One meta tag flickered and deindexed 3,000 pages
A docs site lost 3,000 pages from the index over a month. No code deploy explained it. The clue: a noindex tag that existed for 200ms. Here's the flicker-hunt.
— I loaded a 'lost' page and watched the <head> in real time via a MutationObserver snippet in console.
— On mount, the default template injected <meta name="robots" content="noindex">, then the route's effect replaced it with index — but only after data loaded.
— Googlebot's render sometimes snapshotted during that noindex window.
The head-management checklist I now enforce:
— Never default to noindex and flip to index in JS
— Set robots in the server response, not client effects
— If using react-helmet/next/head, render the final value server-side
— Verify in Search Console's rendered HTML, not the browser
We moved robots to the SSR payload, defaulting to index.
Result: deindexed pages recovered 3,000 → fully restored in 6 weeks, and a recurring 'Excluded by noindex tag' in Coverage dropped from 3,100 URLs to 4.
Hydrate Diaries
@HydrateDiaries
One meta tag flickered and deindexed 3,000 pages
Этот пост опубликован в Telegram-канале Hydrate Diaries. Подписаться можно по ссылке: @HydrateDiaries.