Information Gain

Where does the term come from?

From a granted Google patent, which is unusually solid ground for an SEO concept. Contextual estimation of link information gain was filed by Google on 18 October 2018 and granted on 7 June 2022. Its abstract defines the score directly: “an information gain score for a given document is indicative of additional information that is included in the given document beyond information contained in other documents that were already presented to the user”.

The scenario the patent describes is the familiar one: a user reads several documents about the same problem and each subsequent one repeats the last. The patent’s answer is to score documents by what they add rather than by how relevant they are.

Does Google actually use it?

Unknown, and anybody telling you otherwise is guessing. A granted patent proves an idea was described and claimed, not that it shipped, and Google has neither confirmed nor denied using it in ranking. Treating a patent as a ranking factor is one of the more common errors in this field.

What makes the concept worth keeping anyway is that it does not need Google to be true. In a world where an answer engine assembles a reply from several sources, a page that repeats the consensus contributes nothing the other sources did not already supply, and there is no mechanism by which it gets selected over them.

That is a stronger argument than the patent, and it applies to every engine rather than to one. It also produces a usable test that needs no tooling: read the top three results before writing, and answer in one sentence what your page will contain that all three lack. An answer that takes effort to produce is the page worth writing, and no answer at all is a page worth skipping.

How do you actually add information gain?

By having something the other results do not, and the honest test is to check rather than assume. Read the top three results for the question first, then answer in one sentence what your page will contain that all three lack. If there is no answer, the page is not ready to be written.

• **A verified specific where everyone else is vague.** The exact user agent string, the exact wording of the vendor’s statement, the date it was read.
• **A correction of a widespread error.** The most linkable kind, because it saves the next writer an argument.
• **A distinction nobody draws.** Mention against citation, training crawler against search crawler.
• **First-hand measurement**, where it genuinely exists, and never where it does not.
• **A number with a date and a source**, where the alternative was an adjective.

Related terms

• Answer BlockAn answer block is a short passage near the top of a page that answers the page’s question completely and on its own.
• BLUF (Bottom Line Up Front)BLUF, or bottom line up front, is the convention of stating the conclusion in the opening sentence and the supporting reasoning afterwards.
• Atomic ContentAtomic content is the practice of writing in self-contained units, each covering one idea completely enough to be understood alone.
• Topical AuthorityTopical authority is the idea that covering a subject deeply and consistently makes a site more likely to rank across it.