What Is Information Gain and Why Does Google Care About It Now?

Most of the content I see losing rankings in 2026 is not spam. It is not thin. It is technically well-structured, keyword-optimised and covers the topic in the right word count range. The problem is that it says exactly what ten other pages already say, in roughly the same order, with roughly the same examples. Google has a name for this now: zero information gain. And it is one of the primary signals the March 2026 core update appears to have tightened.

What is Information Gain?

Information Gain is a concept Google patented and has discussed in research papers going back several years. The core idea is straightforward: how much genuinely new information does a page add compared to what already exists in the top results for that query?

It is not about length. A 3,000-word article that rephrases ten other articles has zero information gain. A 900-word article with one genuinely novel angle, one original data point, or one practical example that is not already in the top ten can have high information gain. Google is trying to evaluate the delta, what does this page contribute that the searcher cannot already get from what is ranking?

The March 2026 core update appears to have increased the weighting on this signal materially. Sites with strong Information Gain are gaining. Sites with low Information Gain, even where the content is technically clean, are dropping.

Why this is harder than it sounds

The trap most content hits is writing to the topic rather than writing from experience of the topic.

If I search for the ten top-ranking pages on technical SEO audits and then write an article that synthesises what they say, I produce something that is accurate, comprehensive and entirely replaceable. Google’s systems are increasingly good at identifying this pattern. The content reads fluently, passes surface quality checks and covers all the right headings. But it adds nothing.

What Google is trying to reward is content that could only have come from someone who has actually done the thing. When I write about crawl errors, I reference specific patterns I see repeatedly in real audits, combinations of redirect chains, blocked CSS files and incorrect canonical tags that tend to appear together on particular CMS platforms. That specificity is not something you can derive from reading other articles. It comes from doing the work across hundreds of sites over 25 years.

That is what Information Gain looks like in practice. Not comprehensiveness. Originality.

How Google detects low Information Gain

Google has not published a detailed breakdown of how Information Gain is calculated algorithmically, but the patent literature and the March 2026 update data together point to a few mechanisms.

The first is semantic similarity at scale. Google’s systems can compare the meaning of your content against the existing corpus of pages ranking for a query. If the conceptual coverage is near-identical, same points, same structure, same examples, that registers as low information gain regardless of whether you used different words.

The second is the Gemini 4.0 Semantic Filter, which the SEO community widely believes was deployed in the March 2026 core update. This does not penalise AI-assisted content outright. What it targets is content produced at scale without meaningful editorial input, pages that read fluently but contribute nothing original. The distinction Google draws is between AI used as a tool by someone with genuine expertise versus AI used as a replacement for genuine expertise. The first produces content with information gain. The second typically does not.

The third is engagement signals over time. Content that fully satisfies a query, that gives the searcher something they could not get elsewhere, tends to produce better engagement patterns. Lower pogo-sticking, longer dwell time, fewer follow-up searches for the same query. These signals reinforce the Information Gain assessment.

What high Information Gain content actually looks like

It tends to share certain characteristics regardless of topic or format.

Specificity that only comes from direct involvement. Real numbers from real work, not “studies show conversion rates improve” but “in twelve ecommerce audits I ran in the past year, the most common cause of Core Web Vitals failures was third-party scripts loading in the critical path, specifically chat widgets and analytics tags.” The detail has to be the kind that you can only have if you have been there.

Proprietary data or original research. If you have run a survey, analysed your own dataset, or tracked something systematically that others have not, that is pure Information Gain. Even a small dataset from your own clients or projects, documented honestly, is more valuable to Google than a synthesis of published research.

A genuinely different angle on a well-covered topic. Not a different title, a different argument. Not “here is what E-E-A-T means” but “here is why anonymous content is the single biggest E-E-A-T gap I see on otherwise well-built sites, and here is what fixing it actually involves.” The angle has to be substantive, not cosmetic.

First-hand case examples with real outcomes. A before-and-after from an actual client project, with specific numbers and the actual changes made, demonstrates information gain in a way that a theoretical walkthrough of the same process does not.

How to audit your own content for Information Gain

The practical test is uncomfortable but useful. Take your ten most important content pages and read them as if you found them in search for the first time. Ask: is there anything on this page that you could not have found by reading the other top results for this query?

If the answer is no for most pages, you have an Information Gain problem. The fix is not rewriting the page in different words. It is adding genuine original content, something from your direct experience, your client data, your own analysis, that is not already in the top results.

A useful secondary test: can you find five places on each page where you reference something specific enough that it could only come from direct involvement? A specific number from a real project. A specific tool behaviour you have observed. A specific pattern in data you have collected. If you cannot find five, the page is probably too generic to score well on Information Gain.

The structural changes that help, better headings, cleaner internal links, faster load times, will not move you if the underlying content offers nothing that the existing results do not. Information Gain is a content quality problem, not a technical one.

What this means for content strategy in 2026

The direction Google is moving is clear and has been consistent across the past three core updates. Volume strategies, publishing high numbers of pages targeting individual keywords, only work if each page contributes something. The March 2026 update has made the consequences of the opposite approach steeper.

What works now is depth over breadth, specificity over comprehensiveness, and genuine expertise over synthesised coverage. That is not a new principle. It is a principle Google now has better tools to enforce.

For anyone building content from a position of actual expertise, with real client work, real data and real experience to draw on, this is a competitive environment that keeps improving. The sites that built rankings on content volume without content depth are losing ground to the sites that built on the back of genuine knowledge. That shift creates space for specialists who have been doing the work honestly.

If you want an assessment of whether your current content is contributing genuine Information Gain or repeating what is already out there, that is something I look at as part of a technical SEO audit.

Frequently asked questions