The benchmark is intentionally small. It is meant to create a useful pre-launch or handoff note, not a long SEO export no one reads.
Root llms.txt is missing or too thin
The public root file is missing, blocked, or does not point answer engines toward priority pages and locale groups.
Why it matters: AI-search crawlers have weaker public context for the pages a client expects to be cited or summarized.
Localized JSON-LD lacks language context
Schema appears on the page, but language fields, locale-specific names, or localized descriptions are missing.
Why it matters: Search and answer systems can flatten localized pages into the default-language story.
CMS metadata drifts across locales
Translated CMS pages still share repeated titles, summaries, descriptions, or empty fields from the default locale.
Why it matters: The site looks translated to visitors while public metadata still looks unfinished to crawlers.
Preview QA is treated as launch QA
Teams review inside Framer or staging, but skip a final check against the published production domain.
Why it matters: Client handoff misses the exact public state that search engines and answer engines can actually fetch.