Automated technical SEO checks that catch indexing blocks and deploy regressions — seotto.com
You add a canonical tag. It points where you want. And Search Console's page indexing report says:
Duplicate, Google chose different canonical than user
In other words: Google saw your rel="canonical", considered it, and picked a different URL as the representative version of this content. The page you inspected won't be shown in search; the one Google chose will.
This feels like Google ignoring you. It's more accurate to say your site is sending mixed signals, and the canonical tag lost the vote.
How Google chooses a canonical
rel="canonical" is one signal among several. Google also looks at:
- Redirects: if URL A redirects to URL B, B is a strong candidate.
- Internal links: which version do your own pages link to?
- Sitemaps: which version is listed?
- HTTPS vs HTTP: HTTPS is preferred.
- Hreflang annotations: which URLs reference each other?
- Content similarity: are the pages really duplicates, or just similar?
- URL cleanliness: shorter URLs without parameters tend to win ties.
When most of these point at one URL and your canonical tag points at another, Google goes with the majority.
Step 1: find out which URL Google chose
Use the URL Inspection tool on the affected URL. Under Page indexing, it shows two values:
- User-declared canonical: what your tag says.
- Google-selected canonical: what Google picked.
Write down both. The fix is always about understanding why the second one looks more canonical than the first.
Step 2: compare the two URLs
Fetch both and compare what you're telling Google about each:
for u in "https://www.example.com/shoes/red" "https://www.example.com/shoes/red?color=red"; do
echo "== $u"
curl -sIL "$u" | grep -iE '^(HTTP|location|link)'
curl -sL "$u" | grep -ioE '<link[^>]+rel="canonical"[^>]*>|<meta[^>]+name="robots"[^>]*>'
done
Then check:
- Does each page's canonical tag agree with the other's? If page A says "canonical is A" and page B says "canonical is B", for near-identical content, Google has to pick one, and it might not pick yours.
- Which one is in your sitemap?
- Which one do your navigation, breadcrumbs and in-content links use? Search your templates for the URL pattern.
- Does one of them redirect to the other in some cases (with or without a trailing slash, with
www, with a parameter)?
The usual causes
1. Internal links point at the "wrong" version. The most common one. Your canonical says /shoes/red, but category pages link to /shoes/red?ref=nav or /shoes/Red. Internal links are a strong signal. Fix the links, not just the tag.
2. The sitemap lists a different URL. Sitemaps generated from the database often include parameter variants or old slugs. Every sitemap URL should be the canonical version, exactly.
3. Pages that aren't really duplicates are canonicalised together. If you canonicalise all colour variants to one product page, but each variant has unique content, Google may decide the variant pages are distinct, or pick one of the variants as the representative. Canonicals should join genuinely duplicate content.
4. The canonical points to a page that redirects, 404s or is noindexed. Google won't use a canonical target it can't index. Check that the target returns 200, is indexable, and itself carries a self-referencing canonical.
5. Canonical tags added by JavaScript, or conflicting tags. If the server HTML has one canonical and JavaScript changes it, or the page has two canonical tags (one from a plugin, one from the theme), Google may ignore both.
6. Cross-domain duplication. If your content is syndicated or scraped, Google may choose the other domain's copy. Ask syndication partners to add a canonical back to your original, and make sure your page was published first and is easy to crawl.
Canonical problems usually start with a template or sitemap change, long before Search Console reports them. seotto.com watches for canonical and indexing inconsistencies after you deploy, so mixed signals get caught while they're still one commit old.
Step 3: line up the signals
Pick the canonical you want, then make every signal agree with it:
- Self-referencing canonical on the chosen URL; canonical to it on every duplicate.
- Only the chosen URL in the sitemap.
- All internal links use the chosen URL, exactly, including case and trailing slash.
- Where duplicates have no reason to exist (case variants, old slugs, tracking parameters), 301-redirect them instead of relying on canonical tags.
- Then request reindexing of the chosen URL in URL Inspection, and wait. Canonical selection updates as pages are recrawled, which can take a few weeks.
When to let Google win
Sometimes Google's choice is fine, or even better:
- If Google chose the HTTPS version over an HTTP canonical you forgot to update, fix your tag to match.
- If it chose a cleaner URL over a parameterised one, consider making the clean URL your canonical.
- If the "duplicate" really is a duplicate and Google's choice is the one users would prefer, there's no ranking penalty for letting Google pick. The content still shows up in search; just under the other URL.
The status is a warning, not an error. It matters when Google picks a URL you don't want users landing on: a filtered view, a print version, a staging host, or a page in the wrong language.
Related reading
- Canonical tags FAQ: 10 questions developers ask about rel=canonical (with straight answers)
- There is no duplicate content penalty (but duplicate content still costs you)
- Case study (hypothetical): the staging site that got indexed, and started outranking production
- How to fix 'Discovered – currently not indexed' in Google Search Console (a developer's list of causes)
- How to catch SEO regressions before they ship: diff titles, canonicals and robots tags between production and a preview deploy (Python, for CI)
What's the most surprising URL you've seen Google pick as canonical? Mine was a ?utm_source=newsletter version that had more internal links than the clean URL, thanks to a template bug.
Top comments (0)