Case study · Home services

Writing for Google and AI at the Same Time

A local heating company competing with national publishers in Google, and now in ChatGPT and AI Overviews too. We built a content system around the one thing they can’t fake: being a real local business.

ClientC1 Boilers

WhereGas Safe heating engineers, Edinburgh and Musselburgh

ServiceSEO Specialist

Timeline2026

checks before any post goes live

25+

pages with entity-first titles

schema errors found and fixed

1 edit

fixed a sitewide phone number conflict

The situation

C1 Boilers installs, services and repairs boilers across Edinburgh and the Lothians. Their old content approach was the usual one: publish blogs on a calendar and hope. But customers now search in two places. They still use Google, and they also ask ChatGPT, Perplexity and Google’s AI Overviews who to call. Generic content can’t outrank national publishers, and AI tools have no reason to quote it.

What we did

A seven-point filter for every post

Nothing gets published unless it passes all seven checks: will it still be useful in 12 months, does it support the core topic (Edinburgh plus heating), does it have a real local or human story, which stage of the buying journey it serves, can it be reused as a Business Profile post or video, what success looks like, and can that success be tracked.

Content that AI can’t invent

Hyperlocal detail: the Musselburgh boiler installation brief opens with the actual local housing, from older stone properties near the High Street to post-war estates in Pinkie and Levenhall, and what each means for heating.

Real questions answered: every brief includes the “People also ask” questions for its keyword, answered in an FAQ section with FAQ schema.

First-hand expertise: a “What we at C1 Boilers would add” section carries the owner’s own view.

Entity-first titles across 25+ pages, pairing service and location with the brand, because AI tools recognise businesses as entities.

Schema markup: how the website talks to Google and AI

Organization schema with the business details, logo and links to its social profiles.

LocalBusiness schema with name, address, phone, price range and real reviews, making the business eligible for star ratings in search.

A machine-readable service list covering installation, servicing and repair.

The errors we found and fixed

An invalid property flagged by Semrush, replaced with the correct structure.

A postcode typo inside the schema (“EF21” instead of EH21) that could confuse local signals.

A forgotten, hidden block in an old footer widget that loaded on every page with a wrong phone number and a misspelled street. Google was receiving two different phone numbers for one business. One edit to the footer fixed it across the whole site.

The results

The groundwork is live and measurable: Google Analytics, Search Console and conversion tracking verified, the blog cluster publishing against the filter, and clean metadata and schema across the site. The business now gives Google and AI tools one consistent, correct story about who it is, where it works and how to reach it.