Here is the question every custom builder asks once they understand what's happening to search: when a prospective buyer types "NetZero custom home builders in London Ontario" into ChatGPT or Perplexity, does our site come up. Right now, for most custom builders, the answer is no. The AI engines cite the volume builders with massive page counts and the regional aggregator directories that have been syndicating builder profiles for years. Your independent custom shop with eight gorgeous models and a NetZero certification is invisible.
This is fixable. The fix is structural, not promotional. AEO for custom builders is the work of making your site machine-readable in a way the AI engines reward, which means structured data, topic depth, and the building-science vocabulary that demonstrates expertise. Here's the playbook we run.
HomeBuilder schema as the spine
HomeBuilder is a Schema.org subclass of LocalBusiness that signals to Google and the AI engines that your business is specifically a custom or production home builder. Most builders ship either no schema at all or a generic LocalBusiness block, which is enough to flag a location but not enough to flag the trade. The HomeBuilder block we ship for Starlit carries the NetZero certification, Energy Star membership, and the LHBA, OHBA, CHBA, and Tarion affiliations as structured credential entries. When ChatGPT answers a query about NetZero builders in Southwestern Ontario, it's reading those credentials as facts, not parsing them out of marketing copy.
Product schema per model with additionalProperty
Every model line in your catalogue should be a Product schema entry with a full additionalProperty list. Square footage, bedroom count, bathroom count, garage configuration, blower-door ACH target, R-value of walls, R-value of attic, HRV or ERV spec, primary heating system. These are the spec-sheet entries a serious buyer scans before contacting and they're the structured facts AI engines pull when answering specification queries. The Starlit build runs Vega, Lyra, Orion, and Draco models in the 1,420 to 2,400 square foot range with the 1.5 ACH or better blower-door standard wired in. That's the spec-sheet rigour the AI engines reward.
FAQPage schema with real buyer questions
The FAQPage block is where most custom builder sites leak the most opportunity. Buyers ask a small set of specific questions before signing: what does Tarion warranty cover, what's the difference between Net Zero and Net Zero Ready, what does 1.5 ACH actually mean, can we finance during construction, what's the timeline from contract to close, what's included in the base price. These belong on indexable URLs with FAQPage schema, not in the founder's inbox three times a week. When somebody asks Perplexity "what does Tarion cover on a new build in Ontario," the builder with a structured FAQPage block gets quoted. The builder without one is invisible.
Per-page topic depth on building-science articles
The building-science blog pipeline is the part that separates a custom builder from a production builder in the eyes of AI search. Topics like thermal bridging, ERV mechanics versus HRV mechanics, radon mitigation, sub-slab insulation strategies, ACH standards, and hybrid heat pump comparisons are searched by serious buyers who are reading three to five pages before contacting a builder. Starlit ships nine articles per quarter on these topics, which is more depth than any other regional Southwestern Ontario builder writes. When an AI engine answers a query about thermal bridging in residential construction, it cites the source with the most credible depth on the topic. That's the citation flywheel.
llms.txt and the AI-bot allowlist
llms.txt is the structured manifest file that tells AI crawlers what content on your site is canonical and how it's organized. It's the AEO equivalent of robots.txt for AI engines. Every builder site we ship has a llms.txt file pointed at the model pages, the building-science blog, the FAQ, and the community pages. The AI-bot allowlist sits beside it and explicitly permits Anthropic's crawler, OpenAI's GPTBot, Perplexity's bot, and the Google AI Overviews crawler. Most builder sites accidentally block these in their robots.txt and never realize it. The free audit shows the allowlist gap before we quote.
Cite-met monitoring across six AI engines
Once the structural work is done, the monitoring layer tells you it's working. Cite-met tracks brand mentions and citation events across ChatGPT, Perplexity, Google AI Overviews, Gemini, Claude, and Copilot. You see when your model pages start getting quoted, which queries are triggering citations, and where the depth gap is. Starlit went from zero AI citations to citations across multiple engines for queries like "NetZero builders London Ontario" within the first quarter post-launch. See cite-met.com for the monitoring tool itself.
Cross-trade overlap matters here
The buyer commissioning a NetZero custom build is often the same buyer commissioning a Steinway-grade home cinema and a multi-zone hybrid heat pump install. The AI engines map these buyer journeys together when the structured data on all three sites is shipped at the same depth. If you have referral relationships with a luxury AV installer or an HVAC contractor in your region, the lift compounds. Our AV installer hub (built on the Zebra Home Cinema case with 190+ URLs and five schema types) and the HVAC contractor sub-hub are the natural cross-references.
The audit shows the gap
The free audit runs three tools: Lighthouse for performance, isitagentready.com for AEO readiness, and a Cite-met citation pull across six engines. The findings land on one page. A zero AI citation count gets stated plainly, a missing HomeBuilder schema gets shown against the competitors who have it, and an absent llms.txt file gets a link to a sample you can copy. Then we quote below what you're paying now. Get the audit here.