Information Gain
TL;DR: What is Information Gain?
Information gain is a measure of how much new information a piece of content adds beyond what is already available, a concept from a Google patent describing scores for documents based on their novel contribution. Content that only repeats consensus scores low; content adding something scores high.
Information Gain explained
The concept entered SEO vocabulary through a Google patent describing "information gain scores": ranking documents partly by how much new information they would give a user beyond documents already seen. Whether and how Google deploys it exactly as patented is unknowable from outside, but the principle it names has become one of the most useful lenses in content strategy, and it visibly rhymes with how AI systems select sources.
The lens is simple to apply and uncomfortable in what it reveals. Take any planned piece of content and ask: what does this add that the existing top answers do not already say? Original data, first-hand experience, a genuinely different framework, documented outcomes, a position the consensus lacks. If the honest answer is nothing, the piece is a rehash, and rehashes now compete against both an infinite supply of AI-generated sameness and retrieval systems with no reason to select a redundant source.
For AI visibility specifically, novel information is citable in a way consensus repetition is not. An answer engine grounding a claim needs a source that actually originates or substantiates something; the fifteenth restatement of common knowledge gives it nothing to credit. Information gain is, in that sense, the content-side answer to the citation question: be the source something traces back to.
In practice
Information gain is the test behind every content decision I push clients toward: what do you know that the existing answers do not? First-party data, client patterns, documented results, an honest contrarian position. It is also why this glossary has an In Practice section on every page; the definitions alone would be commodity, and commodity is precisely what retrieval systems have least reason to cite. If a planned piece has no gain, my advice is consolidate or kill it before it dilutes the pages that do.
Common misconception
People often equate more comprehensive with more valuable. Actually, comprehensiveness that restates known information adds no gain; a shorter piece contributing one original finding is more citable than a long one contributing none.
Want the bigger picture? Start with the Founder’s Guide to SEO.