Google updated its canonicalization troubleshooting documentation this month and added a short new section at the top about timing. Barry Schwartz caught it at Search Engine Land on July 10.
The line that matters: after fixing content issues, Google might hold pages in a duplicate cluster for up to two weeks.
That is the whole announcement. No algorithm change, no penalty, no new tag to install. Just Google finally writing down how long the waiting part takes.
It is also probably the most useful thing Google has published all year, because it puts a number on the exact moment where most people ruin their own fix.
First, What Canonicalization Actually Is
Skip this if you already know. It will take thirty seconds.
When Google finds several pages that look basically the same, it does not index all of them. It groups them into what it calls a duplicate cluster, picks one to be the official version, and shows that one in search results. The chosen page is the canonical.
You get a vote. The rel="canonical" tag, your internal links, your sitemap, and your redirects all tell Google which page you would prefer.
Google treats your vote as a strong hint rather than a command, which is the polite way of saying it will overrule you if it disagrees. This is a common source of emotional distress among people who own websites.
The New Part: The Two Week Hold
Here is what Google added. Once you fix whatever caused the duplication, the pages do not immediately spring apart. Google may keep holding them in the same cluster for up to two weeks while it re-crawls, re-evaluates, and decides whether your change was real.
Google also noted that pages separate faster when the difference between the new content and the other clustered pages is clear and significant.
Read that second part twice, because it is doing a lot of quiet work. It is Google saying that a rewrite so subtle only you can detect it may not be enough to break the tie. If two pages were merged because they were basically the same, changing eleven words does not un-basically-the-same them.
Why This Is Genuinely Useful
The stated reason Google published the timing is to stop people from repeatedly changing pages before Google has had a chance to process the first change.
That behavior is extremely common and it goes like this.
Monday, you find the problem and fix it. Correct canonical tags, real content differences, the works. Good work, genuinely.
Thursday, you check. Nothing has changed. You assume the fix did not take.
Friday, you try a different approach. Maybe you noindex one of the pages. Maybe you 301 it somewhere. Maybe you rewrite it again, harder.
The following Tuesday, still nothing, so you do a third thing.
Now you have three overlapping changes at different stages of being crawled, and Google is evaluating a page that has been a moving target for eight days. You have not fixed anything three times. You have reset the clock three times and destroyed your ability to tell which of the three ideas worked.
Two weeks of doing nothing would have beaten all of it. This is deeply annoying to accept and it is true anyway.
What Causes This on Normal Business Websites
Google's docs list six causes, and a couple of them are exotic. Nobody reading this is losing sleep over hreflang annotations across regional domains.
Here are the ones that actually hit small and local businesses.
Location pages that are the same page wearing different hats
This is the big one. You have twelve service area pages. They are the identical eight paragraphs with the city name swapped out. Miami, Coral Gables, Hialeah, Doral, and so on.
You think you have twelve pages. Google may think you have one page and eleven impersonators, cluster them, and index whichever one it likes best. Then you wonder why the Doral page never ranks in Doral.
The fix is not a synonym pass. It is actual different content: different neighborhoods, different jobs you did there, different pricing realities, different photos, different questions people in that area actually ask. Clear and significant, per Google.
Your CMS doing something you never asked it to do
Google specifically calls out content management systems misapplying canonicalization. A theme update, a plugin, or an SEO plugin talking to another SEO plugin can start pointing every page's canonical at the homepage. Nobody notices for four months because the site looks completely fine to humans.
The same page living at multiple addresses
The classic set: www and non-www, HTTP and HTTPS, with and without a trailing slash, and every URL with a tracking parameter stapled to the end. Same content, several addresses. Usually a redirect and canonical problem rather than a content problem.
Server misconfiguration
Google mentions hosting errors causing unexpected cross-domain selection, or a server that returns the same near-empty page for lots of URLs, which Google reads as soft 404s. Staging sites left crawlable are a popular way to accidentally compete with yourself.
People copying you
Two flavors. Copycat sites lifting your content, where Google occasionally picks their version as canonical. And syndication, where you gave a partner your article and their copy outranks yours. Google's note here is worth knowing: the canonical tag is not the recommended way to handle syndication. Partners should block indexing of your content instead.
Hacking
Less common, more alarming. Attackers inject cross-domain canonical tags or redirects pointing at spam. If your pages are suddenly canonicalizing to a domain you have never heard of, that is not a duplicate content problem, that is a security incident.
What This Means If You Run a Local Business
Three practical things.
Your service area pages are the likeliest culprit. If you have city pages that are near-copies, that is where to look first, and the fix is real writing, not a rewrite tool. This is unwelcome news because it is work, but it is also the version that actually holds.
Two weeks is the normal cost of a fix, not evidence of failure. When your site changes and nothing happens for ten days, that is not necessarily broken. Google put the number in writing specifically so people would stop panicking on day four.
You can check it yourself in about a minute. Open Search Console, run the URL Inspection tool on the page you care about, and look at which URL Google says is the canonical. If it names a different page than the one you submitted, you have found your problem. You do not need to know what to do about it to know that it is happening.
What to Expect If You Pay an Agency
Two weeks is a real timeline. It is also a wonderful thing to hide behind, because it renews indefinitely if nobody is keeping dates.
"Give it a couple weeks" is a completely correct answer the first time. The second time it is a yellow flag. The third time about the same page, someone is either not checking or hoping you stop asking.
Four things to ask for. None of them require you to understand SEO.
- What date did the fix actually go live? The clock starts there, not from when it got discussed on a call.
- What does URL Inspection say Google picked as the canonical? This is a screenshot. It takes a minute to produce.
- What specifically changed, in plain words? "We improved the content" is not an answer. "We rewrote the Doral page with local job examples and different FAQs" is.
- Have we changed this page again since the fix? If yes, the two weeks restarted and everyone should know that.
A good answer sounds like: fixed on the eleventh, inspected on the twenty-fifth, Google is now picking the right canonical, here is the screenshot. That is it. That is the whole standard.
How We Stay On Top Of This
Mostly by writing things down and then having the discipline to sit still.
We log the date a fix ships, because a two week window is meaningless if nobody remembers when it opened. We change one thing at a time, so that when something moves we know what moved it. We re-inspect at the end of the window rather than every morning, because checking daily produces anxiety and no information.
We also read the documentation changes. Google updates these pages constantly and most updates are nothing, but occasionally they quietly add a number that saves everyone a month of guessing. This was one of those.
And when we fix duplicate content, we make the pages genuinely different, because Google now says out loud that clear and significant differences resolve faster. Doing the shallow version and then waiting two weeks to find out it did not work is the most expensive way to save an afternoon.
If your rankings are stuck and nobody can tell you which page Google actually has indexed, Miami Web Lab will look and tell you plainly. It is usually a canonical problem, a content problem, or an impatience problem, and all three are fixable.