Part of the Technical SEO Audit series. For the full picture, start with Technical SEO audit: the complete guide and checklist.
A planned website outage SEO plan starts with one question: can the site stay readable while the affected feature is paused? A checkout freeze or deployment window does not always require turning every URL into a maintenance page. Google’s temporary site guidance prefers limiting functionality and keeping content available. This guide addresses the short maintenance window itself; SEO regression testing covers the wider release checks.
What is the safest option for planned maintenance?
Keep the website online and disable only the function that cannot work safely. For example, pause checkout while product and category pages remain accessible, then tell users when purchasing will resume. Google recommends this approach for temporary business interruptions because people and crawlers can still access useful information.
If you must take the whole site offline for a very short period, serve an informational page with HTTP 503 Service Unavailable and a reasonable Retry-After header. Google’s guidance describes whole-site shutdown as an extreme measure for a few days at most. A maintenance banner served with status 200 on every URL can make all those URLs look like the same content; a 503 makes the temporary condition explicit.
| Situation | Response strategy | SEO risk to control |
|---|---|---|
| One feature is paused | Keep public pages live; explain the limit | Stale offers or broken transactions |
| Short full-site maintenance | 503 page with Retry-After | Outage running longer than planned |
| Extended closure | Keep an indexable 200 homepage with clear information | Whole site disappearing from Search |
Do not treat this table as a promise of ranking continuity. Google’s systems can recover from short interruptions, but a long full-site outage can still affect crawling and indexing.
Which maintenance responses cause avoidable problems?
Google’s best-practice list warns against several common switches: sitewide noindex, a robots rule that disallows the entire site, and temporary use of permanent 403, 404 or 410 responses. These controls can remove URLs from Search or prevent Google from seeing the real pages when service returns.
Keep robots.txt accessible and valid during a 503 window. Do not return 503 for it. Also avoid a sitewide redirect to one unrelated maintenance URL. That can obscure which pages are temporarily unavailable and make restoration harder to verify.
If an application returns a friendly error screen with status 200, a browser screenshot will not expose the problem. Check the HTTP status, not just the visible page. The HTTP status code guide explains why a soft error and a true 503 send different signals.
How do you prepare and verify the outage?
Make a small URL sample before work begins: homepage, a category, a product or service page, robots.txt, an image and any critical API or checkout route. Save their expected normal responses. Agree on an owner, a rollback condition and the latest acceptable restore time.
During the window:
- Fetch the sample URLs and confirm the intended status, not only the maintenance page’s appearance.
- Keep
robots.txtavailable; avoid carrying stagingnoindexinto production. - If using 503, include a realistic
Retry-Aftervalue and a lightweight page with useful user instructions. - Monitor origin and CDN responses so a cached maintenance page does not outlive the deployment.
- After restoration, re-fetch the sample and check for 200 pages, intended redirects, canonicals and index controls.
Google notes that it cannot refresh a page’s titles, metadata or structured data while that page returns 503. Do not schedule a major metadata launch and assume Google will process it during the outage. If the maintenance accompanies a migration, also follow the site migration checklist for redirects and new URL verification.
What should you check after the site returns?
Compare the same URL sample with the baseline. Check that the CDN no longer serves cached 503s, robots.txt allows the intended crawling, and the real page content has returned. Review server errors and Search Console over the following days, but avoid treating one delayed report as proof the outage broke rankings.
If the outage lasted longer than planned, check important pages with live URL Inspection and investigate persistent 5xx patterns. Google’s crawl-error guidance explains that prolonged unavailability can cause Google to slow or stop crawling affected URLs.
Planned outage SEO checklist
- Keep public content online with reduced functionality if possible.
- If a brief full shutdown is unavoidable, return 503 and Retry-After.
- Keep robots.txt available; do not ship sitewide noindex.
- Verify status codes from outside the browser UI during maintenance.
- Restore normal responses promptly and clear cached maintenance pages.
- Run post-outage HTTP and indexability checks on critical templates.
Frequently asked questions
Should a maintenance page return 200 or 503?
For a brief whole-site outage, Google's guidance recommends an informative 503 response, not a normal 200 response pretending the maintenance message is the page. If a longer closure is unavoidable, Google suggests keeping an indexable 200 homepage rather than disabling everything.
Should robots.txt return 503 during maintenance?
No. Google's temporary-site guidance says to keep a valid robots.txt available and not return 503 for that file. A robots failure can interfere with crawling beyond the page you meant to pause.
Does Retry-After guarantee Google will preserve rankings?
No. Retry-After tells clients when you expect service to resume, but it does not guarantee a crawl schedule or search outcome. Keep the outage short, send the correct status and verify recovery.
Can I add noindex to every page while the site is down?
Do not do that for a temporary outage. Google warns that sitewide noindex and permanent 403, 404 or 410 responses can remove URLs from Search. Restore access rather than using index controls as a maintenance switch.