If you have switched on ChatGPT, Claude, Gemini or a similar tool somewhere in your business this year, you are in good company. Recent research shows AI use among New Zealand businesses has passed 80%, putting Kiwi businesses among the fastest adopters anywhere in the world. On paper, that looks like a technology success story.
But there is a catch, and it is worth understanding before you invest further: national productivity figures have not moved with it. Labour productivity and multi-factor productivity both went backwards over the same period AI adoption surged. Something doesn’t add up – and the explanation matters for any small business or non-profit deciding what to do next.
Switching AI on is not the same as changing how you work
The gap comes down to depth, not access. For most teams, “using AI” means someone has started leaning on a summarise button in their email client, or a draft-this option in a word processor – genuinely useful, low-effort additions bolted onto a process that has not otherwise changed. That makes individual tasks a little faster. It does not make the business meaningfully more productive, because the process underneath was never redesigned.
The businesses actually pulling ahead are doing something different: rebuilding a specific, high-impact part of how they work – quoting, customer follow-up, donor administration – around what these tools can now do, rather than sprinkling AI on top of an unchanged workflow.
The financial case for doing this properly is real. Analysis prepared by Deloitte Access Economics found that small and medium businesses using AI earned substantially more in the past financial year than comparable businesses that had not adopted it – a gap in the hundreds of thousands of dollars for the average SME. Separate research from AI Forum NZ found that the large majority of AI-using businesses report genuine efficiency improvements and lower operating costs. The upside is there. Most businesses just are not structured to capture it yet.
Why so many businesses get stuck at the shallow-adoption stage
It is not usually about willingness. Government-cited research shows that lack of expertise is one of the most common reasons small businesses give for not going further with AI, and separate industry survey data has found that while the large majority of small and medium businesses report using AI in some form, only a small fraction have anything like a formal AI policy or training programme behind it.
In practice, that means:
- A tool gets adopted individually, by whoever is curious enough to try it – with no process, and no consistency across the team
- Customer, donor or volunteer information stays scattered across spreadsheets, inboxes, and whoever happens to remember it
- There is no time (or appetite) to work through the noise of which specific tools are actually worth paying for, versus which are just well-marketed
For a five-person trades business or a volunteer-run non-profit, none of this is a knowledge problem so much as a time and structure problem.
What closing the gap actually looks like
Getting from “switched on” to “redesigned around it” doesn’t require a big technology budget. It requires, in order:
- An honest look at what you are actually using now, and where time is genuinely being lost – not a generic list of trending tools
- One or two specific tools chosen to fit your actual bottleneck – not a stack of subscriptions nobody has time to learn
- Basic systems underneath it – a proper CRM or contact system, so information is not living in someone’s inbox
- Training that goes until people are actually comfortable, not a single workshop that gets forgotten by the following week
This is, deliberately, the exact sequence we work through with clients in our AI & Automation services – because it is the sequence the data says actually works, not just the one that is easiest to sell.
For non-profits specifically
Non-profits face a version of this gap that is rarely talked about in the broader “AI for business” conversation. Donor records, volunteer coordination, and reporting are often held together by one or two long-serving staff or volunteers, with no system behind them at all. AI and light automation can meaningfully reduce that load – donor acknowledgement, volunteer scheduling reminders, report drafting – but only once there is a basic system in place for the AI tools to plug into. That is usually where we start.
The bottom line
New Zealand businesses are not behind on AI adoption – if anything, the opposite. The opportunity now is in the second step almost nobody is taken yet: turning “we use AI” into “we have changed how we work because of it.” That is a smaller, more specific project than it sounds, and it is exactly what we help small businesses and non-profits do.
Sources:
- Deloitte Access Economics, research commissioned by 2degrees (2026) — AI adoption and SME revenue impact
- AI Forum NZ (2025) — efficiency and cost outcomes among AI-using businesses
- MBIE, citing Datacom research (2024) — barriers to AI adoption among non-users
- EMA member survey (2026) — AI adoption vs. formal AI policy gap among SMEs
- Reporting on NZ productivity figures alongside AI adoption trends (2026)