Duplicate content is substantially the same content available at more than one URL, on your own site or across sites. There is no duplicate content penalty. What happens instead is that Google picks one version to index and folds the others into it, and if it picks the wrong one, or splits links between them, you lose some of what you have earned.
How it happens without anyone meaning it to
A page reachable with and without www, over http and https, with and without a trailing slash, with tracking parameters attached, as a print version, in a paginated series, under two categories. A product in three colours with three URLs and one description. A blog post syndicated to a partner. None of this is malicious. All of it gives search engines several copies to choose between.
What it costs
Google's guidance describes the problems plainly: it wants to consolidate signals such as links onto one preferred URL, split versions make your own metrics harder to read, and crawling duplicates wastes effort that could go on new pages. If some sites link to one version and some to another, the credit is divided. If Google chooses the version you did not want, that is the one that appears.
What it does not do is punish the site. Google's staff have said for years that there is no penalty for ordinary duplication. Copying other people's content wholesale is a different matter and falls under the spam policies for scraping.
How to deal with it
Declare a canonical URL on every page pointing at its preferred form. Redirect the www and non-www, http and https variants to one. Keep tracking parameters out of internal links. Where two of your own pages genuinely say the same thing, merge them and redirect the loser. When content is syndicated, ask the partner to canonicalise to you.
Where this bites hardest
Online shops, where filters and sort orders generate thousands of near-identical URLs, and multi-location businesses that copy one service page per town with the name swapped. The first needs canonicals and parameter handling. The second needs pages that are actually different, or fewer of them.
Sources
Where this shows up in the tracker
Every term in this glossary is something the product actually measures or acts on. Plans start at £29 a month, with rank tracking, AI visibility, the daily AI readiness scan, security and domain health, Search Console, backlinks and audits on every one of them.
Related terms
Canonical
A canonical tag is a line in a page telling search engines which address is the official version, for when the same content is reachable at several different URLs.
Thin content
Thin content is a page that offers a visitor little or nothing they could not get elsewhere: empty category pages, copied product descriptions, auto-generated text, a paragraph wrapped around an advert. Google has no minimum word count and says so. A short page that answers a question is not thin. A long page that says nothing is.
301 redirect
A 301 redirect is a permanent instruction that a page has moved: anyone requesting the old address, including search engines, is sent to the new one. Google treats it as a signal that the new address should be the canonical version and shows the new address in results. Use it whenever a URL changes for good, and keep it in place.
Cannibalisation
Cannibalisation is when two or more of your own pages compete for the same search, so Google keeps swapping which one it shows. Both pages are weaker than one good page would be.