Local SEO
Local SEO Audits
Website Development
Google Local Service Ads
CRO
Key takeaways
- Answer engines retrieve rather than rank. They pull every mention of your business they can find, so being in one place is not enough.
- A business that exists only on its own website and a Google profile does not give a model enough to justify recommending it.
- New Zealand is an unusually open field, because most content answering local service questions was written for American search.
- The signals shift constantly, but the work does not: be present where models look, and be easy to quote when they get there.
AI search is not replacing Google for local businesses outright, but it is changing who gets recommended, and it is happening faster than most trade businesses have noticed. Our own estimate, based on running this across twelve local service businesses, is that within two to three years roughly 30 to 50 percent of find me a local plumber style searches will run through an answer engine, and a meaningful share already does in major metros.
The important part is not the percentage. It is that answer engines choose businesses by a different mechanism than Google Maps does, so the businesses winning there are not always the ones winning the map pack.
How is AI search different from Google?
Google ranks pages. An answer engine retrieves passages. Instead of ordering a list of results, it pulls every mention of your business it can find across the internet and assembles an answer from them. That single difference drives everything else.
So the question it is effectively asking is not are you the best plumber in Hamilton. It is closer to: do I have enough independent material about this business to justify naming it? A business with a website and a Google profile and nothing else does not clear that bar, regardless of how good it is.
Do AI Overviews take my clicks?
For informational searches, yes, and that trend is not reversing. Someone asking how long a hot water cylinder lasts now often gets the answer without visiting anyone's website.
For local service searches the effect is different and, for a trade, more interesting. Nobody gets their drain unblocked by an AI answer. The answer names a business, and that business gets the call. The click matters less than the mention, which is a genuinely different thing to optimise for.
Where do AI answers get local businesses from?
From wherever there is material. Across the local service businesses we have audited, the citations come back as roughly the same mix every time.
| Source | Why it gets cited | What to do about it |
|---|---|---|
| Best-of and roundup articles | They directly answer who is the best X in Y | Get into the ones that already exist and are already cited |
| Forums and discussion threads | Real people naming real businesses | Find the threads being cited and reply properly, naming your city |
| Directories and citations | They verify the business exists | The New Zealand set: Yellow, Neighbourly, NoCowboys, Builderscrack |
| Your Google Business Profile | Reviews, categories and services are structured and specific | Reviews that name the job and the suburb do double duty here |
| Your own website | It answers the question directly, if it is written to | One question per page, answered in the first line |
| Video | Transcripts are indexed and almost no trade is there | Short videos titled with the exact question |
Why does this favour New Zealand businesses right now?
Because the material answering New Zealand service questions is thin. Most of what exists was written for the United States: water heaters instead of hot water cylinders, county permits instead of building consents, prices in the wrong currency. When someone in Tauranga asks an AI what a reroof costs or whether they need a consent, the model often has very little genuinely local to draw on and answers with something imported and slightly wrong.
That gap is an opportunity with a time limit on it. A business publishing genuinely local answers becomes the obvious thing to cite while it is open, and competing for the same citation in two years will be considerably harder.
How do I check whether AI already mentions my business?
Directly, in about fifteen minutes, and it is worth doing before you spend anything. Open ChatGPT in a temporary or incognito chat so your own history is not influencing the answer, then ask the questions your customers would ask, one at a time rather than batched.
- Ask who is the best plumber in your city, then the two or three highest value questions in your trade.
- Scroll past the answer to the sources underneath. Those URLs are the reason it answered the way it did.
- Log the exact page, not just the domain, because the model is lifting one specific passage from it.
- Ask a follow up: why did you pick that business, and what signals led you to recommend them. It will usually just tell you, often naming review counts and specific profile details.
- Repeat monthly. The answers move, and the log is what shows you whether anything you did made a difference.
If your business does not appear at all, that is the expected result for most New Zealand trades right now and it is the reason the opportunity exists. If a competitor does appear, the sources list tells you precisely what you would need to be present in.
What should a New Zealand business do now?
- Find out what is being cited today for your ten to fifteen highest value questions, by asking the engines directly and logging the sources.
- Fix the profile first. Reviews that name the service and the suburb feed both the map pack and the answer engines.
- Publish one page per question, with the answer in the first line, which is set out in how to write blogs that rank in AI search.
- Get mentioned somewhere other than your own website: directories, forums, roundups, video.
- Re-check every quarter. The specific signals move constantly, so treat any checklist, including this one, as current rather than permanent.
Read next: how to get recommended by ChatGPT and AI search and how to write blogs that rank in AI search.
