Information Gain Audit

The Information Gain Audit helps you discover where your website is missing out beyond consensus. You get the main subjects your website is missing, named and ordered, across the whole domain, up to 300 URLs by sitemap. Larger domains on request. Text generation is your task or can be commissioned separately.

The audit stays away from the result page. Tools that watch the result page describe what competitors already published, which is how everybody ends up writing the same thing once more. The subjects worth adding come from what customers want answered and nobody has answered yet.

More words do not move the score. Figures nobody else has do.

150 pages across 50 search queries in 10 verticals. The median page holding a top-3 position carries four such figures. Study by On-Page.ai, published through Similarweb, June 2026.

The list stops at the main subjects. Everything beneath them, the subtopics, the questions your customers ask there, the order of work and a briefing per topic, is the domain-wide audit.

Delivery, one list of the missing subjects, three to five working days after the written agreement, through your email address.

The price depends on the website and is agreed on request before anything begins: info@johannesfaupel.com. For business customers only.

The whole domain, up to ten hub pages, and what each of those hubs is missing. Where the Information Gain Audit names the topics, the domain-wide audit works through them: which are worth adding, in what order, what adding each one requires, and what it is worth if it stays missing.

What you get

• Up to ten hub pages examined, with the settled part and the missing part separated for each.
• The missing topics across the domain, ranked by what they are worth to your customers.
• An order of work, with the reasoning for the order.
• A written briefing per topic, ready to hand to whoever writes.
• A review of how your hubs connect, including the connections nobody has written.

More than ten hub pages on request. The price depends on the website and is agreed before anything begins.

The domain-wide audit is settled by bank transfer against an invoice. There is no card payment for it, because the scope is agreed first and the invoice follows the agreement.

Take the Information Gain Audit if you want to know whether there is anything there. It answers one question: which topics are missing, and what people would ask about them. Most domains have more than the owner expects.

Take the domain-wide audit if you already know there are topics missing and need an order of work, a budget and briefings your team can act on. It is the version that produces a plan rather than a list.

For both audits, write to info@johannesfaupel.com and name the domain. The website decides the scope, the scope decides the price, and the invoice follows. Payment is by bank transfer.

The only offer you can order immediately is the thematic gap stack at overlookedrevenue.com.

Prices are quoted net, plus VAT where it applies. All offers on this page are for business customers only, subject to change and non-binding. A contract comes into being only with the written agreement on scope and price. Johannes Faupel, Frankfurt am Main, Germany.

Is the Information Gain Audit the same as “information gain” in machine learning?

No. In machine learning, information gain measures how much the entropy of a dataset falls when it is split on one feature, and it is used to choose the splits in a decision tree. The Information Gain Audit measures something else: the difference between what a subject field already treats as settled, including what an AI model has absorbed as settled, and what is genuinely still missing there. The two share a name and an intuition, that gain means the removal of uncertainty, and nothing beyond that.

Why does a search for the term return decision trees and entropy?

Because the machine learning sense is the older one and currently the only one with a public cluster of questions behind it. Every follow-up question that search engines attach to the term today asks about entropy, Gini impurity, C4.5 or the arithmetic of a split. No cluster exists yet for the sense used here. That is a statement about the state of the public record rather than about the method.

Where does the term come from in search?

From Google. Two granted patents carry the title “Contextual estimation of link information gain”: US 11,354,342 B2, granted in June 2022, and US 12,013,887 B2, granted in June 2024, both held by Google LLC. They describe an information gain score for a document, defined as the additional information that document carries beyond what the reader has already seen. That is the machine learning idea carried over to reading, and it is the reason the term belongs in search at all.

What counts as already seen?

This is where the methods part company. The patent takes it to be the documents the reader has already viewed. Most SEO tooling takes it to be the pages currently ranking in the top five. The Information Gain Audit takes it to be the settled understanding of the subject wherever it has been recorded, including what an AI model has absorbed as settled. Three different reference points, and only the third still means anything for a reader who arrives already holding an answer.

What does the audit compare a domain against?

Against the settled understanding of the subject. It stays away from the result page on purpose. A method that takes the top five results as the reference can only ever return what those five already published, which is why domains that follow such advice converge on the same text.