01
1. Make key pages reachable without a human session
Start with ordinary web access. Important public pages should return a successful HTTP response without login, CAPTCHA or a browser-only challenge. Review robots.txt and the CDN or WAF separately: an allowed robots rule does not help if infrastructure still returns 403.
Crawler controls are not one switch. For example, OpenAI documents OAI-SearchBot for search discovery separately from GPTBot controls. Decide deliberately which public paths each named crawler may request, and keep private paths protected.
02
2. Establish one clear URL for each important page
Use stable HTTPS URLs, a self-referential canonical where appropriate and a sitemap containing the canonical pages you want discovered. Keep redirects, canonical declarations and sitemap URLs consistent.
Do not use robots.txt as a substitute for noindex or authentication. Crawling, indexing and access control solve different problems.
03
3. State the business entity in visible content
The page should say who the business is, what it offers, for whom, where it operates and how a customer takes the next step. Keep the public name, location, contact paths and core offer consistent across the homepage, contact page and relevant service pages.
Add truthful structured data that represents the visible page. JSON-LD can make entities and relationships more explicit, but valid markup does not guarantee a rich result, recommendation or ranking.
04
4. Write blocks that can stand on their own
Use descriptive headings followed by a direct answer. Add concrete scope, conditions, location, prices or dates when they are verified and relevant. A machine should not need decorative layout or three previous sections to understand the claim.
Treat citability checks as versioned heuristics, not proven ranking factors. They help identify ambiguous blocks; only later measurements can show whether observed visibility changed.
05
5. Verify the exact fix
Record the URL, observed fragment and rule before editing. Preview the proposed change, publish it through the site's normal owner-controlled workflow, then run a new readiness check. Preserve the old result as the baseline.
A readiness increase means the site meets more of the disclosed technical and content criteria. It does not by itself mean AI systems now mention the brand more often. That conclusion requires a separate measurement cycle with the same Configuration Lock.
- Check up to five key pages rather than only the homepage.
- Keep llms.txt diagnostic-only; Selena gives it zero scoring weight.
- Do not expose private accounts or add paid provider calls to a free readiness check.