First, confirm which copy search prefers
Three checks settle the question. Search for an exact sentence from the article in quotation marks and note which address appears first. Open the search performance report for the original's URL and compare impressions for the article's main queries before and after the copy went live; a fall that starts on the copy's publication date is the signature of this problem. Then inspect the original's URL: the crawling and indexing guide explains that URL inspection shows both the canonical you declared and the one Google selected, and a selected canonical on the partner's domain is direct evidence that search treats the copy as the main version.
Then read the copy itself as a crawler would. Open the partner's page source and note any canonical element and any robots meta tag, check the response headers for a robots directive, and check the partner's robots.txt for a rule covering the page. Compare the copy's title, headings and length with the original. With those facts written down, the table below names the cause; the options for protecting an original, and when each suits, are set out in the guide to syndicating and republishing content and are not repeated here.
Diagnosis table
Diagnosis
| Symptom | Check | Likely cause | Fix |
|---|---|---|---|
| The copy was found and indexed before the original existed | Publication dates and first-crawl dates of both pages | The partner's copy went live before the original | Publish first on your own site; for live copies, add a directive |
| A canonical points at the original, yet the copy is shown | The copy's title, sections and length against the original | A canonical on a copy edited until it is no longer a duplicate | Restore a faithful copy, or replace the canonical with a noindex |
| A noindex is on the copy, yet it still appears | Whether a canonical sits beside it, or robots.txt blocks the page | Canonical and noindex on the same copy, or a robots.txt block hiding the noindex | Keep one directive, and let the crawler fetch the page |
| The copy carries no canonical and no noindex | The copy's source and headers for any directive at all | Only an attribution link, which says nothing about which version to prefer | Ask for a canonical or noindex; otherwise stop syndicating that piece |
| Directives are correct, yet the copy still wins | Internal links pointing at the original on your own site | The original weakly linked within its own site, so the copy carries more signals | Link the original from hubs and related pages with descriptive anchors |
Fix, verify and prevent each cause
A copy that existed first
If the partner published before you did, the copy was the only version for a while, and search had no original to prefer. The syndication guide's written terms put timing first for that reason: the original goes live on your site before any copy appears.
Fix: for a copy already live, the timing cannot be undone, so the remedy is a directive on the copy that search can read; ask the partner for a canonical on a faithful copy or a noindex, as the agreement allows.
Verify: the exact-phrase search starts returning the original, and the original's impressions for its main queries recover.
Prevent: write the publication order into every agreement, and send the partner the original's live address rather than a draft.
A canonical on a copy that is no longer a duplicate
A partner that retitles the article, adds sections or cuts it down has made a different page, and the canonical on it describes a relationship that no longer exists. The canonical tag versus permanent redirect comparison records that a canonical is a hint weighed against other signals, and that a duplicate whose content differs substantially may be treated as its own page. Google's documentation on consolidating duplicate URLs is the source those pages draw on.
Fix: either ask the partner to restore a faithful copy, moving its additions to a separate article, or replace the canonical with a noindexA robots directive telling search engines not to show a page in their results. that does not depend on the pages matching.
Verify: URL inspection of the original shows your own URL as the selected canonical again.
Prevent: agree in writing what the partner may change, and review the copy whenever the partner redesigns its templates.
Directives that contradict or cannot be read
Two set-ups look protective and are not. A copy carrying both a Canonical TagTells search engines the preferred URL. and a noindex sends signals the crawling and indexing guide describes as contradictory. A copy with a noindex behind a robots.txt rule is never fetched, so its noindex is never read; Google's help page on blocking indexing with noindex explains that the crawler must be able to access the page to see the directive, and the noindex versus robots.txt comparison adds that a blocked but linked page can still be indexed from links alone.
Fix: keep one directive on the copy and remove any robots.txt rule that stops it being fetched.
Verify: the copy's source shows a single directive and the partner's robots.txt allows the page.
Prevent: add "one directive, crawlable" to the checks you run when each copy goes live.
Nothing but an attribution link
A credit line and a link back are the minimum in any arrangement, and they help readers find the source. They do not tell search which version to show, so a copy on a larger, busier site may simply win. The syndication guide is plain that when the link is the only protection, the copy may be the version shown.
Fix: ask the partner whether its system can add a canonical or a noindex; if it cannot and the article matters for search, end syndication of that piece or replace the copy with an excerpt that links to the original.
Verify: either a directive appears on the copy, or the full copy is replaced and the original regains its queries.
Prevent: confirm before agreeing which directive the partner's system supports, and decline full copies of search-critical pieces where it supports neither.
An original its own site ignores
Sometimes every directive is right and the original still looks weaker. A copy linked from a busy partner's homepage and newsletter, against an original reachable only through an archive page, gives search a lopsided picture. The internal linking guide explains that pages linked from important pages are found sooner and refetched more often, and that internal links distribute the authority the site earns from each BacklinkA link from another site to yours..
Fix: link the original from its hub, from related articles where the topic comes up and from the newsletter, with anchors that describe it.
Verify: a crawl shows the original's inlinks rising, and the original's impressions recover for its queries.
Prevent: before any piece is offered for syndication, check that it is already well linked on your own site.
A prevention routine for syndicated pieces
Keep a register of every syndicated article: the original's address and publication date, the partner, the copy's address, the directive agreed, the date the copy went live, and the date it was last checked. Publish on your own site first, link the original into its hub and related articles, and only then send it to the partner. On the day the copy appears, read its source and headers and the partner's robots.txt, and record what you found in the register.
Each month, run the three confirmation checks for every live copy: an exact-phrase search, the original's search performance for its main queries, and URL inspection of the original. Re-read the copy's directives whenever the partner changes its design, since a new template is the commonest way for a canonical to disappear. When a partner asks to edit, extend or retitle a piece, treat it as a change of arrangement: agree a noindex or a separate article before the edit goes live, not after the copy starts appearing in your place.
Finally, end arrangements that cannot be protected for pieces that matter in search, and keep syndicating only the pieces where reach is the point. Copies made without permission are not part of this routine; they belong to the spam policies guide. The content troubleshooting section lists the neighbouring symptoms, such as an archive losing traffic over many months, that have different causes.
Pibblesprout Home Brewing wins back its own guides
Pibblesprout Home Brewing, an invented home-brewing supplier, lets a hobby magazine republish six of its brewing guides under a cross-domain canonical. Within two months, three of the magazine's copies appear for the guides' main queries instead of the originals. An exact-phrase search returns the magazine first for all three, the originals' impressions for their main queries have fallen from about 4,800 a week to 1,300, and URL inspection shows the magazine's address as the selected canonical. Reading the copies shows why: the magazine retitled all three, added reader tips and an equipment box, and roughly doubled their length. The other three copies, left as published, still defer to the originals. Pibblesprout and the magazine agree to switch the edited copies to a noindex and to allow the crawler to fetch them. Pibblesprout also links each original from its brewing hub, where they had been reachable only through a dated archive. Six weeks later the originals appear for their queries again, and weekly impressions are back to about 4,600.
Common mistakes
Asking for a canonical and then letting the partner rewrite the piece. Stacking a canonical and a noindex to be doubly safe. Blocking the copy in robots.txt. Treating an attribution link as protection in search. Sending the partner a piece before it is live on your own site.
