Google Says Canonical Re-Evaluation Can Take Up to Two Weeks

Google's canonicalisation troubleshooting doc has a line in its "Things to keep in mind" section that's worth flagging:

"Re-evaluation takes time: Even after fixing content issues, Google might hold pages in a duplicate cluster for up to two weeks."

That's a real number in an official Google doc, not a figure someone extrapolated from forum posts and Search Console screenshots. It's specific to one situation, so it's worth being precise about what it covers before you start quoting it to clients.

The problem it's answering

Google groups pages with the same or very similar main content into what it calls a duplicate cluster, then picks one URL from that cluster to actually show in search results. The rest get folded in as duplicates. Most of the time Google picks a sensible URL. Sometimes it doesn't: a print-friendly version outranks the real page, a parameter-heavy URL beats the clean one, or an international subdomain cannibalises the main domain.

When that happens, the fix is usually to make the pages genuinely distinct, rewritten content, a clearer unique value on each page, not just swapped paragraph order. What's never been clear is how long to wait after making that fix before you can tell whether it worked.

What the two weeks actually covers

Google's wording ties the two-week figure to content fixes specifically: pages held in a duplicate cluster because their content was too similar, then differentiated. It's the ceiling, not a guarantee, and Google notes elsewhere in the same doc that pages split out of a cluster faster when the new content is clearly and significantly different. A thin rewrite that just reorders sentences won't necessarily get the full two weeks, it may just stay clustered.

One thing worth being careful about: Google's doc doesn't explicitly say redirects or rel=canonical tag corrections follow a different timeline, it just doesn't mention them in that line at all. Those are picked up on Google's next crawl and re-render of the affected URLs, which is a different mechanism to content re-evaluation, so treating them as separate is a reasonable inference. It's our reading of how the mechanics work, not a direct quote from Google, so don't attribute that specific split to them if you're citing the source.

How we're using this

Setting client expectations. "How long until this is fixed" used to be unanswerable. Now there's a real, linkable source: up to two weeks for a content-based fix, and we'll check back once that window has passed rather than daily.

A quality bar, not just a clock. Since distinct content splits out faster, the two-week window is also a decent test of whether the rewrite was good enough in the first place. If a human reader wouldn't see two clearly separate, useful pages, Google likely won't either, and the fix will need another pass regardless of how long you wait.

Not a reason to stop checking technical fixes properly. If the actual problem is a redirect loop, a wrong canonical tag, or a server misconfiguration, this two-week figure isn't the right tool, and quoting it there would be misapplying someone else's timeline to a different kind of fix.

The part worth remembering

Google's canonicalisation guidance opens by pointing out that its pick might genuinely be the better one, and asks you to consider whether your preferred URL is actually the best result for users before fighting to change it. The two-week number is useful for planning. It doesn't change the underlying fact: you get a vote on which URL is canonical, Google gets the final decision. The only way to make that vote count is content that's obviously, unarguably distinct, to a reader and to a crawler.

FAQ