What it means
If ten pages already say the same thing, an eleventh page saying it again adds nothing. A page with a new fact, test, number or view adds information. That's information gain.
Google has a patent describing an "information gain score" for this idea. Whether and how it's used in ranking isn't public, but the idea fits how both search and AI work.
Why it matters for AI
An AI writing an answer reads several pages. If they all say the same thing, it only needs one of them. The page with something extra, like a price nobody else lists, a test result or a clear comparison, has a reason to be quoted.
How to add it
- Your own data: numbers from your product, a survey, a test.
- First-hand experience: what happened when you actually used it.
- Specifics: prices, limits, dates, names.
- A clear opinion, with reasons.
Example
Fifty articles explain what a CRM is. One adds a table of what 12 CRMs actually cost for a five-person team, checked this month.
That table is new information. It's the page AI quotes when someone asks about CRM pricing.
Frequently asked questions
Does rewriting other articles in my own words add information gain?
No. It's the same information. This is what commodity content means.
Is original research expensive?
It doesn't have to be. Data you already have, like usage trends or support questions, can be enough.
Does information gain help with ChatGPT too?
The idea does. Any AI combining several sources gets more from the one that says something new.