Google may assess not only what makes an article unique, but also how much relevant new information it adds beyond what a user has already read. Media and industry authors’ interpretations of the patent suggest one possible approach: compare documents, estimate information gain, and then adjust subsequent search results. But a patent is not the same as a deployed ranking mechanism. These interpretations alone do not establish that Google has adopted any particular formula.
Uniqueness and information gain: what is being compared?
In a 2025 article, Semrush explains information gain in SEO as how much unique information an article provides compared with similar content a user has already read. Uniqueness can take the form of original research, expert commentary, or useful original images. Information gain places greater emphasis on the question: “After reading the previous article, what else does this one add?” The same article therefore may not offer the same added value to users with different reading histories.
Information foraging theory uses the concept differently: it focuses on the value of information learned after subtracting the cost of obtaining it. Making an article easier to understand can improve its value to readers, but that improvement cannot be treated directly as a Google ranking score.
A possible process: comparing read and unread documents
In his 2026 interpretation, Search Engine Journal author Harry Clarkson-Bennett explains the process using a document already read, d1, and a document yet to be read, d2:
- Convert the documents into vectors or other semantic representations so their content can be compared.
- Estimate how much relevant new information d2 provides beyond d1.
- Use that added value to decide whether to show d2 in the user’s subsequent searches.
For example, if a user has already read a basic introduction to growing apple trees, an article that adds different cultivation methods may be a better choice to show next than one that repeats the basics. This is an example the author uses to explain the mechanism, not an observed ranking result. Under his interpretation, documents may be reordered, excluded, or demoted. But the focus here is on subsequent results; this approach cannot be treated as a rule for all initial search results.
What do public clues suggest, and what remains an inference?
Semrush noted in 2025 that the relevant patent was granted in 2022, but that it was still unclear at the time whether Google used information gain as described in the patent. The possible scenarios it listed included a user quickly returning to the results page or making a slightly different follow-up search. These illustrate how the mechanism might work; they do not mean that every return to the results page triggers reordering.
The 2026 Search Engine Journal article argues that public and leaked information about Google’s systems contains clues related to originality and content creation effort. These clues, the author suggests, make the existence of some assessment mechanism more likely. However, the author also acknowledges that it is unclear whether Google uses an information gain score, another predictive score, or indirect interaction signals. This speculation cannot substitute for evidence of actual deployment.
What can content creators do instead of chasing an unknown score?
A sounder approach is to first identify search intent and gaps in existing content, then add original research, expert input, or visual explanations that help answer the question. Semrush also cautions that more unique elements are not necessarily better. Even if a travel list includes places others have not covered, it should not sacrifice the quality of its recommendations or cram in too many options. Useful differences matter more than simply being different.
In our content services at Procrustes, we turn the questions clients genuinely care about into articles, with ongoing planning, writing, and updates.
How Google might assess content uniqueness and information gain: perspectives from across the web
Does an information gain score of zero mean an article has no value?
Not necessarily. The Search Engine Journal author uses a hypothetical scenario to explain that if a second article provides no additional information beyond the first article a user has already read, its score could be zero, and that user may be less likely to see it later in their journey. This is a comparison tied to a particular reading history. It does not mean the article has no value to any reader, nor is it a confirmed current ranking rule.
Could original content affect rankings in an initial search?
Some authors suggest this possibility, but the material is insufficient to confirm it. Semrush argues that original elements could improve rankings in subsequent results, or even initial results, only if information gain actually influences Google's algorithm. Search Engine Journal's interpretation of the patent, meanwhile, emphasizes subsequent results. An improvement in initial rankings therefore cannot be treated as a proven effect.
Is an estimate of content creation effort the same as an information gain score?
No. Search Engine Journal describes contentEffort as using a large language model to estimate the effort put into creating an article. The same article interprets an information gain score as a measure of how much information a document adds beyond documents on the same topic that a user has already seen. The two measure different things, and the author cannot determine which mechanism Google actually uses.
References
- How Google May ‘Understand’ Unique Contentwww.searchenginejournal.com · 2026-07-15T13:30:00+00:00 — Discusses: Does an information gain score of zero mean an article has no value?; Could original content affect rankings in an initial search?
- What Is Information Gain in SEO & Does Google Measure It?www.semrush.com · 2025-02-26T16:26:00Z — Discusses: Could original content affect rankings in an initial search?


