Information gain

Information gain, in SEO, is the amount of new information a page adds beyond what a reader has already seen in other results on the same topic. This SEO sense is tied to a Google patent on ranking documents for search and AI assistants, and differs from the machine learning measure of the same name used in decision trees.

Why information gain matters

AI answers grounded in search results draw on several sources at once, so a page that repeats the others gives the engine nothing new to use. Google says so in its guide to generative AI features: "Our AI systems take a look at a variety of sources, so it can be helpful to have a unique viewpoint that stands out." It contrasts commodity content, such as "7 Tips for First-Time Homebuyers", with content that offers "unique expert or experienced takes that go beyond common knowledge." Expertise and experience are both part of Google's E-E-A-T framework.

Data points the same way, with caveats. In SE Ranking's study of 216,524 pages, pages rich in statistics (19 or more data points) averaged 5.4 ChatGPT citations, against 2.8 for pages with few, and pages with expert quotes averaged 4.1 against 2.4 without. SE Ranking rates these effects as "relatively minor compared to others."

What the Google patent describes

Google's patent "Contextual estimation of link information gain" (US11354342B2), filed in October 2018 and granted in June 2022, defines an information gain score as "indicative of additional information that is included in the document beyond information contained in documents that were previously viewed by the user." A machine learning model computes it, and documents can be re-ranked as the user reads, with those that add little "excluded (or significantly demoted)" from the results. Its first claim describes an automated assistant answering natural language requests, a close match for AI assistants; other claims cover a search results page.

The word "contextual" matters: the score depends on what this user has already seen, so the same page can score high for one reader and low for another. And a patent shows what Google has explored, not what it runs in Search.

Information gain in SEO vs machine learning

The two meanings share a name and little else. If a source talks about entropy or decision trees, it uses the machine learning sense.

	Information gain in SEO	Information gain in machine learning
What it measures	New information a document adds for a reader	Drop in uncertainty (entropy) when data is split on a feature
Where it is used	Ranking documents, in Google's patent	Decision trees and feature selection
What scores high	A page that says what others don't	A feature that separates classes well

How to add information gain to a page

Take the prompt "How long does a heat pump last?" (illustrative). If every ranking page gives the same lifespan range, another page with that range adds nothing. A page from an installer that reports what failed first in the units it services adds something no other page has.

List what the top results and AI answers already say, then write what only you can add: tests, your own data, prices, customer cases.
Answer Google's own questions: does the content "provide original information, reporting, research, or analysis"?
Find the gaps that cost you: Hawkscan's action plan flags pages that cite your competitors but not you, your citation gap. Compare them with your page and note what they say that you don't.