Local roofing

How BYLTup got 73% ChatGPT mention on a Long Beach roofing site

ChatGPT named BYLTup on 73% of answers after eight weeks and cited it on 64%. OpenAI crawler hits rose from 2,409 to 6,188 versus late July.

73%ChatGPT mention after 8 weeks
64%ChatGPT cite after 8 weeks
2.5xOpenAI crawler hits vs late July
0→88%Licensed Long Beach roofer named
After eight weeks, ChatGPT: 73% mentioned, 64% cited. OpenAI crawler hits rose 2.5x. Licensed Long Beach roofer still went from never named to 7 of 8 answers.
AP
Appear probe measurement36 paired questions, early July vs late August

The Challenge

BYLTup is a Long Beach roofer. The questions that matter are local and specific: licensed in Long Beach, emergency leak, GAF-certified, who to call tonight.

Early Appear probes on those prompts named nobody, or named a competitor. The site had no answer pages that engines could cite.

Gemini still almost never cites. By late August it mentioned BYLTup on 83% of answers and cited it on 3%. That gap is the product, not a rounding error.

Before Appear

  • Licensed Long Beach roofer: 0% mentioned, 0% cited.
  • Emergency roofer Long Beach: 0% mentioned, 0% cited.
  • GAF-certified Long Beach: 25% mentioned, 0% cited.
  • No published answer pages for engines to quote.

After Implementation

  • Licensed Long Beach roofer: 88% mentioned, 63% cited.
  • Emergency roofer: 75% mentioned, 50% cited.
  • GAF-certified: 88% mentioned, 38% cited.
  • 18 published pages. After eight weeks, ChatGPT: 73% mentioned, 64% cited. OpenAI bots 2.5x.

Platform Performance

Citation rate growth across major LLMs after implementation.

ChatGPT mention64% cite after 8 weeks
Baseline: 68%Current: 73%
Perplexity mention+22 pts vs paired baseline
Baseline: 68%Current: 90%
Claude mention+11 pts vs paired baseline
Baseline: 68%Current: 79%
Gemini mention+15 pts vs paired baseline
Baseline: 68%Current: 83%

Implementation Timeline

22 May 2026

First live pages

Appear published the first BYLTup answer pages into WordPress.

May to August

18 pages shipped

Licensing, emergency, GAF, and local service pages went live and stayed live.

Late August

Paired-question lift

On the 36 questions that existed in both the early-July and late-August windows, mention rose 68% to 79% and cite rose 42% to 55%.

Local service, still unnamed when people ask AI?

Appear writes the pages engines can cite, then measures the same questions until the name shows up.

Book a Demo