Do Reviews Affect AI Search Recommendations for NZ Businesses?

Customer reviews directly affect what AI search tools recommend. When someone asks ChatGPT, Gemini, Perplexity, or Google AI Overviews for a good plumber in Wellington or the best physio in Christchurch, those tools draw on review platforms as part of how they decide who to mention. A 2026 study analysing over 800,000 responses from AI search tools found that businesses with no review presence were cited just 1 percent of the time, while businesses with active review profiles were cited up to 75.3 percent of the time. For a New Zealand service business, reviews are one of the most practical visibility signals you can influence without a marketing budget.

How does AI search actually use customer reviews?

AI search tools read review text, not just star ratings, to decide which businesses to recommend. They scan multiple sources, including review platforms, to build a picture of what your business does, where it operates, what problems it solves, and whether customers consistently trust it. That picture shapes whether your name appears in an AI answer.

When someone types a question into an AI tool, the system draws on a wide pool of sources to assemble its response. It is not simply copying a ranked list from Google Maps. It is reading patterns: which business names appear consistently, what services are mentioned alongside those names, how recently customers have written about their experiences, and whether the business owner engages with that feedback.

Review and trust platforms account for 14 percent of all citations in AI search responses, making them the second most-referenced source category overall, behind only the general web. This finding comes from an overseas analysis of ChatGPT, Gemini, and Perplexity outputs, and the precise weighting may differ for NZ-specific queries. The underlying principle holds, though: AI systems treat third-party review platforms as independent evidence of real-world quality, and they draw on them systematically rather than occasionally.

If you want to understand how AI search approaches trust and sourcing more broadly, How is ChatGPT search different from Google? explains the core differences in how each tool decides what to surface.

Does the number of reviews I have change whether AI recommends me?

Yes, but volume alone is not the full story. A 2026 study by Trustpilot and SEER Interactive analysed over 800,000 responses from AI search tools and found a steep relationship between review presence and citation rates. Businesses with no review profile were cited just 1 percent of the time. Brands with a median of 13 reviews reached a 54 percent citation rate. Brands with a median of 81 reviews reached 75.3 percent. The jump from having no reviews to having an active presence is the biggest single gain available.

The move from zero to active is larger than the move from 13 to 81 reviews, but both matter. If you currently have fewer than 20 reviews, getting that number up is one of the most direct steps you can take to improve your AI visibility.

Some 2026 research points to an approximate threshold of around 150 reviews across platforms where AI visibility becomes more consistent. NZ-specific conditions change this picture significantly, however. In a regional NZ city like Invercargill, Rotorua, or Whanganui, there may be five to ten businesses in your category, several of which have fewer than 20 reviews. That competitive landscape means 20 to 50 detailed, specific reviews can be enough to stand out, because there are fewer competitors with strong review profiles for AI to weigh you against.

Do star ratings matter more than the words inside reviews?

Star ratings are a secondary signal for AI tools. AI extracts meaning from review text, not from numbers alone. A business with a 4.2-star average and 40 detailed reviews describing specific services, problems solved, and outcomes will typically outperform a competitor with a 4.9-star average and 200 one-line comments that simply say “Great!” or “Five stars.” Star ratings signal general credibility, but without descriptive language they give AI nothing useful to match to a search query.

This is the single most important misconception to correct about reviews and AI search. When someone asks an AI tool “who is the best electrician in Dunedin for rewiring an older home?”, the tool is looking for reviews that mention electrical rewiring, older homes, reliability, and Dunedin locations. A string of five-star ratings without any of that context does not help the system match you to that query.

What carries high value for AI in review text:

• Specific service names, such as “roof restoration”, “deep tissue massage”, or “commercial refrigeration repair”
• Location references, for example “they came out to our place in Papamoa” or “the team services the whole Hawke’s Bay region”
• Problem descriptions in the customer’s own words, such as “we had a burst pipe on a Saturday night”
• Outcome language, such as “fixed within two hours” or “explained everything clearly before starting the job”

Generic reviews like “Highly recommended!” or “Amazing service!” have very low value as AI signals because they contain no information about what your business actually does or where it operates. They give AI nothing to work with when forming a recommendation.

Which review platforms does AI search look at in New Zealand?

Google Business Profile is the most important review platform for most NZ businesses, but it is not the only source AI tools draw on. Third-party platforms, where reviews come from independent customers rather than appearing on your own website, carry extra credibility as corroborating evidence. In New Zealand, the platforms that matter most for local AI search differ somewhat from standard global advice, which tends to focus on Yelp and Trustpilot.

For NZ trade businesses, including builders, plumbers, electricians, and roofers, No Cowboys is worth treating as a priority platform alongside Google. For broader consumer service businesses, ProductReview.co.nz and Facebook recommendations both appear in local search contexts. Industry-specific directories and professional association listings can also be picked up by AI tools, particularly for health and professional services.

Note that most research on which platforms AI prioritises is based on overseas data. How AI tools weight NZ-focused platforms like No Cowboys has not been publicly studied in the same way as global platforms. The practical approach is to maintain a consistent, active presence on Google Business Profile plus two or three platforms that are genuinely relevant to your trade or profession in your region.

One important nuance: reviews on your own website are less reliable as an AI signal than third-party reviews, especially if those reviews are displayed through JavaScript widgets, image carousels, or embedded third-party tools rather than as plain, crawlable HTML text. AI tools may simply not be able to read them. If you have strong testimonials, adding them to your service pages as plain text with schema markup gives AI a direct path to that content.

For practical guidance on keeping your Google Business Profile well maintained, Google Business Profile: the first thing NZ firms fix covers the setup from the ground up.

Do negative reviews hurt my chances of being recommended by AI?

A negative review on its own is not the problem. An unanswered negative review, or a consistent pattern of unresolved complaints, is. AI tools can interpret an owner response as evidence that a business is active and takes its customers seriously. A thoughtful, specific response to a complaint adds indexed, crawlable text that can reinforce your service category, location, and professionalism.

What genuinely signals risk to AI:

• Long periods with no new reviews, which suggests the business may be dormant or struggling
• Negative reviews with no response at all
• A cluster of very similar reviews posted in a short window, which can pattern as inauthentic
• Inconsistent business details across platforms, such as different trading names, outdated hours, or missing phone numbers

A genuine, specific response to a complaint adds useful information. It restates the service type, acknowledges the issue, and shows ongoing engagement with customers. That kind of response is far better for your AI visibility than silence, and it gives any reader of your review profile a reason to trust you.

The risk of fake or manufactured reviews is also worth noting. Beyond the ethical problem, AI tools and review platforms are increasingly capable of identifying patterns of inauthenticity. A suspicious profile with sudden bursts of identical-sounding reviews may be treated with reduced confidence. Genuine reviews, requested from genuine customers in a natural way, are both more effective and more sustainable over time.

Does the NZ market work differently from what global advice says?

Yes, in several important ways. Most research on AI search and reviews comes from the US or broad global studies. The dynamics in New Zealand are meaningfully different, primarily because the market is smaller.

In the US, a single trade category in a major city might have dozens of businesses each with hundreds of reviews. In Hamilton, Palmerston North, or Nelson, there may be six to fifteen businesses in your category, several of which have fewer than 20 reviews. That concentration means each review you earn carries more weight per review than it would in a larger overseas market.

It also means a single detailed negative review can have a proportionally larger impact for a regional NZ business. This is not a reason to avoid negative reviews, but it is a reason to respond to them promptly and specifically rather than leaving them unanswered.

The other NZ-specific consideration is consistency across platforms, sometimes called NAP consistency: your Name, Address, and Phone number. In a smaller market, inconsistencies are easier for AI to detect. If your business appears as “Kapiti Landscaping Ltd” on Google and “Kapiti Landscaping” on Facebook and “Kapiti Landscaping Services” on ProductReview.co.nz, AI tools may treat those as separate entities or reduce their confidence in recommending you. Use exactly the same business name, address, and phone number across every platform, without variation.

According to Yext’s 2026 Consumer Search Behaviors Report, 47 percent of US adults used AI to find a local business in the past month. That figure is US-specific and comparable New Zealand data is not publicly available at the same level of detail. But the direction of travel is consistent with what NZ business owners are observing in their own enquiry patterns, and the underlying signals that shape AI recommendations work the same way regardless of geography.

What does a review actually need to say to help with AI recommendations?

A review that helps AI is one that describes a specific problem, the service provided, and the outcome, in the customer’s own words. The more context it contains, the more effectively AI can match your business to the right search queries. Generic praise is almost worthless as an AI signal.

Consider the difference between these two reviews for a Tauranga plumbing business:

Review A: “Really good. Would definitely recommend. 5 stars.”

Review B: “We had a blocked stormwater drain that was flooding our driveway every time it rained. The team came out the next morning, identified the problem in about 20 minutes, and had it cleared and pressure-washed by lunchtime. Very clear about the cost before they started. We’re in Papamoa and they were punctual and left the site tidy. Would absolutely use them again.”

For AI answering “reliable plumber for blocked drains in Tauranga” or “same-day drain clearing Papamoa”, Review B provides direct, matchable evidence: blocked drain, plumbing, Tauranga, Papamoa, same-day response, pricing transparency. Review A tells AI nothing it can act on.

You do not need to script your customers. A simple message sent after each job works well: “When you write your review, it helps if you can mention the problem we fixed, what we did, and how it went. Even one or two sentences is great.” Most customers are happy to be specific when they know exactly what would be helpful. This applies whether your customer is a homeowner in Porirua, a business owner in Queenstown, or a patient attending a clinic in Hamilton.

What practical steps can a NZ business take right now?

The practical steps are straightforward: build a steady flow of specific reviews on the right platforms, respond to every review, keep your business profile accurate and consistent, and make sure your review content is accessible to AI tools. None of this requires technical expertise or significant spend.

Build a monthly rhythm, not a one-off burst. Ask for a review within 24 to 48 hours of a completed job, while the experience is fresh. A short text message or email template is enough. Aim for 8 to 12 new reviews per month rather than a rush of activity followed by silence. AI tools weight recency, and a steady flow signals an active, ongoing business far more effectively than an old spike.

Give customers a simple prompt. “Could you mention the problem we fixed, what we did, and how it went?” Sent as a follow-up message alongside your review link, this produces much more useful reviews without directing anyone to be dishonest. The best prompts are brief enough to feel natural and specific enough to actually help.

Respond to every review within a week. Keep responses specific. Thank the customer by name if they used one. Reference the job or service they mentioned. For negative reviews, acknowledge the issue and explain what you have done about it. Every substantive response adds text that AI can read and use as evidence of an engaged, active business.

Be consistent across every platform. Check that your business name, address, phone number, and trading hours are identical on Google, Facebook, ProductReview.co.nz, No Cowboys if you are a tradesperson, and any industry directories you are listed in. Run this check at least every six months, since details drift over time.

Add reviews to your website as crawlable text. Pick 8 to 10 strong testimonials and add them to your relevant service pages as plain text, not inside a widget, image, or third-party embed. This gives AI tools a direct path to your review content that does not depend on third-party platform access.

Test your own AI visibility regularly. Open Perplexity, ChatGPT, or Gemini and ask: “Who is the best [your service] in [your city]?” or “Recommend a reliable [service type] near [suburb].” See whether you appear and what the AI says about you. This feedback loop is more informative than any report, because it shows you exactly what a prospective customer would see.

For a deeper look at the broader signals that shape which businesses AI tools choose to cite, How to get your NZ business cited in ChatGPT, Gemini and Perplexity covers the full picture of what AI tools look for when forming a recommendation.

The shift toward AI-mediated local search is still developing in New Zealand, but the underlying principle is not new: genuine, specific customer endorsements shape what people find and trust. AI search has made that dynamic more direct and more consequential for local service businesses. A builder in New Plymouth or a physiotherapist in Dunedin who consistently earns and manages specific, detailed reviews is building exactly the kind of profile that AI tools are designed to surface, whether the next search arrives through Google, ChatGPT, or whatever tool comes after that.