wiseseo:~/news/mueller-200-status-during-outage-503 $

date
category
[CRAWLING]
impact
MED
status
STATEMENT
read
4 min

Mueller: a 200 'back soon' page during an outage can cost you indexed URLs

John Mueller warns a 200 status on a downtime page can get the error text indexed and pages folded together as duplicates. Use a 503.

Google's John Mueller has restated one of the oldest rules in outage handling, and this time he spelled out exactly what goes wrong when it is ignored. In a LinkedIn comment reported by Search Engine Roundtable on 17 November 2026, he was asked whether a maintenance page that returns a 200 status with a "back soon" message is risky if it only lasts a few hours, or only during longer outages.

His answer listed two separate problems. First, the page "can get indexed with whatever 'try again later' text is visible," which he said fixes itself once the real content returns. Second, it "can get dup-eliminated" alongside other pages showing the same text, and that part is slower to undo, because "duplicates tend not to get crawled as often."

Why the second problem is the expensive one

A short-lived indexed error message is embarrassing but self-correcting: Google recrawls, sees the real page, and updates the snippet. Duplicate elimination is different. When every URL on a site serves the same "we'll be right back" body with a 200, Google can treat those URLs as copies of each other and keep only one. Getting the rest "un-deduplicated" depends on Google recrawling them, and Mueller's point is that URLs already judged to be duplicates are crawled less often. So a few hours of a bad maintenance page can turn into a much longer recovery for deep pages that are not crawled frequently anyway.

This is not a new position. Mueller made the same argument in a 2013 post, still published on his site, saying an error page with "200 OK" leads Google to index the changed content and, if every page returns the same error page, to "assume that these URLs are duplicates." What the new comment adds is the explicit asymmetry: the indexed-text problem clears on its own, the duplicate problem lingers.

What Google's documentation says

Google's HTTP status code documentation explains why 503 is the safer signal. For 5xx and 429 responses, Google's crawlers slow down, already indexed URLs "are preserved in the index, but eventually dropped," and any content served with a 5xx is ignored. When the server returns to 2xx, Google gradually raises the crawl rate again. The same page notes that URLs which persistently return a server error are removed from the index, so 503 is a short-term tool, not a parking state.

For 200 responses, the documentation says Google passes the content to indexing, and if that content looks like an error or an empty page, Search Console reports a soft 404.

Google's guide to pausing a site draws the time line clearly: to disable a site urgently for 1-2 days, return an informational error page with a 503 instead of the content; for longer periods, it suggests an indexable placeholder home page served with a 200. It also says disabling a whole site is an extreme step that should last a few days at most.

Confirmed vs. speculation

Confirmed by Google documentation: 5xx content is ignored, indexed URLs are kept for a while during 5xx errors, persistent server errors lead to removal, and 503 is the recommended response for a short outage. The duplicate-elimination risk and the slower recovery come from Mueller's statements, not from documentation. Mueller did not give a number of hours after which a 200 maintenance page becomes harmful, and we do not know of any published threshold.

what to do

  • Check what your maintenance or outage page actually returns. Request a few URLs during a planned window (or on staging) with curl -I and confirm the status is 503, not 200 or a redirect to a single holding page.
  • Look at your CDN, load balancer and hosting "maintenance mode" settings. Many default to a branded page served with 200; switch them to 503, optionally with a Retry-After header.
  • Make sure robots.txt keeps returning its normal content during maintenance rather than the outage page.
  • After any outage, open Search Console's Page indexing report and watch "Soft 404" and "Duplicate without user-selected canonical" / "Duplicate, Google chose different canonical than user" for spikes dated to the outage.
  • Use URL Inspection on a sample of important URLs to confirm the Google-selected canonical points to the page itself and the last crawl shows real content.
  • In Settings > Crawl stats, check the response breakdown for the outage period: a burst of 503 is expected; a stretch of 200 with tiny page sizes is a warning sign.
  • Keep any full-site disablement to a day or two. For longer shutdowns, follow Google's placeholder home page guidance instead.

For more on how Google treats URLs it decides not to keep, see our brief on why shrinking a site will not fix a mass not-indexed problem. For URLs Google has found but not crawled, see our guide on Discovered - currently not indexed.

refs

  1. [1] HTTP status codes, network and DNS errors and Google Search (Google Search Central)https://developers.google.com/search/docs/crawling-indexing/http-network-errors
  2. [2] Temporarily pause or disable a website (Google Search Central)https://developers.google.com/search/docs/crawling-indexing/pause-online-business
  3. [3] John Mueller: 503shttps://johnmu.com/503s/
  4. [4] Google On Why Serving A 200 Status Code When Your Site Is Down Is Bad (Search Engine Roundtable)https://www.seroundtable.com/google-200-status-code-site-down-42542.html