Why doesn't ChatGPT mention my business?

Most small sites never appear because they never answer the question the user typed, they hide the facts inside long posts, they have no author or date, and a Wikipedia entry, directory, or larger competitor is simply easier to cite. None of this is personal and none of it means an algorithm is against you. It means the engine had an easier source available and took it.

Five concrete reasons

1. No page matches the question. You have a Services page and an About page. The customer asked "who does emergency septic pumping in Ulster County on a weekend?" Nothing on your site answers that sentence.
2. The answer is buried. The fact exists, but it is in paragraph nine of a 1,800-word post, phrased as a hint rather than a statement.
3. No author, no date. Several engines prefer sources that show who wrote it and when it was last checked. A page with neither looks unmaintained.
4. No structure to lift. No headings that match the question, no FAQ block, no schema — so the compression step has nothing self-contained to reuse.
5. An easier source exists. Directories, chambers of commerce, franchise sites, and Wikipedia are already trusted and already answer-shaped. You are competing with them, not with the model.

What we build instead

• — One URL per real customer question, with that question as the H1.
• — The answer in the first 80 words, with any number, price range, or condition stated plainly.
• — A short "who this is for / not for" block so the engine can tell when your answer applies.
• — Sources for anything factual, and no source for anything we cannot support.
• — Visible author and updated date on every page.
• — Four FAQs with matching FAQPage JSON-LD.

Things that are usually not the problem

Site age is rarely the blocker. Neither is domain authority in the way SEO tools describe it. We regularly see three-year-old, low-traffic sites get cited for narrow local questions because nobody else wrote the page. The two real blockers we do find are technical — a site that renders only in JavaScript with no server HTML, or a robots rule that blocks the crawlers the engines use.

How to check in ten minutes

1. Ask an engine the five questions your customers ask most, using their words, not yours.
2. Note who gets named and which URLs are cited.
3. Open your site and try to find the answer to each question in under fifteen seconds.
4. Check whether each of those pages shows an author and a date.
5. View source on one page and confirm the text is in the HTML, not injected by script.

If steps three and four fail, that is your project. If step five fails, fix rendering first — no amount of copy will help a page an engine cannot read.

Sources

Crawler and indexing behavior claims should be verified against Google Search Central and each engine's published crawler documentation, including OpenAI's and Perplexity's crawler pages.

What changes after you fix it

Expect movement on narrow questions first. A page answering "do you take same-week appointments and what is the deposit?" can be picked up within weeks because almost nobody else wrote it. Broad category questions move slowly or not at all, because the engine already has a default it trusts. Track the difference: keep the same fifteen questions, score them the same way, and judge the work on the narrow set. If nothing moves after two reruns, the usual cause is that the pages are still not answering in the first screen, or the site is not being crawled at all.