Google's information gain score rewards one thing: data the searcher has not already encountered. Better prose, more thorough coverage, and a fresher publish date do not move it.
That makes it an inventory problem.
The standard content refresh workflow starts by reading what ranks, extracting what those posts cover, and rewriting the same points with better prose and a current date. Every team running that workflow arrives at the same output.
Google holds a patent on scoring the delta between what a searcher has already seen and what your page adds. When every page adds the same information, the delta is zero. The teams doing the most refreshes are often the ones this system was designed to deprioritize. The gap is structural.
Google Scores Every Page on How Much New Information It Adds
The scoring mechanism is specific. It comes from a patent for "Contextual estimation of link information gain", granted in 2022. The patent language is direct:
"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."
Translation: Google tracks what a searcher has already seen and scores each subsequent result on how much it adds. Restate what the previous three results covered, score low. Contain data the searcher has not encountered, score high.
The patent also specifies the consequence: "one or more documents may be excluded (or significantly demoted) from the search results based on the new information gain scores." Pages that add nothing new can be removed from results entirely.
Google's Helpful Content documentation reinforces this with a question every content team should answer honestly: "Does the content provide original information, reporting, research, or analysis?" And a second question that cuts deeper: "If the content draws on other sources, does it avoid simply copying or rewriting those sources, and instead provide substantial additional value and originality?"
Google's Quality Rater Guidelines instruct raters to mark regurgitated content as lowest quality. Patent, rater instructions, public self-assessment: the signal is documented from three directions.
Comprehensiveness Is the Baseline, Not the Differentiator
For years, the winning strategy was coverage. Write longer. Include more subtopics. Build the most thorough result. That era rewarded the content refresh workflow because the goal was completion.
Coverage is now table stakes. When 10 pages cover the same 15 subtopics with the same borrowed statistics, thoroughness distinguishes none of them. What separates results is whether a page holds something the others do not.
Standard Content Refresh Adds Zero Information Gain
You open a post that peaked eight months ago. Traffic is down 30%. The playbook says refresh. You pull up the top five results, note what they cover, identify gaps, and start rewriting. Two time zones away, a competitor's content lead is doing the same thing with the same five results.
That is the content freshness lie at its most mechanical. The workflow begins by reading what ranks. Everything downstream inherits the same inputs. The output cannot diverge because the starting point is designed to converge.
This is freshness theater: the publish date changes, the substance does not.
Each content refresh adds to the content debt instead of reducing it. The team pours editorial hours into reproducing what already exists. Rankings do not move, because the page says nothing new. The effort is real. The output is indistinguishable.
Zombie Statistics and the Information Gain Floor
Some statistics never die and never update. A conversion rate benchmark from 2019. A market size projection revised three times but still circulating at its original figure. A user behavior stat borrowed from a report that borrowed it from another report.
Call them zombie statistics: they propagate without provenance long after the reality underneath them has changed.
When five domains cite the same Gartner number in posts published within the same quarter, Google's system sees five pages contributing identical information. The stat entered the ecosystem once. Every subsequent citation is a copy.
The pages carrying those citations score zero on the delta that matters.
That floor drops further every time a content refresh pulls the same stat from the same shared source. The refresh makes the page look current. The data is inherited.
We Scored 961 SaaS Blog Posts on Originality
Theory says the refresh workflow produces convergence. I wanted to know by how much, so we measured it.
We extracted 5,034 claims from 961 SaaS blog posts across 46 domains and classified each by origin: Original (backed by the publisher's own data), Sourced (attributed to a named external source), or Unattributed (stated without any citation). 31% were Unattributed, and 65.5% were borrowed from third-party research. The Sourced claims drew from a narrow pool of shared references: Gartner, McKinsey, HubSpot, and a handful of industry reports that appeared across multiple domains.
A separate source verification study traced 3,299 third-party claims. Of those, 70% have no external link in the claim's own paragraph that a reader can follow to the data. Of the 30% that do carry a verified link, about 20% are dead, gated, or broken. The named source does not link to the data. The URL is missing. The original report is paywalled or gone.
A link is not proof of a source. It is proof that someone once typed a URL.
That chart is the delta visualized. When the same unverifiable statistics appear across dozens of posts, the novelty each one adds drops toward nothing. The citations are orphaned data: cut off from their source, impossible to verify, shared too widely to differentiate anything.
Charts are claims. Statistics are claims. Borrowed from shared sources without verification, they carry zero information gain no matter how good the prose wrapped around them.
The ratio of Original claims to total claims produces an Originality Score. The domain-level scores make the gap concrete.
GitHub's Originality Score: 91%. Twilio: 100%. Both teams write about their own products, their own engineering decisions, their own usage data.
At the other end of the range, domains scoring near 12% to 21% lean on Unattributed claims borrowed from the broader industry. I have looked at every domain in that study, and the split is rarely about writing talent. It tracks whether the publishing system generates data or borrows it, the exact thing Google patented a way to score.
Infrastructure Is the Scalable Source of Information Gain
Three categories of publishing infrastructure generate information gain as a byproduct. Polls collect zero-party data from readers who interact with the page. Living content blocks synthesize that data into prose that Google indexes on its next crawl. Charts connect to live sources and update without a re-export or re-upload. The loop is mechanical: a reader votes, a paragraph rewrites, a crawler finds different content than last week, and the page contains data that did not exist on the previous crawl.
Infrastructure that generates proprietary data through reader interaction is the only scalable answer to a system that penalizes shared inputs. LiquiChart is one implementation. It changes the category of content a team produces.
Google's own Helpful Content documentation asks: "Does the content provide original information, reporting, research, or analysis?" A living poll that generates its own data answers that question with every response it collects. The data is original because it was collected from readers who chose to contribute. The analysis is original because it reflects responses no other page received.
That is the mechanism.
JSON-LD Dataset schema on the embed gives Google a machine-readable signal that the page holds structured original data. The AI Insights tab turns the response distribution into analysis no competing page can run, because no competing page holds the responses.
None of this content existed before readers arrived. All of it is indexable.
The Question That Exposes the Gap
What percentage of the data claims on your site come from original research? I have never seen a team track that number. That gap is the information gain problem made personal. A poll that generates a dataset where none existed before is information gain in its purest form: the aggregate will say something no competing page in this SERP can, because no competing page asked.
What the Responses Show
What content teams report about their own original research ratios puts a number on the gap.
As readers weigh in above, the distribution will sharpen. The measurement gap is already clear: most publishing teams have never quantified how much of their published data originated from their own systems versus borrowed sources. That ratio determines whether a page generates information gain or reproduces it. Infrastructure that collects proprietary data through reader interaction makes the ratio measurable for the first time. Until teams know the number, they cannot change it.
How Bad Is It for Your Content?
The poll measures the industry. Your content is specific. Different question.
Knowing the industry runs low on original research does not tell you which of your own posts carry the highest convergence risk, or which claims you share with the competitors ranking beside you. Scan your published content for that. The Content Health Scanner extracts claims from any URL, classifies each as Original, Sourced, or Unattributed, and returns your Originality Score, the same metric behind every domain in the study above. The diagnosis takes under a minute.
The scanner measures the structural problem. Knowing the gap is the prerequisite for closing it.
Information Gain Is the New Unit of Content Differentiation
The teams investing in content infrastructure that generates original data will pull ahead of the teams still running the refresh playbook. The gap is scored. Every crawl widens it.
A page that collects 500 reader responses contains data no competitor can replicate by reading it. A paragraph that rewrites itself based on live input says something different every week. A chart connected to a live source reflects the current quarter, not the one from which it was exported.
Content built from shared inputs scores zero on the metric Google built to separate results.
Information gain is an infrastructure problem. The teams that build publishing systems capable of generating proprietary data will own the delta. Everyone else will share it.
The delta does not shrink.