Work through this checklist over the next week. Each item links to the article that explains the reasoning.
The basics (do these first)
- Your pages can be crawled. Check robots.txt does not block Google or the AI crawlers you want, such as GPTBot, ClaudeBot and PerplexityBot. (AI crawler basics)
- Your site is built properly: clean HTML, JavaScript that search engines can read, fast pages, and no heaps of duplicate content. Google lists these as priorities in its AI guide.
- Your local or shop data is current: Google Business Profile and Merchant Center feeds, if they apply to you.
The content
- Each key page answers the main question near the top, in direct language.
- Each page adds something original: your data, your experience, your examples. Google calls this “non-commodity content”.
- Follow-up questions are answered on the same page, with clear headings. (Query fan-out)
- You do not make a page for every tiny variation. That risks Google’s scaled content rules.
Skip these
- llms.txt. Google says it is not needed. (Details)
- Special “AI markup”. Google says there is no special schema for AI features. Keep normal structured data for rich results.
- Chopping content into tiny pieces for robots.
Beyond your own site
- Get mentioned where AI looks: roundup articles, helpful community threads, YouTube, and other publications. (Reddit and AI citations)
- Check your Bing presence, since ChatGPT depends on it for some brands. (Why each AI differs)
Measure
- Look at the AI report in Search Console, impressions only. (How)
- Test 20 customer questions monthly in ChatGPT, Perplexity and Google. Note who is cited.
- Compare before and after for pages hit by AI Overviews. (What the studies say)
If you only do three things: let crawlers in, add one original insight to your best page, and start the monthly prompt test.