Written by 4SYSTEMPublished on Updated on
9 min read
The canonical tag tells Google which URL you prefer for duplicate or very similar content. It does not replace a redirect and is not an exclusion directive. To avoid hiding a useful page, first compare the contents, then choose the treatment, then control the signals sent by the site.
A canonical is a signal, not an order. Keep useful variations accessible, redirect pages that are actually replaced, and reserve noindex for pages whose results you want excluded.[24]
Compare contents before URLs
Two addresses that talk about the same subject are not necessarily duplicates. A guide on choosing a material and a sheet presenting its maintenance constraints can meet different needs. Designating one as canonical to the other would express a consolidation preference unsuitable for these distinct needs, without imposing this choice on Google. Google presents canonicalization as a treatment of duplicate or very similar pages, not as a thematic grouping.[24]
For each suspicious set, note the full URL, parameters, language, title, main content and function for the visitor. Add HTTP status, indexing directives, declared canonicality, incoming internal links and presence in the sitemap. Also note the hreflang annotations when they exist.
The comparison must focus on what the reader obtains: text, proposed references, availability of information and possible actions. A different layout does not necessarily mean different content. Conversely, a seemingly technical parameter can modify a product selection. Note the observed equivalence and its justification before choosing a target.
Canonical tag, redirection or noindex: the tree of choice
Here is a choice tree proposed for editorial and technical arbitration. It translates the documented distinctions, but its acceptance testing criteria constitute a working method, not additional rules of Google.
- Is the page permanently replaced? If its useful content now exists at another address and the old one no longer has a role, choose a permanent redirection to this relevant replacement. Google recommends permanent server-side redirection for sustainable movement.[25]
- Should the variant remain accessible? If it has equivalent content, indicate the preferred version with a canonical. The visitor retains access to the variant; you express a preference for the search, without imposing the final choice on Google.[24]
- Should the page be excluded from the results? Retain noindex for this purpose, not to designate another URL. Google deprecates noindex as a way to choose a canonical on the same site, because it blocks the page in search.[24]
- Is the content distinct and intended for search? Keep a standalone page, with a consistent self-referencing canonical. Do not force its consolidation to a simply neighboring page.
For removed pages, the redirection plan 301 helps document the destination without defaulting readers to home.
A final, indexable and correctly declared target
In this method, the chosen target must be an absolute, final and indexable URL: it displays the expected content, without new redirection or noindex directive. Reject a canonical to an 404 error. If the target has disappeared, look for a real equivalent; do not invent a replacement from a neighboring theme.
For a HTML page, place the element in a <head> valid. Google recommends an absolute URL rather than a relative path and a self-referring canonical on the preferred page.[24] Example of declaration for our fictitious catalog:
<link rel="canonical" href="https://catalogue.example/guides/choisir-matiere">
Choose a single declaration path, HTML or HTTP header, to limit contradictions. Google accepts both, but recommends retaining one, their combination being more prone to errors.[24] Also inspect the server response: an extension and a hosting setting can send two different preferences.
If the site uses JavaScript, compare the source code to the rendered HTML. Google recommends declaring the canonical in the source HTML and not modifying it with JavaScript.[24] A correction of the template is not enough if a script then rewrites the target.
Parameters, filters and languages: processing useful variants
Order settings by their actual effect, not by their name. A tracking parameter can leave the guide unchanged: the variant can then designate the own URL. A sort or filter can modify the available elements. In this case, compare the main content before deciding if the pages are close enough.
For a variant dedicated to internal use, the exclusion may be relevant if you do not want its presence in the search. However, it does not become a duplicate of the general category. Therefore separate the objective of visibility from the judgment of equivalence.
Also align languages. With hreflang, Google requests a canonical in the same language, or the best possible substitute when it does not exist.[24] In your acceptance testing, favor final linguistic URLs that are consistent with the links and the sitemap. Do not make all the translations converge into French just because they cover the same subject.
A fictional catalog, explicit decisions
The following catalog is entirely fictitious. It offers a guide to choosing materials, accessible at its own address and via a campaign link. An old advice page has been replaced by this guide. These situations illustrate decisions, not results obtained.
https://catalogue.example/guides/choisir-matiere: reference page preserved, canonical to itself, internal links and sitemap aligned to this address.https://catalogue.example/guides/choisir-matiere?utm_source=lettre: content assumed identical after comparison, canonical to the own guide; the variant remains accessible for arrival from the countryside.https://catalogue.example/conseils/matieres: useful content included in the new guide, permanent redirection to it, without detouring through the reception.https://catalogue.example/selection/compare: personalized comparison tool, not equivalent to the guide; noindex if the editorial decision is to exclude it from the search.
The editorial manager validates the equivalence and replacement here. The technical team then verifies that the actual behavior corresponds to the recorded choice.
Keep a record of contradictions
An isolated signal is not enough to explain an unexpected canonicality. Google qualifies the redirection and rel="canonical" as strong signals, the inclusion in the sitemap as a weak signal, and asks not to send contradictory preferences.[24] The register below extends the fictitious example with hypothetical defects.
| URL or variation | Supposed contradiction | Correction decided | Expected proof |
|---|---|---|---|
| Guide with tracking | Sitemap parameterized, clean canonical | Remove sitemap variant | Export of the sitemap containing the own URL |
| Old page | Redirection to home, replacement identified | Redirect to guide | Response HTTP with the correct destination |
| Clean Guide | Canonical to old URL redirected | Install a self-referencing canonical | Source and rendering designating the guide |
| Custom tool | noindex and canonical to the non-equivalent guide | Keep the exclusion, remove this consolidation | Guidelines and content verified |
Add to each line the person responsible for the correction, the date of the check and the document kept. Do not mark a line “validated” solely on the basis of a setting recorded in CMS: check the response actually served.
Check in the correct order and keep the proof
Execute the checks in this order to avoid validating a markup placed on the wrong destination:
- Open source and target. Identify HTTP responses and main content. Eliminate errors, loops and destinations that do not meet the identified need.
- Check the desired indexability. Examine the target's indexing directives and its accessibility to the robot. Do not block a URL in robots.txt to choose its canonical: Google can index a banned URL without its content.[24]
- Read the declarations. Compare canonical of HTML, possible header HTTP and rendering JavaScript. The destination must match the decision, including the protocol, host, and parameters retained.
- Realign discovery paths. Fix internal links, sitemap and language annotations. Google recommends linking the canonical rather than a duplicate in the internal linking.[24]
- Preserve the evidence. Archive response HTTP, markup extract, content comparison and decision signed by the maintainer. If the canonical retained by Google is observable in your tools, record it separately from that declared.
A compliant technical acceptance testing does not prove that Google will adopt your preference. If the observed choice differs, review the register and equivalence of the content before adding tags.
Frequently asked questions
How to process a PDF equivalent to a HTML page?
For a PDF, the HTML element is not suitable. Google accepts a HTTP header Link with rel="canonical" for non-search-supported HTML documents.[24] If the PDF and web page are truly equivalent, document the preferred version, then check the header served by the file URL and the accessibility of its target.
Is a permanent redirection necessary during maintenance?
No, if you plan to revert to the original URL. Google distinguishes between temporary and permanent redirects: the former do not serve as a signal to make their target canonical.[25] Choose the mechanism according to the actual duration of the change, keep the decision in the register and provide control of the return to the initial page.
Can we protect a reserved guide with noindex?
No. A noindex directive concerns presence in search results, not permission to view the document. For a reserved guide, have access controlled on the application or server side. Don't treat a canonical tag as a protection: an alternative URL can still open content if there are no access restrictions preventing it.
Take action on a set of variants
The next action is to choose a set of variants, name its reference page and populate the registry before any global modification of CMS. Have the contents validated, apply the decided treatment, then keep the proof of acceptance testing. If several templates or teams are involved, the web services and SEO of 4SYSTEM can accompany this framing and technical controls.
