GEO Content Structure and Formatting Test

GEO content structure test board comparing paragraph, table, definition list, and source block formats with scope, caveat, evidence, and action scores — A controlled formatting run can reveal where a page loses scope, caveat, evidence, or action when the same recommendation is summarized in a GPT-style answer.

Search Console can show that a page appeared for a query like “geo content structure and formatting.” It cannot tell you whether the page format helped an answer system keep the important details. For that, I use a small content-formatting test: the same technical recommendation is written in four formats, then reviewed for claim, scope, caveat, evidence, and action retention.

This is a GEO experiment, not a ranking study. It does not prove that a table will earn a citation or that a source block will improve traffic. It answers a narrower editorial question: which format makes this recommendation hardest to distort?

The recommendation used in the test

A good test starts with one claim. If the source text tries to test five recommendations at once, the result becomes noisy. I used this claim because it is common in GEO work and easy to overstate:

Recommendation:
Use a source block when a GEO recommendation depends on platform behavior,
crawler access, structured data output, screenshots, or a reproducible test.

Scope:
Technical articles, documentation pages, and implementation notes.

Caveat:
A source block improves verifiability. It does not guarantee ranking,
indexing, inclusion in an AI answer, or citation.

Evidence:
The rendered page should keep the claim, source note, date, limitation,
and verification step in the same visible section.

Four formatting variants

The variants below use the same information. Only the structure changes. That is important because the test is about formatting, not about whether one version secretly contains more useful facts.

Variant	Format	Why test it
A	Paragraph	Fast to write and common in articles, but easy for summaries to flatten.
B	Table	Good for comparison, but can feel dense and may split the recommendation into cells.
C	Definition list	Keeps claim, scope, caveat, and evidence close without a heavy layout.
D	Source block	Best when the claim depends on proof, a screenshot, a command, or platform behavior.

The screenshot-style test board

The image for this article is not decoration. It is a compact visual record of the test design: four format variants on the left, extraction scores on the right, and the editorial decision below the score table. The same information is repeated in text so the article still works without images.

Variant A: paragraph

The paragraph version is readable, but it creates the most extraction risk because the claim and caveat sit inside a continuous sentence. A short answer may keep the recommendation and drop the limit.

<p>
  Use a source block when a GEO recommendation depends on platform behavior,
  crawler access, structured data output, screenshots, or a reproducible test.
  This is useful for technical articles and documentation pages, but it does
  not guarantee ranking, indexing, inclusion in an AI answer, or citation.
</p>

Variant B: table

The table version is stronger for scanning. It makes the fields visible. The weakness is that the reader may need to connect the cells mentally, and on narrow screens the table can become harder to read.

<table>
  <tr><th>Claim</th><td>Use source blocks for GEO claims that depend on evidence.</td></tr>
  <tr><th>Applies to</th><td>Technical articles, docs, and implementation notes.</td></tr>
  <tr><th>Evidence</th><td>Keep source note, date, limitation, and verification step together.</td></tr>
  <tr><th>Limit</th><td>No ranking, indexing, AI answer, or citation guarantee.</td></tr>
</table>

Variant C: definition list

The definition list was the best compact format in this run. It kept each fact close to its label without making the section feel like a large comparison table.

<dl class="geo-extraction-facts">
  <dt>Recommendation</dt>
  <dd>Use source blocks for GEO claims that depend on platform behavior,
  crawler access, structured data output, screenshots, or reproducible tests.</dd>

  <dt>Applies to</dt>
  <dd>Technical articles, documentation pages, and implementation notes.</dd>

  <dt>Evidence</dt>
  <dd>Keep the source note, date, limitation, and verification step in the same section.</dd>

  <dt>Limit</dt>
  <dd>A source block improves verifiability; it does not guarantee visibility or citation.</dd>
</dl>

Variant D: source block

The source block version scored highest when the recommendation needed proof. It is not the right shape for every paragraph. It works when a reader should see the claim and the evidence before acting.

<aside class="source-block" aria-label="Evidence for this recommendation">
  <strong>Recommendation</strong>
  <p>Use source blocks for GEO claims that depend on platform behavior,
  crawler access, structured data output, screenshots, or reproducible tests.</p>

  <strong>Checked evidence</strong>
  <p>Rendered HTML keeps the source note, date, limitation, and verification
  step in the same visible section.</p>

  <strong>Limit</strong>
  <p>This improves verifiability. It does not guarantee ranking, indexing,
  AI answer inclusion, or citation.</p>
</aside>

Prompt used for the review

I used one fixed prompt for all four variants. Changing the prompt between runs would make the formatting comparison useless.

Summarize the provided page section in 4 bullets.
Preserve:
- the recommendation
- where it applies
- the evidence or verification step
- any limitation or caveat
- the next action a technical editor should take

Do not add claims that are not present in the section.
Do not turn a limitation into a guarantee.

Scoring notes from the run

This is the kind of scoring note I would keep in an editorial log. The point is not to publish a scientific benchmark. The point is to decide which format should ship on the page.

Format	Scope	Caveat	Evidence	Action	Editorial decision
Paragraph	1/2	0/2	0/2	1/2	Do not use alone for claims with limits.
Table	2/2	2/2	1/2	1/2	Useful for comparison, but add a short text summary.
Definition list	2/2	2/2	1/2	2/2	Use for compact implementation facts.
Source block	2/2	2/2	2/2	2/2	Use when the recommendation needs proof.

What I changed after the test

The experiment changed my default formatting rule. I no longer put caveated GEO recommendations inside ordinary paragraphs when the reader needs to act on them. I use a paragraph for context, then a definition list or source block for the actual recommendation.

Working rule

If the advice can be misunderstood without scope or evidence, do not bury it in prose. Label the claim, applies-to boundary, evidence, and limit.

Screenshot protocol

For a real article, I would keep one screenshot of the test board and one screenshot of the rendered page section. Do not publish private account data or prompt history that includes client information.

Screenshot file: geo-formatting-test-run-2026-07-03.png
Shows: four formatting variants and extraction scores
Redact: account names, private prompts, unrelated tabs, internal URLs
Keep in text: test prompt, scoring table, editorial decision
Do not claim: model-wide performance, stable citation behavior, or ranking impact

How to use this in a GEO content brief

Add a formatting decision to the content brief before drafting. This prevents every article from falling back to the same paragraph-plus-table pattern.

GEO format decision
Main recommendation: source blocks for verifiable claims
Risk if summarized badly: caveat may be dropped
Chosen format: source block
Fallback text: one-sentence summary before the block
Screenshot needed: yes, if the evidence comes from rendered HTML or a tool check
Internal links:
- /articles/geo-source-blocks-verifiable-claims/
- /articles/geo-answer-extraction-testing/
- /articles/geo-content-structure-examples/