How to Show Up in ChatGPT and Google AI Overviews Answers: Central Bombas' Schema.org Setup
Schema.org for AI search: how we implemented LocalBusiness and FAQPage so a small business shows up in ChatGPT and Perplexity answers.
The client
Showing up on Google isn't enough anymore — more and more people ask ChatGPT, Perplexity, or Google's AI Overviews where to find a service near them, and those answers come from somewhere. They come from structured data, not guesswork. That's the problem we solved for Central Bombas, a washing machine and water pump repair service in Porto Velho, Rondônia, Brazil — a client who came through a referral from Ygor Lopes Calisbino, already served on the Calisbino Imóveis & Arquitetura project. Scope closed at R$400 (roughly USD 70), 7-day deadline, and one specific goal beyond the site itself: schema.org structured for generative AI to understand exactly who the business is, where it's located, and what it solves.
The decision to skip third-party themes
The technical call, made before a single line of code was written, was to build a 100% original theme — no child theme of Astra, GeneratePress, or any other foundation. Functional plugins (SEO, cache, security, forms) went in as usual, but the visual layer is exclusive Central Bombas code, deployed via SSH/SCP directly to the live theme in production.
A courtesy add-on
During execution, the client asked to also include water pump service — outside the original washing-machine-only scope. Crazy Diamond took it on as a courtesy add-on, with no change to price or deadline: a hero section, a dedicated tab under /services, an FAQ, schema.org markup, and the Mueller brand added among the brands serviced, all shipped to production.
SEO built for AI, not just for Google
Schema.org LocalBusiness + FAQPage implemented via functions.php, with consistent, structured NAP data (name, address, phone) — not just for rich snippets on Google, but to surface in answers from ChatGPT, Google AI Overviews, and Perplexity when someone asks about pump repair service in Porto Velho. Rank Math, Wordfence, and UpdraftPlus configured via WP-CLI, with automated backups tested and confirmed.
Performance tuning
The original 2.8MB logo was reduced to 35KB, homepage images were compressed from a combined 15MB down to 8.7MB, and real lazy-loading for the Swiper carousels was implemented after an initial version that loaded everything at once. The hero section also needed a fix: a full-bleed background image with minimal overlay and a height calculated via clamp() so the text wouldn't get cut off on smaller screens.