Google’s Original Content Systems: Why the Source Should Outrank the Copy, and When It Doesn’t

Few things frustrate publishers more than seeing their own work rank below someone else’s copy of it. It feels like the system has failed.

Sometimes it has. More often, the original never made itself clearly identifiable as the original. Understanding what Google looks for turns this from bad luck into something you can manage.

What Google says

Google’s guide to its ranking systems describes systems that help ensure original content is shown prominently, including original reporting, ahead of pages that merely cite it.

It also points to canonical markup: a tag creators can use to tell Google which page is the primary version when the same content appears in several places. Originality is something Google tries to detect, but it also accepts help in identifying it.

The 2019 change for original reporting

On 12 September 2019, Google’s Vice President of News, Richard Gingras, announced changes to elevate original reporting. Google said it had updated its ranking and its rater guidelines to recognise original reporting better, show it more prominently and keep it visible for longer.

• The problem it addressed: a scoop would generate dozens of follow-up stories, and the latest or longest version tended to replace the story that started it.
• The rater guidance: raters were told to give the highest rating to reporting that provides information that would not otherwise have been known had the article not revealed it.
• Reputation: raters were also asked to consider a publisher’s overall record of original reporting, including recognition such as major journalism awards.
• Scope: the change applied across Search, News and Discover.

Discover-specific recovery for publishers is covered in news publisher Discover recovery.

What counts as original

Originality is not limited to journalism. Any page that contributes something that did not exist before can qualify.

• Your own data: surveys, experiments, measurements, sales or usage figures you are able to share.
• First-hand testing: products used, services tried, places visited, with evidence.
• New analysis: conclusions drawn from public data that nobody has drawn before.
• Primary material: interviews, documents and photographs you obtained yourself.
• Case studies from your own work, with real outcomes.

The test is simple: what would be missing from the web if this page did not exist? If the honest answer is nothing, the page is not original, however well it is written.

What does not count

• Rewriting other articles in different words.
• Aggregating the top results for a search into one longer page.
• Summarising coverage of someone else’s study without adding anything to it.
• Adding a paragraph of opinion to a copied or syndicated piece.
• AI-generated overviews of what already ranks.

These pages can still be useful to readers, and some rank. But they compete on convenience, not originality, and they are exactly what large-scale publishing produces. Where volume turns into a policy problem is covered in scaled content abuse recovery.

A worked example

Imagine two sites writing about remote working habits.

• Site A surveys two thousand people, publishes the results with its method and a downloadable table, and explains what surprised it.
• Site B reads the news coverage of Site A’s survey and publishes a longer, well-formatted article summarising the findings.

Site A is the original, and it is what Google’s systems are designed to favour. Site B may still rank for some searches, especially if it is a larger site, but every figure it quotes depends on Site A, and the method, the data and the context only exist in one place.

Site A can make its position stronger by publishing first on its own domain, making the data easy to cite, and asking the outlets that cover it to link to the original rather than to each other.

When the copy outranks the original

• The copy was crawled first. If a partner or scraper publishes before Google sees your page, the copy can look like the source.
• The copy sits on a much stronger site. Authority can outweigh originality, particularly for syndicated articles.
• Syndicated copies are indexable. Google’s current guidance suggests partners keep their copies out of the index with noindex if you want your version shown.
• Your page is thinner than the copy. If the republisher added context, images or updates, its version may genuinely be more useful.

How Google chooses between near-identical versions is covered in Google’s deduplication systems, and a real example of one site winning over another in the duplicate content case study.

If you syndicate on purpose

Syndication can be a good deal: a larger outlet brings readers you would never reach alone. The trade-off is that its copy may outrank yours.

• Decide what you want from it. If reach matters more than your own rankings for that piece, accept the trade. If not, set terms.
• Ask for noindex on the partner copy, or at minimum a prominent link back to your original.
• Delay the partner version by a day or two, so Google crawls and indexes yours first.
• Keep your strongest pieces exclusive, and syndicate the ones where reach is worth more than ranking.

Agree all of this before anything is published. Changing terms after the copy is live is much harder.

Scrapers and copied content

Scraped content is listed in Google’s spam policies, alongside other ways of producing pages without adding value. Most scrapers never outrank the original, because they add nothing and usually sit on weak sites.

• If a scraper outranks you, check first that your own page is indexed and crawled before theirs.
• For copyright infringement, a valid removal request is the formal route. Google’s systems also take account of sites that receive large numbers of such requests.
• For spam, Google accepts spam reports, though it uses them mainly to improve its systems rather than to act on individual pages.

The full list of policies is covered in every Google spam policy explained.

Checking whether your content is being copied

• Search for a distinctive sentence from your page in quotation marks and see which sites appear.
• Run a reverse image search on your original photographs and charts.
• Look in Search Console for searches where your page lost ground to an unfamiliar site carrying similar text.
• Check new linking sites in the Links report, since some scrapers keep your internal links intact.

Do this for your most valuable pages, not every page. A quarterly check on the pieces that earn the most traffic catches most problems that matter.

How to make originality obvious

• Publish on your own site first, and make sure it is crawled before copies appear.
• Show who did the work and when, with a named author and a clear date.
• Explain your method for any data or testing, so others can see it is yours.
• Use your own images, which are harder to copy convincingly than text.
• Set canonicals on your own duplicates, and agree indexing terms with any syndication partner.
• Make the original easy to cite, with a clear title, a summary of findings and a stable address.

Originality and AI answers

AI answers in search are built by combining sources. A page that repeats what ten other pages say gives them nothing they could not find elsewhere.

A page that is the only place a figure, finding or first-hand detail exists is different. It cannot be replaced by a summary of other pages, which makes original material one of the more durable advantages a site can build. Review content follows the same logic, as covered in Google’s reviews system.

Where original content systems sit among Google’s other systems is covered in Google’s ranking systems explained.