Case Study: How a Local Business Went from Invisible to #1 on ChatGPT in 8 Weeks

When Marcus Reilly, owner of Reilly HVAC Services in Austin, Texas, first searched ChatGPT for "best HVAC company in Austin", he expected to see his business. Instead, he saw three competitors cited by name - companies he knew his work consistently outperformed on customer satisfaction scores and review ratings.

That discovery, in January 2026, was the start of an eight-week project that took Reilly HVAC from zero AI search visibility to the first brand cited for their primary target query. This case study documents exactly what was done, in what order, and what the results looked like at each stage.

Note: This case study represents a composite based on the common patterns seen in local business AEO implementations. Business name and specific metrics are illustrative.

The Starting Position

Business: Reilly HVAC Services - residential and light commercial HVAC installation and repair, serving Austin and surrounding areas.

Online presence at the start: A 12-page website built in 2022, strong Google reviews (4.8 stars, 247 reviews), active Google Business Profile, page 2-3 rankings for most local service keywords. No structured schema markup. Content pages were primarily short service descriptions with no FAQ structure and no author information.

AI visibility baseline (measured via AI Rank Lab):

• ChatGPT citation rate across 15 target queries: 0%

• Perplexity citation rate: 0%

• Gemini citation rate: 7% (appeared in one borderline query)

• Primary competitor citation rate: 47% average across the same query set

The AEO/GEO audit score was 8 out of 30 across the 15-item framework, with the most critical gaps in schema markup (no schema of any type on any page), content structure (all pages used long prose paragraphs with no direct-answer structure), and third-party authority signals (not listed on Yelp, Thumbtack, or HomeAdvisor despite strong Google review presence).

Week 1-2: Technical Foundation

The first two weeks focused entirely on technical implementation - no new content, just fixing the structural gaps that were preventing AI crawlers from extracting information from the existing site.

Actions taken:

• Added Organization schema to the homepage: business name, address, phone, service area, business hours, logo, and sameAs links to Google Business Profile, Facebook, and Yelp (after creating the Yelp listing)

• Added LocalBusiness schema (HVAC Contractor type) to the homepage and contact page

• Added Service schema to each of the 8 service pages: service name, provider, area served, and description

• Added FAQPage schema to the About page after restructuring it with 5 common customer questions and answers

• Fixed a robots.txt misconfiguration that was partially blocking the Googlebot and other crawlers from the service pages

• Created Yelp, Thumbtack, Angi, and HomeAdvisor listings - platforms that AI engines heavily reference for local service provider citations

Week 2 citation rate check: ChatGPT 0%, Perplexity 13% (up from 0% - the schema improvements and Yelp/Angi listing creation had already produced a measurable change in Perplexity, which uses real-time retrieval), Gemini 7% (unchanged).

Week 3-4: Content Restructuring

The second fortnight focused on restructuring the existing service pages to lead with direct answers and question-formatted sections - without writing any new content, just reorganising what already existed.

Actions taken:

• Restructured each service page to open with a two-sentence direct answer to the primary query the page targets ("Reilly HVAC provides residential AC installation in Austin with same-day availability. We service all major brands and include a 2-year parts and labour warranty.")

• Converted all prose section headings to question format: "Our Process" became "How does HVAC installation work?", "Pricing" became "How much does HVAC installation cost in Austin?"

• Broke long prose sections into bullet-pointed lists and short direct-answer paragraphs

• Added Marcus Reilly's author byline and a 150-word bio establishing 14 years of HVAC experience and NATE certification to the About page and two main service pages

• Added the current year and source citation to all statistics referenced on the site

Week 4 citation rate check: ChatGPT 21% (first real ChatGPT citations appeared - for the "same-day HVAC repair Austin" query type), Perplexity 34%, Gemini 21%.

Week 5-6: New AEO-Optimised Content

With the technical foundation in place and existing pages restructured, weeks five and six focused on publishing two new content pieces targeting the highest-value uncited query patterns.

Actions taken:

• Published "How much does HVAC repair cost in Austin in 2026?" - a comprehensive guide with current pricing data by service type, structured as a complete answer to the most common service question, with HowTo schema for the "how to get an accurate estimate" section

• Published "Best HVAC companies in Austin: What to look for (and what Reilly HVAC does differently)" - a structured comparison piece covering the criteria customers use to evaluate HVAC contractors, demonstrating expertise and establishing differentiation without making unsupported claims

• Added FAQPage schema to both new articles, each with 6 structured Q&A pairs covering the questions prospective customers ask most commonly

• Updated the homepage with the dateModified field in schema and refreshed the statistics in the "Why choose us" section with 2026 figures

Week 6 citation rate check: ChatGPT 47%, Perplexity 61%, Gemini 38%.

Week 7-8: Authority and Consistency

The final two weeks focused on consolidating the citation gains and extending authority signals.

Actions taken:

• Requested reviews from the 20 most recent customers across Google, Yelp, and Angi simultaneously - the review recency and breadth boost was visible in Perplexity citations specifically

• Published a case study page documenting a recent complex commercial HVAC installation: the problem, the solution, the timeline, and the outcome - with HowTo schema for the diagnostic process

• Added Reilly HVAC to the local Austin business directory maintained by the Austin Chamber of Commerce (high-authority local citation)

• Contacted two local Austin home improvement blogs about featuring the pricing guide as a reference resource

Week 8 citation rate check (final):

• ChatGPT: 68% (first citation in primary query "best HVAC company Austin")

• Perplexity: 74%

• Gemini: 52%

• Combined average: up from 2% to 65% across the 15-query monitoring set

What Changed in the Business

Within three weeks of the ChatGPT citation improvement, Marcus noticed a pattern in new customer enquiries: a higher proportion of callers mentioned they had "looked it up" before calling - phrasing consistent with AI search research rather than a direct Google search. Two callers in week 10 specifically mentioned ChatGPT by name.

By the end of the two-month period, inbound lead volume was up 23% compared to the same period in the previous year - against a market where competitors reported flat or declining inbound enquiries. The attribution was not perfect, but the correlation between AI citation improvement and inbound lead growth was visible in both the data and the direct customer feedback.

The Key Lessons

1. Technical schema implementation was the fastest win. The Perplexity citation improvement in week two - before any content changes - demonstrated how much citation opportunity was being left on the table by the absence of basic structured data.

2. Content restructuring without new content produced measurable gains. Reorganising existing pages to lead with direct answers was more impactful than it appeared on paper - the content quality was already there, it just was not structured for AI extraction.

3. Third-party listing creation mattered more for local than expected. AI engines rely heavily on review platform data for local service recommendations. Creating Yelp and Angi listings where none existed was among the highest-impact single actions in the program.

4. The 8-week timeline is achievable but requires focused execution. Every week had a clear priority list and each item was completed before moving to the next. Programs that try to do everything simultaneously tend to run slower and produce less clear cause-and-effect data.

The Reilly HVAC program used AI Rank Lab for all citation tracking, the AEO/GEO audit, and the new content pieces generated in weeks five and six. Run your own AEO audit to see where your baseline gaps are - the audit output maps directly to the implementation sequence that produced these results.