SquishPixel Blog · SEO · Front-end · Web performance

A React site fixes thin pages and empty HTML after Google indexes only 30 of 547

According to DEV Community, a developer who runs GAGPedia, a fan-made calculator and guide site for the Roblox game Grow a Garden, found that Google had indexed only 30 of its 547 pages after eight months. Search Console listed 508 pages as "Discovered – currently not indexed". Bing, whose index also feeds Yahoo and DuckDuckGo, sent roughly ten times more search visits than Google.

The first problem was thin, repeated content. The site made a page for each of 372 pets and 115 plants from one template. Each page had about 325 words, with only one or two unique sentences. Two pages compared were 80% identical. The author kept the pages, because Bing ranks them and players use them, and added a noindex rule that only Googlebot obeys. Links on those pages still count. The sitemap shrank from 547 URLs to 60.

The second problem was that the guide articles, 1,000 to 3,000 words long, looked like about 125 words to a crawler. The static HTML held only a short summary, and the full text appeared after JavaScript ran. The author now renders the real React components to HTML at build time, with a fallback to the old summary if rendering fails. Guides now ship 800 to 2,400 words of real HTML. This caused a brief flash of unstyled static content on reload, which two lines in the page head now hide for browsers that run JavaScript.

Other fixes followed. Loading the Supabase client on demand cut the main bundle from 210 KB to 152 KB gzipped. IndexNow, a protocol for telling search engines about new URLs, now pings Bing after each production deploy. A broken audit script, which wrongly flagged links on Windows because of backslash paths, was fixed and exposed one real bug: every plant page linked to a missing page. The author says it is too early to see results.

Why it matters

Google's "Discovered" and "Crawled – currently not indexed" labels can point to quality problems, not crawling faults. Pages generated from one template with little unique text may not earn a place in the index, especially on a young site with few backlinks.

The case also shows that the HTML a crawler first fetches can differ from what visitors see. If your content only appears after JavaScript runs, a search engine may judge a much shorter page than you think you published.

Source: DEV Community. This brief summarises their reporting; read the original for the full detail.