Information gain is the extra, genuinely new information a page contributes beyond what a reader has already seen in other results for the same question. In SEO, the phrase usually points to a Google patent that describes scoring documents by how much novel information they offer.
This page explains where the term comes from, what the patent does and does not prove, how it lines up with Google's public guidance on helpful content, and how to build pages that add something new rather than rewording the top ten.
What is Information Gain?
The term originally comes from machine learning and statistics, where it measures how much a piece of data reduces uncertainty. Marketers adopted it after Google filed a patent titled "Contextual estimation of link information gain." The application was filed in October 2018 by Google engineers Victor Carbune and Pedro Gonnet Anders and granted as US Patent 11,354,342 in June 2022. A related patent with the same title, US 12,013,887, was granted in June 2024.
In plain terms, the patent describes a system that looks at documents a user has already seen, then uses a machine learning model to estimate how much new information other documents would add. Documents can then be ranked or presented based on that score. Much of the patent is framed around automated assistants, where the goal is to avoid repeating what the user was already told, but it also discusses reordering search results after someone views a page.
In everyday SEO use, information gain has become shorthand for a simple question: what does your page offer that the pages already ranking do not?
Why Information Gain matters
Here it helps to separate what is confirmed from what is industry interpretation.
Confirmed: Google holds patents describing information gain scoring. Separately, Google's own guidance on creating helpful content asks creators whether their content provides original information, reporting, research, or analysis, whether it goes beyond the obvious, and whether it offers substantial value compared with other pages in search results. It also asks whether content that draws on other sources adds real value instead of just rewriting them. Google's ranking systems guide notes that the helpful content system became part of its core ranking systems in March 2024.
Interpretation: Google has not said that it calculates an information gain score for organic rankings or for AI Overviews. Patents often describe ideas that never ship. Search Engine Journal's analysis of the 2024 patent pointed out that it is framed mainly around assistants and follow-up content, and that nothing in it proves pages with more information outrank competitors.
So why care? Because the practical advice holds up either way. Helpful content guidance rewards originality, and AI answer engines that pull from several sources have little reason to cite a page that repeats what five others already said. Adding something new is a sound bet whether or not a literal score exists.
How Information Gain works
The patent's model
The patent describes three basic steps: identify documents the user has already consumed, estimate how much new information each candidate document would add relative to that set, and use those estimates to decide what to show next. The score can be computed from a document's content or from a mathematical representation of it, such as an embedding.
How SEOs apply the idea
Most people use information gain as an editorial test rather than a technical metric. If a reader has already opened the top three results, would your page teach them anything they have not seen? Things that tend to pass that test include original data, firsthand testing, expert commentary, real examples, and angles the current results skip.
The AI search connection
AI systems that break a question into sub-queries, a process called query fan-out, gather passages from many pages and combine them. A page that contributes a unique fact, number, or perspective has a clearer reason to be included than one that duplicates the consensus. This is a reasonable inference about how retrieval works, not a documented Google ranking rule.
Information Gain in practice: an example
Imagine a hypothetical company that sells standing desks and wants to rank for "how tall should a standing desk be." The top results all offer the same height chart and the same ergonomic tips.
A low information gain page would rewrite that chart in new words. A higher information gain page might add measurements the company took from 200 customer setups, a short video showing common mistakes, notes from a physical therapist the company works with, and a calculator for people who use a monitor arm. Each of those adds something a reader cannot get from the existing results.
No one can promise the second page will rank higher. It does give searchers, Google, and AI tools a clear reason to prefer it.
How to add Information Gain to your content
- Read what already ranks. Open the top results and any AI answers for your target query. List the points every page covers so you know the baseline.
- Find what is missing. Run a content gap analysis, check forums and support tickets, and note questions the current results leave open.
- Bring first-party evidence. Use your own data, customer surveys, test results, case studies, or photos. Original evidence is the hardest thing for competitors to copy.
- Add real expertise. Include the judgment of people who have done the work, which also supports E-E-A-T signals.
- Cover the baseline briefly. Readers still need the basics, but keep them tight so your new material stands out.
- Make the new parts easy to find. Use clear headings, labeled data, and specific claims so both readers and AI systems can pick out what is unique.
- Revisit over time. What was novel last year may be standard now. Update pages when the field catches up.
Common mistakes and misconceptions
"Google confirmed information gain is a ranking factor." It has not. Google holds patents on the concept and its public guidance values originality, but there is no official statement that an information gain score drives organic rankings.
"Longer content has more information gain." Length is not novelty. A 5,000-word page that repeats the consensus adds less than a 900-word page with one original dataset.
"Rewording competitors creates something new." Paraphrasing produces different sentences, not different information. Google's guidance specifically asks whether content does more than copy or rewrite other sources.
"Information gain means skipping the basics." Readers who land on your page first still need context. The goal is baseline plus something extra, not only the extra.
Information Gain vs. content gap analysis
Content gap analysis is a research method: you compare your site with competitors to find topics, keywords, or AI prompts they cover and you do not. Information gain is a quality standard for the page you then write. A gap analysis can tell you to create a page on a topic. Information gain is what makes that page worth more than the ones already out there.
How Butter approaches Information Gain
We treat information gain as an editorial discipline, not a trick. Before recommending new pages, we look at what already ranks and what AI tools currently say, then focus on the evidence, examples, and expertise a client can uniquely provide.
Through our SEO services and our GEO plans, which start at $399/month and include monthly AI prompt testing and on-page recommendations, we help brands identify where their own knowledge can fill gaps in both search results and AI answers. Fundamentals first: original, useful content tends to outlast any single algorithm theory.
Frequently asked questions
Is information gain a Google ranking factor?
Google has not confirmed it. The company holds patents describing information gain scores, and its helpful content guidance rewards original value, but there is no public evidence the patented method is used in organic ranking.
What is the Google information gain patent?
It is a patent titled "Contextual estimation of link information gain," filed in 2018 and granted as US 11,354,342 in 2022, with a related patent granted in 2024. It describes using machine learning to estimate how much new information a document adds beyond what a user has already seen.
How do I measure information gain in my content?
There is no public tool that reproduces Google's score. A practical approach is to list what top results already cover, then count the facts, data points, examples, and perspectives your page adds that none of them include.
Does information gain help with AI search visibility?
It plausibly helps, since AI answers combine information from several sources and unique details give a page a reason to be included. This is a reasonable inference, not a documented rule, so track results with regular prompt testing rather than assuming.
Sources and further reading
- US Patent 11,354,342: Contextual estimation of link information gain, Justia Patents
- US12013887B2: Contextual estimation of link information gain, Google Patents
- Creating helpful, reliable, people-first content, Google Search Central
- A guide to Google Search ranking systems, Google Search Central
- Google's Information Gain Patent for Ranking Web Pages, Search Engine Journal
- What Is Information Gain in SEO?, Ahrefs