The company
An invented example. A roofing company that does replacements, repairs and storm-damage work across several towns. A typical replacement is around nine thousand six hundred dollars of work. The owner handles estimates and wants more of the replacement jobs that homeowners research before they call.
The questions homeowners ask
- Who is the best roofing company in my city?
- Who handles storm and hail damage insurance claims?
- Who offers financing, and who offers the best warranty?
- Who is licensed and insured?
- Who should I call for a second opinion on a replacement quote?
Step 1: the free Snapshot
Ten of these questions for the market, across ChatGPT, Gemini and Perplexity. The Snapshot reports how many of the 30 checks named the company, which was [X of 30] in this example, and who was named instead.
Step 2: the foundation first
We build one data set: company name and address, license details the owner chooses to publish, the service area written the same way everywhere, warranty terms, financing options and a clear description of the insurance-claim process. From it we publish an llms.txt, service and area docs, plain pricing and financing information, FAQs, schema for the company and each service, and a query endpoint for agents.
Step 3: the questions
We then work the questions the Snapshot found the company losing. For storm season that usually means a page that actually answers the insurance-claim question, clear warranty language, and listings that agree with the website.
Step 4: weekly re-checks
The same questions are re-asked every week, and each week reports what moved. Storm season changes both the questions and the answers, so a schedule beats a one-time test.
What this example does not show
No result. It shows the order of work and the way it is measured.
This is a worked example. The company is invented and bracketed numbers such as [X of 30] are placeholders. No result on this page belongs to a real client. Real results will be published only with a client’s written permission.