By 2026 a meaningful share of "find me a plumber in Newcastle" type questions never reach Google's classic search results. They're answered directly by ChatGPT, Google AI Overviews, Perplexity, or whichever assistant the user happens to be in. If you're not in the cited sources, you're not in the answer - and the user never sees you.
The good news: the things that get a local business cited by AI engines overlap heavily with classic local SEO. The bad news: most local businesses aren't yet doing the specific extras that tip the balance. Here's what works.
1. How AI engines actually pick which businesses to cite
Across our testing of ChatGPT, Perplexity and Google AI Overviews on local-business queries, four factors dominate which businesses get named:
- Indexable, structured pages. The engine has to be able to crawl the page and extract clear facts about your services, area, and prices. Schema markup is doing more of this work in 2026 than it did in 2024.
- Third-party corroboration. Trustpilot, Checkatrade, FreeIndex, local press mentions, trade body listings - independent sources confirming the same facts about your business.
- Recency. Pages updated in the last 12 months are dramatically over-represented in citations vs older content of equivalent quality.
- Q&A-style content. Pages structured around specific user questions get cited more than pages structured around marketing themes.
2. Schema strategy specifically for AI
Beyond the standard local SEO schema set, two additions punch above their weight for AI citations:
- FAQPage schema on any page with three or more genuine questions and answers. AI engines lift these almost verbatim.
- HowTo schema on any process content (e.g. "how to prepare for a boiler service"). Less competitive than article content and disproportionately cited.
Make sure the JSON-LD validates in Google's Rich Results Test. Invalid schema is worse than no schema - engines distrust the whole page.
3. The content shape AI engines prefer
From running the same query through multiple engines and inspecting which sources get cited, the pattern is consistent:
- One question per H2.
- A 2–3 sentence direct answer immediately under the H2.
- Optional supporting detail after the direct answer.
- Concrete numbers, prices, timeframes - vague claims aren't quoted.
- Author byline with a real name and role (this article has one - see the top).
4. Build your third-party footprint
AI engines weight independent corroboration heavily. The minimum viable third-party footprint for a UK local business in 2026:
- Google Business Profile, fully populated.
- Trustpilot or Checkatrade (or both, depending on trade) with at least 20 reviews.
- Listing on the relevant trade body (Gas Safe, NICEIC, FMB, NHBC etc.) with your URL.
- Listing in your local council's business directory.
- One mention in the local press a year - most regional papers will cover a 'community sponsorship' story for free.
5. The prompt-test methodology
You can't track AI citations the way you track Google rankings. What you can do is run a repeatable monthly prompt test. The protocol:
- Pick five queries a customer would actually type. Two service-plus-area, one comparative ("best plumber in [town]"), one price ("how much for X in [town]"), one problem ("my boiler is leaking").
- Run each query through ChatGPT, Google AI Overviews, and Perplexity. Use a fresh chat / incognito window each time.
- Record: did your business get named? Did it get linked? Was the source page accurate?
- Note which sources did get cited. Those are your direct competitors and your link-building targets.
- Repeat monthly. Look for trend, not single readings - a single chat is too noisy.
6. A 30-day plan to start showing up
- Week 1: Add LocalBusiness, Service and FAQPage schema across your site. Validate with Rich Results Test.
- Week 2: Restructure your top three service pages around customer questions, with direct 2–3 sentence answers under each H2.
- Week 3: Audit and fix NAP consistency across 15+ UK directories. Submit to any trade-body or council directory you're missing.
- Week 4: Run the prompt-test methodology, baseline your scores, and set a monthly diary reminder to repeat.
Keep reading
- The complete local web design guide (pillar)
- The 30-point local SEO checklist
- Google Business Profile optimisation
FAQs
Is AI search actually sending real customers to local businesses?
Yes, and the share is growing fast. By mid-2026 we're seeing 5–15% of organic traffic to client sites originating from ChatGPT, Perplexity and Google AI Overviews - and the conversion rate of that traffic is higher than classic organic because the user has already had their question partly answered before they click.
Does AI search replace Google?
Not yet, but it's eating into it. Google AI Overviews now occupy roughly the top quarter of the desktop SERP for many local queries. Show up there and you're effectively above position 1.
How do I know if I'm being cited?
Manually, for now. Run the prompt-test methodology in this article weekly. There are no good 'AI rank tracking' tools in 2026 - most of the ones that claim to work are sampling too small to be reliable.
Does schema actually help with AI search?
Significantly. AI engines parse structured data faster and more confidently than unstructured prose. A page with proper LocalBusiness, Service, and FAQPage schema is roughly 3–5x more likely to be cited in our testing than the same content without.
Will AI search eventually let me 'pay to be included'?
Probably yes for the major engines within 18–24 months. Build the organic moat now while the playing field is still level.