Homeowners were getting their answers from AI before they reached a website. A windows and doors company is a source those answers cite, more often than two competing installers.
The Client and the Problem
Our client is a Mid-Atlantic company that sells and installs windows, doors, siding and roofing. Every sale starts with a homeowner doing research, and that research has moved.
A homeowner who wants to know which siding holds up best, or what a new roof should cost, now often gets the answer from AI right on the results page, written by Google’s AI Overviews or by ChatGPT. They may never click through to anyone’s website.
The client’s own data showed the gap. On one common window search, one of its window articles averaged a top-3 position on Google over 90 days, with more than 60,000 impressions and only 21 clicks. Google now puts an AI-written answer at the very top of that search.
Ranking on Google is still part of being found. But when the answer is written for the homeowner, the businesses that get seen are the ones the answer cites.
1. Fix the Foundation AI Reads From
We cleared about 3,000 technical warnings, repaired broken structured data, fixed duplicate headings on core pages and rebuilt a broken service-area section. When the site moved to a new platform, we mapped 124 redirects for pages that already ranked.
A site search engines and AI tools can crawl and read cleanly
2. Answer the Questions Homeowners Actually Ask
We built the blog around real homeowner questions: siding comparisons, roofing costs, window types. We published 23 new posts from June to August 2025 and 25 more from April 1 to June 15, 2026. In February 2026 we updated 125 existing posts.
Pages written to be the answer, not just to rank
3. Make Every Page Easy for AI to Understand
We added article markup to the 125 updated posts, business and FAQ markup to its core window pages, and service and FAQ markup to the main city pages. That markup tells search engines and AI tools what a page covers, who it is for and where the company works.
AI can tell what each page answers and where the company works
4. Build a Page for Every Market
We built city and neighborhood pages across the Mid-Atlantic, including neighborhood pages in its largest metro area for each service it sells there.
Local pages for the markets the company serves
The Results
How often AI answers cite the client’s website, compared with two competing window installers. Counts measured November 2, 2026:
| Our client | Competing installer A | Competing installer B | |
|---|---|---|---|
| Citations in Google's AI Overviews | 1,520 | 1,294 | 895 |
| Citations in ChatGPT answers | 299 | 273 | 64 |
Which pages get cited, and what homeowners see when they ask:
| Measure | Result |
|---|---|
| Most-cited pages in the 200 highest-volume Google AI Overview answers (October 30, 2026) | #1 a siding comparison and #2 a roofing cost guide, both written in our content program |
| Perplexity, asked for the best window replacement company three times in each of two metro areas | Named the client in 6 of 6 answers |
| ChatGPT, asked the same question three times in each of four metro areas | Named the client in 4 of 12 answers: 3 of 3 in its home metro, 1 of 3 in a second, 0 of 3 in the other two |
| Visits from ChatGPT, January 1 to October 29 (paid ads excluded) | 594 in 2026, up from 381 in 2025 (+56%) |
The naming tests were run on November 2, 2026. AI answers change from day to day, so they show where the client stood that day, not a fixed ranking.
Key Takeaway
Homeowners still research before they buy. More of that research now ends with an answer written by AI, and the companies in those answers are the ones homeowners see first. For this client:
- Cited more than two competing installers in both Google’s AI Overviews and ChatGPT
- The two most-cited pages in the 200 top Google AI answers we measured came from our content program, written around homeowner questions
- ChatGPT sent more visitors this year: 594 from January 1 to October 29, up from 381
Being the source AI cites is now part of being found.
More on how we work as a home services marketing agency, and how we approach answer engine optimization and generative engine optimization.
When a homeowner asks AI about the work you do, is your company the source it cites?