Apprendre
Duplicate Content: What Google Actually Does About It
The duplicate content penalty does not exist. What exists is Google picking one version and ignoring the rest — and picking a version you did not intend.
Lancez un audit DomainLens frais et utilisez le rapport comme liste de priorités.
There is no penalty
Google has stated this repeatedly and it remains one of the most persistent SEO myths. Duplicate content does not trigger a penalty. What happens is duplicate detection: Google groups near-identical pages, picks one as canonical, and indexes that one.
The cost is real but different from a penalty. Signals split across several URLs instead of accumulating on one, and Google may pick a version you did not want — a parameter URL, a print view, or a syndicated copy on someone else's domain.
Where duplicates actually come from
Almost none of these are editorial. Duplicate content is overwhelmingly a URL problem generated by templates and parameters, which is why writing more content rarely fixes it.
| Source | Example | Fix |
|---|---|---|
| URL variants | www / non-www, http / https, trailing slash | Redirect to one form |
| Parameters | ?utm_, ?sort=, ?sessionid= | Canonical to the clean URL |
| Faceted navigation | Every filter combination | Index a chosen few, block the rest |
| Pagination | Page 2 canonicalised to page 1 | Each page self-canonical |
| Product variants | One product per colour | Canonical to the parent |
| Syndication | Your article on a partner site | Cross-domain canonical |
| Boilerplate | The same 300 words on 500 pages | Write distinct content |
Choosing the right tool
The mistake is reaching for canonical when the answer is a redirect. If nobody needs the old URL, redirect it — a canonical leaves it reachable, crawlable and available to be picked despite your preference.
- 301 redirect — when only one version should exist at all. The strongest signal.
- Canonical tag — when both must remain reachable but one should be indexed.
- noindex — when a page must exist for users and never appear in search.
- Parameter handling in the URL itself — strip what does not change content.
- Rewriting — when the pages are genuinely different and should stay separate.
Finding duplication that matters
The Google-selected canonical is the most reliable signal that duplication is affecting you, because it is Google telling you which version it chose. See canonical tag problems when the choice disagrees with yours, and URL Inspection for reading it.
- 1Crawl the site and group pages by identical or near-identical title and description; that catches most template-level duplication.
- 2Look for the same content at multiple URLs by comparing content hashes rather than reading pages.
- 3Check URL Inspection on a sample: a Google-selected canonical differing from yours names a duplicate you did not know about.
- 4Search a distinctive sentence in quotes to find external copies.
- 5Prioritise by value — duplication on your top category page matters, duplication among tag archives usually does not.
Boilerplate is not duplicate content
A shared header, footer and sidebar appear on every page and Google handles that without difficulty — it identifies the main content and evaluates that. Sites do not need to worry about navigation being repeated.
The line is crossed when the unique portion is small relative to the repeated portion. Fifty location pages differing only in a city name are duplicates in substance, and no canonical fixes that — the pages need genuinely different content or they need to not exist.
How DomainLens contributes
DomainLens reports the canonical a page declares and whether it resolves, plus duplicate title and description signals — the mechanical half of duplicate detection. Judging whether two pages are substantively the same is the half that needs a person. See thin content for the related question of whether a page has enough to say.
- Is there a duplicate content penalty?
- No. Google groups duplicates and indexes one version. The cost is split signals and losing control of which version is chosen, not a penalty.
- How much duplicate content is acceptable?
- There is no percentage. Shared navigation and boilerplate are fine. The problem starts when the unique part of a page is small relative to the repeated part.
- Canonical or redirect?
- Redirect when only one version should exist. Canonical when both must remain reachable but only one should be indexed.
- Does syndicating my content hurt me?
- Only if the copy outranks the original. Ask the partner for a cross-domain canonical or at minimum a link back to your version.