Editorial strategy · SEO · AI search

Information Gain & The Death of Commodity Content

By LinkinGrow · November 6, 2026 · AI-assisted, source-checked editorial analysis

Research desk contrasting a stack of blank sheets with annotated evidence being examined by hand — AI-generated editorial illustration; the documents and chart are not research evidence.

When everyone can produce a competent explanation, competence alone stops being a distinctive reason to read—or reference—your page.

That is the argument behind the “death of commodity content.” It is not a claim that basic explainers no longer rank, that Google has announced a new information-gain score, or that every article needs a laboratory experiment. It is an editorial and commercial argument: a page that adds nothing beyond existing answers has a weak reason to exist.

For publishers, marketers, and online businesses, the useful question is not “How do we make this article sound different?” It is “What can a reader understand, verify, or do after reading our page that they could not do as easily before?” This guide separates documented search guidance from our analysis and turns that question into a practical publishing discipline.

The short answer

Information gain, in this editorial context, is the useful additional understanding a page contributes relative to the information already available to its intended reader. It can come from original observations, a transparent calculation, a better synthesis of primary sources, or a clearer explanation of an overlooked constraint.

Commodity content is interchangeable: remove the logo and almost any competitor could publish it unchanged. AI makes this material easier to produce and summarize. Our view is that publishers should invest less in interchangeable coverage and more in evidence, interpretation, and decision support. That is a strategy, not a guaranteed ranking mechanism.

What information gain means—and what it does not

The term has several meanings that should not be collapsed into one. In technical information theory, information concerns uncertainty and probability. Shannon’s foundational paper explicitly separates the engineering problem from semantic meaning. Its mathematics does not provide a formula for the quality of a blog post.

In SEO discussion, “information gain” often refers to whether a document adds something beyond other documents a searcher might encounter. In editorial practice, it is a useful lens for testing a page’s contribution. These meanings are related by analogy, not by a published conversion from word count or novelty into search ranking points.

Additional information is also relative. A plain definition may be highly useful to a beginner and redundant to a specialist. An advanced benchmark can be original but useless to a buyer if it ignores price, setup effort, or the conditions under which the test applies. Originality must meet audience need.

Novelty is not the same as truth. An unsupported statistic is new information only in the most literal sense; it is not useful evidence. Nor is a contrarian opinion automatically a contribution. Distinctiveness earns its place when readers can inspect the reasoning behind it.

What makes content a commodity?

A commodity page usually repeats the same definitions, headings, advice, and examples as existing coverage without improving the answer. “Know your audience,” “create quality content,” and “measure your results” may be correct, but they leave the reader to solve the actual problem.

The issue is not whether AI helped write it. A human can produce an interchangeable article; AI can help analyze a genuinely useful dataset. The distinction is the contribution, not the typing method. Warning signs include:

Headings copied from the top-ranking pages, with no missing question identified.
Advice that could apply to any industry, audience, or product.
Statistics repeated without dates, original sources, or definitions.
A “2026” title attached to unchanged material.
Claims of experience unsupported by observations, methods, or real authorship.
Multiple pages targeting small wording variations of the same need.

Basic reference material still has a place. A concise, accurate answer with exceptional usability can be more valuable than a sprawling original essay. Commodity content is not a synonym for short content, introductory content, or content on a popular topic.

What Google actually documents

Originality and added value appear in its guidance

Google’s people-first content guidance asks whether a page provides original information, reporting, research, or analysis; goes beyond the obvious; avoids merely copying or rewriting sources; and provides substantial value compared with other search results. These are useful editorial questions. They are not a checklist that guarantees a higher position.

Scaled low-value publishing can violate spam policy

Scaled content abuse concerns generating many pages primarily to manipulate rankings rather than help users. Google’s examples include content with little or no original value. The policy applies regardless of the production method. Publishing at scale is not inherently abusive; scale, purpose, and value need to be considered together.

Google’s generative AI guidance similarly recognizes research and structuring uses while warning about generating many pages without added value. “AI content is banned” and “AI content is automatically safe” are both misleading shortcuts.

A patent is not proof of a live ranking factor

The Google-assigned patent “Contextual estimation of link information gain” describes estimating additional information a document may provide relative to documents a user previously viewed. It is relevant to the concept, but a patent is not evidence that the technique is deployed in Google Search today.

It does not justify claims that every page has a publicly established information-gain score, that Google measures novelty with a known formula, or that a particular writing tactic will improve that score. The practical case for useful originality stands on published quality guidance and reader value; it does not require a speculative algorithm story.

Why AI search changes the economics of generic answers

An answer engine can assemble a serviceable summary of widely repeated facts without giving any one publisher a distinctive role. Our editorial inference is that being another source of interchangeable definitions offers less differentiation than being the source of a useful method, observation, or explanation that the summary cannot replace as easily.

That does not mean answer engines always cite the most original source. Their full selection processes are not public; retrieval, availability, query interpretation, freshness, and other factors can affect what appears. Original work may be overlooked, summarized imperfectly, or cited without generating visits.

Google says ordinary SEO practices remain relevant to AI Overviews and AI Mode, with no special additional requirements. Useful originality is not a substitute for an indexed page that is eligible to show a snippet, accessible text, or a site that readers can navigate.

The business implication is to create material worth visiting after an answer: the complete methodology, an interactive calculation, a maintained reference table, the underlying evidence, or a nuanced decision guide. The summary can introduce the topic; the source should justify the next step.

Six forms of useful originality

1. Original observations

Document what you actually observed: a repeated AI-search question set, a migration diagnostic, or a product implementation issue. Record dates, conditions, sample selection, exclusions, and limitations. A small transparent observation is more defensible than a large unsupported claim. Do not turn a handful of sessions into a claim about all users or all engines.

2. Reproducible comparisons

Compare alternatives against the same criteria, using dated primary specifications and a stated method. Distinguish documented features from tested behavior. If you did not test a product, describe the work as a specification comparison—not a hands-on review or benchmark.

3. Primary-source synthesis

Originality does not require collecting new data. A careful synthesis can reconcile conflicting definitions, connect policies to practical choices, or explain where sources disagree. Preserve attribution and distinguish the source’s claim from your interpretation. Rewriting six articles into a seventh is not the same thing.

4. Decision tools and worked calculations

A transparent cost model, implementation checklist, or decision tree can help readers act. Show assumptions, units, and failure conditions. A calculator with hidden assumptions offers precision without accountability; a simple worked calculation can be more useful.

5. Constraints, exceptions, and negative findings

Explain when a familiar recommendation does not apply. A guide that says when not to use structured data, when a migration should wait, or why a test was inconclusive helps prevent mistakes. Negative findings are valuable only when they are real and the method is disclosed.

6. Maintained reference assets

A dated crawler-policy matrix or licensing reference can remain useful because someone verifies it and records changes. Updating the year in a title is not maintenance. Keep a change log, revisit primary sources, and explain whether a changed fact affects the reader’s decision.

A worked editorial example: from advice to a decision

This is a hypothetical publishing exercise, not a customer case study or measured result. Consider a company writing “How to choose payroll software for a multi-state business.” Its draft lists price, features, integrations, and support. All are sensible; none resolves the hard choice.

An improved brief asks narrower questions: Does the stated price include each state registration service? Which responsibilities remain with the employer? Are support availability and tax-filing coverage documented? What should a buyer verify before signing? The team can build a dated comparison from official terms and mark unknowns rather than fill them with guesses.

For a hypothetical cost illustration, assume a $50 monthly base fee and $6 per employee for 25 employees: $50 + (25 × $6) = $200 per month, or $2,400 per year, before any additional fees. Those numbers are invented assumptions, not vendor prices. The useful addition is the explicit calculation and the reminder to check excluded costs.

The contribution is not that the article became longer. It now gives a buyer a verifiable comparison method, identifies responsibility gaps, and separates known facts from unanswered questions. No traffic lift or AI citation outcome is claimed.

An information-gain audit for existing pages

The following is our editorial framework, not a Google scoring system. For each important page, write down the intended reader, the decision they need to make, what existing sources already answer, and the specific contribution your page makes. If the contribution cannot be stated in one sentence, the brief probably needs work.

Need: Is the question real and specific? Use search queries, support questions, and interviews where available, not invented demand.
Contribution: Which observation, explanation, tool, or synthesis adds value beyond available answers?
Evidence: Can a reader inspect the sources, method, dates, and assumptions?
Utility: Does the addition change understanding or help a decision, rather than merely decorate the page?
Limits: Does the page explain what the evidence cannot establish?
Maintenance: Who will verify information that can become outdated?

Choose an action rather than a vanity score. Keep pages that answer a distinct need well. Improve important pages with identifiable evidence gaps. Consolidate overlapping pages when one stronger resource would serve readers better. Retire obsolete material when there is no useful replacement; do not redirect unrelated topics simply to preserve traffic.

An evidence-first production workflow

Start with a contribution brief, not a target word count. Define the question, intended audience, proposed addition, available evidence, and the boundary between fact and interpretation. A useful stop condition is: if the promised contribution depends on data you do not have, gather the data or change the brief.

Build a source ledger before drafting. For each important claim, record its primary source, the relevant passage, the date checked, and any qualifications. Obtain consent before using private customer material; anonymization alone does not create permission. Do not manufacture quotations, testimonials, tests, or company experience to make an article appear authoritative.

Use AI where it improves the work: organizing notes, comparing terminology, suggesting missing questions, drafting from verified material, and checking internal consistency. It should not invent the evidence. Keep a clear distinction between a proposed experiment, a completed observation, and a hypothetical example.

Put the answer first, then show the reasoning. Keep limitations close to the claims they qualify. Explain unusual terms, use meaningful headings, and link sources where readers need them—not only in a bibliography. More citations do not compensate for irrelevant citations.

A human editor should check factual support, originality, conflicts of interest, disclosure, and usefulness before publication. The quality gate is not “Does it read smoothly?” It is “Are we adding knowledge to the internet, or merely creating another page?”

How to measure the result without inventing a score

There is no universally established public information-gain score for websites. Evaluate the editorial output and its distribution separately. First, ask whether readers can use it: do support teams reference it, do relevant readers return, do independent sources link to the method, and are corrections handled transparently?

Then examine search and business outcomes: query-level impressions and clicks, qualified visits, meaningful actions, and conversions where attribution is available. A successful reference page may have modest traffic but help a high-value decision. Page count and total word count are not useful substitutes.

For AI visibility, keep a fixed question set and record engine, date, location or session context, full answer, cited URL, and whether the citation supports the actual claim. Repeat samples because answers vary. A mention, a citation, an accurate description, and a recommendation are different observations.

Before-and-after changes cannot establish causation on their own. Search updates, seasonality, competing pages, distribution, and technical changes may explain the result. When possible, compare similar unchanged pages and keep a change log. Report uncertainty rather than crediting every improvement to “information gain.”

A 90-day improvement plan

Days 1–15: identify redundancy and real needs

Audit a manageable set of important pages. Review existing search queries and audience questions. Identify overlap, unsupported claims, stale information, and the decisions each page should help. Establish a measurement baseline before changes.

Days 16–40: build one useful contribution

Choose a page where better evidence is feasible. Produce a dated comparison, a transparent worked calculation, or a source-led explanation of a difficult issue. Publish the method and limitations alongside the result. Verify that the page remains accessible and indexable.

Days 41–65: improve the surrounding library

Link supporting definitions and relevant guides to the new resource. Consolidate genuinely overlapping pages where appropriate. Remove unverified statistics and misleading experience claims. Assign owners to references that require updates.

Days 66–90: evaluate and decide

Review reader feedback, independent references, search performance, and repeated AI-answer samples. Decide whether the contribution deserves expansion, maintenance, or a different question. Do not respond to weak results by automatically multiplying pages.

Common questions

Is information gain a confirmed Google ranking factor?

The cited public sources do not establish a deployed ranking factor under that name. Google encourages original, useful content; a relevant patent describes a possible technique. Those are different kinds of evidence.

Is commodity content really dead?

No universal disappearance is established here. The title describes declining differentiation, not a documented elimination from search. Accurate introductory material can still be useful. The problem is redundancy without a reader benefit.

Does every article need original research?

No. A clear synthesis, maintained reference, or practical decision tool can add value. If you claim original research, disclose what you actually did and avoid presenting a small convenience sample as representative.

Can AI-assisted content have information gain?

Yes, when it helps produce a truthful, useful contribution from real evidence. Google’s policies do not categorically prohibit AI content. Human accountability remains important for checking the work and deciding whether it deserves publication.

Will longer articles perform better?

Length is not the contribution. Use enough detail to answer the question and support the claims. Repetition, unrelated sections, and an inflated word count can obscure the useful part.

Should a small business stop publishing basic FAQs?

No. Genuine questions about pricing conditions, availability, compatibility, and responsibilities deserve clear answers. Business-specific accuracy and completeness can be valuable even when the writing is simple. Do not invent questions solely to expand the site.

Sources and disclosure

This AI-assisted, source-checked article distinguishes published guidance from LinkinGrow’s editorial interpretation. The audit and 90-day plan are proposed practices, not validated ranking formulas. The payroll example and its prices are hypothetical. The image is an AI-generated illustration. No original benchmark, customer outcome, or private dataset is claimed.