Skip to content

Insights ·

How to make your website easier for search engines and AI to cite

To have a company website found on Baidu and cited by AI assistants such as Doubao and DeepSeek, the key is not publishing more articles but five things: state who you are, let crawlers in, put the content in the page itself, keep dates and URLs trustworthy, and have each page answer one specific question.

More and more people no longer page through search results. They ask AI directly: "Which studios in Beijing build brand websites?" The AI's answer comes from web pages it can crawl, understand and feel confident citing. Whether your website appears in those answers is largely decided when the site is built.

When we redesigned our own website this year, we broke this down into five items, each of which can be checked before launch. Here they are in order.

1. Say "who you are" in the same sentence everywhere

When AI introduces a company, what it cites most often is a one-sentence definition: what kind of company it is, when and where it was founded, what it does, and who it has worked for. If the homepage, About page and footer each have a different version, AI treats the inconsistency as uncertainty and either leaves the company out or stitches together a wrong description.

The fix is to write one definition first and use it everywhere the company is introduced across the site, including the description metadata that visitors don't see but search engines read. Keep facts such as the company name, founding year and address consistent across all your own channels.

If the company has several business lines or several websites, also state how they relate to each other; otherwise search engines may merge two unrelated businesses into a single entry.

2. Let search and AI crawlers in

Many websites block some crawlers by default through their site template, without the owners knowing. Our approach was to list, one by one in the site's crawler rules file (robots.txt), the crawlers of search engines such as Baidu, Sogou, 360 and Bing (the first three are Chinese search engines) and of AI products such as ChatGPT, Claude, Perplexity and Doubao (ByteDance's AI assistant), and allow them explicitly rather than relying on each one's default settings.

We also provide a sitemap listing every page, so crawlers can find everything in one pass.

3. Put content in the page, not behind animation

Polished entrance animations and content that only appears when clicked are fine for visitors, but some crawlers read only the page's original HTML and don't wait for animations to finish. If text only appears after a script runs, a crawler may see a blank page.

Our principle: all text is in the page from the start. Animation only controls "how it appears", never "whether it's there". If a script fails to load, the content is still visible.

4. Keep dates and URLs trustworthy

Each page in a sitemap has a "last modified" date. Some websites mark every page as updated today each time they redeploy; when search engines find that this field can't be trusted, they ignore it entirely. We update the date only when a page's content actually changes, and leave it out when it isn't available.

When a redesign changes URLs, map every old URL to its new page with a permanent redirect. This is the step that carries the old site's indexing and inbound links over to the new site, instead of leaving a pile of broken links.

5. Have each page answer a specific question

What AI cites is usually a paragraph that directly answers a question. An article about "how professional we are" is unlikely to be cited; an article about "what to prepare before a website redesign" has a chance, as long as the answer is in the first two paragraphs.

The same goes for case study pages: stating the client, industry, year and scope of work makes a page easier to retrieve than a few large images alone. We have also prepared a site description that AI can read directly (llms.txt), listing company information, services and main pages, and it updates automatically with the website's content.

Pre-launch checklist

  • Is the company introduced in the same sentence across the whole site?
  • Do the crawler rules block any search engine or AI crawlers?
  • With JavaScript turned off in the browser, is the page text still there?
  • Are the dates in the sitemap genuine?
  • Does every old URL redirect to its matching new page?
  • Can you say in one sentence what question each important page answers?

Getting these right won't guarantee that you appear in AI answers straight away, but it does mean that when AI comes looking, it can find your pages, understand them and feel confident citing them.

← All articles