Organise pages around a real question
Start with the decision a prospective client needs to make. A PR page should explain earned media outreach, newswire distribution and paid partnerships. A case page should describe the brief, the team's role and the results available to share. A substantial subject deserves a page people can visit and pass on directly.
Accordions and modals can make browsing easier. Important subjects should also be reachable through clear links, with a useful route from a service to relevant work and then to an enquiry. Creating more URLs is not a reason to split a few sentences into near-identical pages.
Check what the first response contains
Google can execute JavaScript and process rendered pages. Its guidance still recommends considering server-side rendering or pre-rendering for users and crawlers, noting that some bots cannot run JavaScript. A brand's essential introduction should not depend on a complex interaction before it can be read.
Inspect the HTML the page returns: are the title, main explanation, service scope and internal links already present? Compare that with the browser view. Motion, galleries and disclosure controls can support the experience without making brand information depend on a click, a completed animation or a successful script.
Keep page signals and brand facts aligned
When the same content has several addresses, a canonical identifies the preferred version. When Chinese and English pages coexist, hreflang describes their language relationship. Internal links and the sitemap should point to the corresponding addresses. These settings clarify page relationships; they cannot replace useful content.
The communications team has another consistency check to make. Do the title, body copy, structured reference and downloads describe the same service? Keep project figures in their original context, and distinguish concepts from delivered work. Include these checks in routine editing so an old company description does not remain in circulation.
Separate access from inclusion
OpenAI provides separate controls for OAI-SearchBot, used for search, and GPTBot, used for training. Perplexity also documents its search crawler, PerplexityBot. Search access does not require the same decision as training access. Check firewall rules and request logs alongside robots.txt, verifying bot identity against the provider's documentation.
Google says its AI search features require no additional AI text files or special markup, and meeting the requirements does not guarantee indexing or display. We treat JSON and llms.txt as supplementary references, not a shortcut to citations. Address the content, page relationships and access first, then monitor actual search performance and enquiries.