Module 04 · Lesson 3 of 4
Information Gain, Evidence and Original Data
Information gain is the useful difference between your page and what the competing sources already say. If a reader who has seen the top three results learns nothing new from yours, your page has zero gain, and there is no reason for any system or person to prefer it. Rewording competitors is not gain. Length is not gain. Gain is new.
What produces gain
| Source of gain | What it looks like | Honesty requirement |
|---|---|---|
| Original data | Your counts, surveys, benchmarks, price tables you gathered | Say how and when you collected it |
| First-hand testing | "We ran this and here is what happened" | Only where testing genuinely happened |
| Practitioner expertise | Decisions, warnings, edge cases from real work | Attributable to a real, named person |
| Unique aggregation | Combining scattered sources into one structured view | Cite every source aggregated |
| Genuine local or niche specificity | The version of the answer for your exact audience | Actually specific, not a keyword swapped in |
Notice what is absent from the table: adding words, adding sections nobody asked for, "10x length" thinking. Padding dilutes. A page with one real dataset and 800 words beats a 3,000-word retelling of what five sites already say, every time, with both audiences.
Evidence discipline
Gain without evidence is assertion. Claims get sourced to their primary origin where one exists; your own data gets a method note (when, how many, how collected); uncertainty gets stated plainly rather than papered over. This discipline costs little and buys the credibility that citation selection responds to. And the integrity rule from the methodology page applies to your content too: unknown stays unknown. If you do not know a number, say so or leave it out; a fabricated case study is a defect even when it makes the page feel richer.
Planning for gain
The planning pass is mechanical and takes twenty minutes. Search the query. Read the top three results properly. List what they cover, then look for the gaps: what questions do buyers ask that none of them answer, what data do they assert without showing, what situation do they ignore? Five candidate gains will usually fall out. Pick the one you can deliver honestly and build the page around it, with passages and naming from the last two lessons.
- Query: "how much does seo cost for a small business". Top three: two agencies with vague ranges, one publisher survey.
- Covered: typical monthly ranges, what affects price, agency vs freelancer.
- Gaps: none show real quotes side by side; none break cost down by business type; none cover one-off project pricing vs retainers; none explain how to sanity-check a quote.
- Candidate gains ranked: your own anonymised quote collection (strong, original), a cost-by-business-type table (strong, aggregable), how to sanity-check a quote (good, expertise-based).
- Build the page around the first two; the third becomes a section, not the headline.
- Take the starred query from Workbench 3.4, your highest-cost absence.
- Read the top three results fully. Write their coverage list.
- Write five candidate gains that none of them deliver.
- Star the gains you can honestly produce, and pick one as the page's centrepiece.
- Draft the page outline: answer passage, gain section, evidence, FAQ. Module 4.4 turns it into architecture.
Is a longer version of what competitors say information gain?
What makes original data honest?
Where does expert commentary count as gain?
Sources used in this lesson
Google Search Central: helpful, reliable, people-first content