ToolsFree, no email.

Assistants rely on structured data to work out what a page is about. Without it they infer, and inference favours sites that did not make them guess.

Conflicting names, addresses or descriptions split your entity in the model's understanding. Pick one form and repeat it.

Marked-up questions are the easiest thing for an assistant to lift and attribute. The content must be visible, not hidden.

Assistants extract a passage, not a page. If the answer arrives in paragraph six, the passage they quote comes from a competitor.

Question headings match the shape of a prompt far better than noun-phrase headings do.

“Founded in 2019 in Leeds” is quotable. “A passionate team of innovators” is not.

Many sites block these accidentally through plugin defaults or a security setting. Blocked means invisible, permanently.

Several AI crawlers do not execute JavaScript. Content that only appears after a script runs may never be seen.

Still the fastest way to get new content discovered by anything that crawls.

Unattributed content is weak evidence. A named author with a traceable history is what distinguishes a source from a page.

Third-party mentions build the entity associations that make a model confident enough to name you.

Undated content gets treated as potentially stale, and stale content loses to fresher sources.

Slow pages get crawled less. Less crawling means less of your content is known.

Duplicates make it ambiguous which page should represent a topic, so neither gets chosen.

Heading structure is how a parser segments your content into passages worth quoting.