engagement path is staged — free call, fixed-fee workshop, scoped deployment — so every step has an exit. Our published work is on this blog; judge it there.

Ready to talk it through?

Book a free discovery call. No preparation required — just tell us what you’re trying to solve.