If you have spent the last decade obsessing over keywords, links, and page speed, here is the uncomfortable truth: AI search engines do not read your website the way you think they do. ChatGPT, Perplexity, Google's AI Overviews, and Gemini are not skimming your prose for relevance. They are looking for structured signals that confirm what your page is about, who wrote it, and whether it can be trusted. That confirmation comes from schema markup.
For NZ businesses trying to get cited in AI answers instead of buried below them, schema is no longer a nice-to-have technical SEO checkbox. It is the difference between being a source the AI quotes and a page the AI ignores. Here is what is actually working in 2026.
What schema markup is, in plain English
Schema markup is structured data added to your website's HTML that tells search engines exactly what each piece of content represents. Instead of an AI guessing whether "Auckland plumber, $185 callout, 24/7" is a service, a price, and an availability window, schema spells it out using a shared vocabulary called Schema.org.
Most schema is added as a small block of JSON-LD inside your page's head. Google, Bing, and the major AI engines all read it. When schema is present and accurate, the engines stop interpreting and start trusting. That trust is what gets you cited.
Why AI search depends on it more than traditional search ever did
Traditional search could rank a page on backlinks, content, and intent signals alone. AI search is different. Large language models are trained to be cautious about misinformation, so they lean heavily on structured signals before quoting a source. When ChatGPT or Google's AI Overviews choose which sites to cite, they prefer pages where:
- The topic is unambiguously defined (Article, Product, Service, FAQ, HowTo)
- The author is identified with credentials
- The publisher is verifiable as a real organisation
- The facts are machine-readable (prices, dates, locations, ratings)
Without schema, your page is just text. With schema, your page becomes a database entry the AI can quote with confidence. That is why our agency has watched plain-text pages slip out of AI citations while properly marked-up pages climb into them, often without any new backlinks at all.
The schema types every NZ business should be using
You do not need to deploy every schema type on Schema.org. Most NZ businesses get the biggest gains from a focused set:
- Organization: confirms who you are, your logo, your address, and your social profiles. This is foundational.
- LocalBusiness: critical for trades, retailers, and service providers. Includes opening hours, service areas, and geo coordinates.
- Article or BlogPosting: identifies blog content with author, publish date, and headline. Helps AI cite your thought leadership.
- FAQPage: pairs questions with answers in a format AI engines love to lift verbatim.
- Product: pulls in price, availability, and reviews for ecommerce.
- Service: defines what you offer, where, and at what price tier.
- Person: gives credibility to your authors and team members, especially with linked credentials.
If your site is built on WordPress, plugins like Rank Math, Yoast, and Schema Pro will handle most of this automatically. The catch is that the defaults are rarely complete. Most sites we audit have schema present but missing the fields AI engines actually use, things like sameAs links, author credentials, and review aggregates.
How to get cited in ChatGPT and Perplexity
AI answer engines treat schema as a credibility shortcut. To get cited, your pages need three things working together:
- Article schema with a real author Person, not a generic "admin" or company name. The author should have a linked Person schema with job title, employer, and ideally a verified social profile.
- FAQPage schema on any page that answers common questions. AI engines lift FAQ answers almost word-for-word into responses, with the source cited. This is the single highest-impact tactic we use for clients chasing AI visibility.
- Organization schema linked across every page with the same NAP (name, address, phone) and consistent sameAs profiles to your LinkedIn, Facebook, and Google Business Profile. Inconsistencies kill trust.
You can validate everything at validator.schema.org and Google's Rich Results Test. If those tools cannot read your schema cleanly, neither can the AI engines.
The mistakes we see most often on NZ websites
Schema is easy to get wrong, and the mistakes are quietly expensive. The most common ones we fix during audits:
- Marking up content that does not exist on the page. If your FAQ schema says "Yes, we service Hamilton" but the page never mentions Hamilton, Google flags this as spam.
- Using LocalBusiness without geo coordinates or opening hours. AI engines need the full picture to recommend you.
- Multiple conflicting schema blocks from competing plugins. Pick one source of truth.
- No author or a fake author. AI search heavily weights authorship in 2026. A real Person schema with credentials is now table stakes.
- Stale dates. Pages with a 2019 datePublished and no dateModified look abandoned to AI engines, even if the content is current.
Where to start this week
If you do nothing else, do this: open your homepage source, search for "@type", and check whether you have Organization or LocalBusiness schema with complete fields. Then check your top three blog posts for Article schema with a named author and current dates. Then add FAQPage schema to your most-visited service or product pages.
That is roughly 80% of the AI visibility win for most NZ businesses, and most of it can be deployed in an afternoon. The remaining 20% is the deep technical layer most agencies do not touch: nested schema, sameAs graphs, knowledge panel claims, and entity consolidation. That work compounds, and it is where the real gap between cited brands and invisible ones now lives.
If you want a hand auditing your site's schema and getting your business cited in AI Overviews, ChatGPT, and Perplexity, get in touch with the MONSTERS_ team. We build AI-ready SEO foundations that turn structured data into measurable visibility.