Sixty-eight percent of New Zealand small businesses had no plans to evaluate or invest in AI, according to a 2024 survey by Spark and NZIER. That figure gets used in two ways. Some treat it as a warning sign — evidence that Kiwi businesses are falling behind. Others treat it as validation — proof that the hype is overblown and most businesses are sensibly sitting it out.
Neither reading is quite right. The more useful question is what sits behind that number, and whether the reasoning in your own business is sound.
Skepticism is a reasonable starting point
Most of the pressure to adopt AI is not coming from an operational problem you need to solve. It is coming from vendor marketing, conference agendas, and the general sense that something important is happening and you might be missing it. That is a poor foundation for a business decision.
A lot of early AI projects have not delivered on their promises. The demos were impressive. The production systems were not. Businesses that moved fast often found themselves with expensive tools they could not measure, workflows that depended on outputs nobody fully trusted, and staff who did not know when to use the system and when to ignore it.
Being skeptical is not the same as being wrong.
The question that actually matters
The right question is not “should we use AI?” It is “do we have a problem that AI solves better than the alternatives?”
That reframe is important. AI is a set of tools, not a strategy. Like any tool, it is useful in specific situations and a poor fit in others. A hammer is not better than a screwdriver. It depends on the job.
The situations where AI tends to create clear value for small businesses share a few characteristics. The task is repeated often. It involves processing text, data, or documents. The output is reviewed by a person before it has any consequence. The current process is slow, inconsistent, or dependent on one person’s knowledge. When those conditions are present, the economics of AI often work. When they are absent, the economics usually do not.
What a genuine signal looks like
If you are asking the right question, the answer tends to be specific. Not “we should do something with AI” but “our team spends three hours every Monday pulling data from three systems to produce a report that nobody reads in full.” Or “every new staff member has to learn the same ten things by asking the same three people, and those people are tired of it.”
A genuine signal is a workflow that is slow, manual, and repeated. It has a clear input, a clear output, and a cost you can name. That is the kind of problem that produces a useful AI pilot.
Four questions to ask before doing anything
Is there a specific task in the business that takes more time than it should, happens regularly, and produces something that gets reviewed before it matters? If yes, that is worth exploring further.
Do you have reasonable access to the data or documents the task depends on, in a form that is reasonably accurate? If the inputs are a mess, AI will make the mess faster, not cleaner.
Is there someone in the business who has time to run a contained experiment, assess the output honestly, and stop if it is not working? AI projects without an accountable owner tend to drift.
Is the timing right? A business going through a restructure, a system migration, or a leadership transition has other priorities. AI projects require attention. If attention is not available, the project will not succeed regardless of the technology.
If the answers are mostly yes, there is probably something worth exploring. If they are mostly no, the honest answer is: not yet.
Not yet is a fine outcome
The businesses that get the most from AI are generally the ones that started when the conditions were right, not the ones that started first. A well-run pilot with a clear outcome is worth more than a rushed project that produces inconsistency and erodes trust in the technology.
If you work through those questions and conclude that now is not the right time, that is not a failure of ambition. It is good judgement. The tools will still be there when the conditions are better.
What matters is that the decision is based on your business, your data, and your people — not on someone else’s urgency.
Want to know more? Contact Us.
