Lighthouse scores, Core Web Vitals, LCP improvements — these are real and worth caring about for human users. But they have almost nothing to do with how bots index your site.
Bots are not browsers. They don't render pages, they don't measure paint timing, they don't care how fast your font loads. Optimizing CWV for bot traffic is, at best, wasted effort. At worst, it gives you false confidence that your pages are discoverable when they're not.
What bots actually measure
When Googlebot crawls a page, it's looking for: indexable HTML content, a valid title and meta description, a canonical URL, structured data, and signals that the page is not being blocked (robots.txt, noindex, redirect loops).
Word count matters more than most people expect. A product page with fewer than 50 words of visible text is unlikely to rank for anything meaningful. Googlebot doesn't give you credit for the 1,200-word description that loads after React hydrates.
The thing that kills indexing
Empty HTML shells. This is the single most common reason JavaScript sites end up stuck in "Discovered — currently not indexed." The crawl succeeds, returns a 200, and the page has a title. But the main content is missing because it's rendered by JavaScript that bots never execute.
CWV optimization does not fix this. Adding more alt text does not fix this. Improving TTFB does not fix this. The fix is getting the content into the initial HTML response.
What actually moves the needle for bots
- Content parity: the HTML bots receive should have the same text your customers see
- Title and meta description present and within length limits
- Canonical URL defined (prevents duplicate content issues)
- Structured data (JSON-LD) for products, articles, or whatever fits your page type
- No noindex or X-Robots-Tag blocking the page
- No redirect chains longer than 2 hops
Run the site audit below on one of your product pages. Look at the structural issues, not the performance score.
Try it on your site
Full tool with more details