> ## Documentation Index
> Fetch the complete documentation index at: https://docs.haloxlabs.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Fan-out search patterns

> How to analyze AI search fan-out from root questions to source paths and citation actions in HaloX.

<Frame>
  <img src="https://mintcdn.com/afterworklab/lupD4-jE5SvWCVgz/images/concepts/page-thumbnails/en-guides-fan-out-search-patterns.png?fit=max&auto=format&n=lupD4-jE5SvWCVgz&q=85&s=811de11f290256dbdbedcf9138b8a334" alt="HaloX Guides page thumbnail summarizing Fan-out search patterns in five cards" width="1600" height="900" data-path="images/concepts/page-thumbnails/en-guides-fan-out-search-patterns.png" />
</Frame>

AI search engines do not answer a user question in one straight line. ChatGPT, Gemini, Perplexity and Claude often expand a question into sub-questions, comparison criteria and source checks. HaloX treats this as a **fan-out search pattern**.

For example, “Which GEO tool should our industry use?” can expand into:

* What GEO means
* Which use cases exist in this industry
* How GEO differs from SEO tools
* What to check before adoption
* Which sources and guides are trustworthy

GEO operations therefore cannot stop at the ranking of one representative keyword. You need to know **how a question fans out and whether your brand and content appear along each path**.

## What should you prepare before starting?

* Make sure your HaloX workspace and target site are ready.
* Keep GSC/GA4 access and any existing SEO or meeting notes nearby.
* Decide where the output will live: internal note, client report, or weekly operating document.

## How do you verify the result?

| Check              | Healthy state                                               | Next action                                             |
| ------------------ | ----------------------------------------------------------- | ------------------------------------------------------- |
| Workflow completed | The requested audit or reporting task is finished           | Reflect the result in Strategy Map and the right report |
| Priority selected  | The next keyword, page, or content opportunity is clear     | Assign an owner and due date                            |
| Evidence recorded  | The metrics and reasoning can be reused in the next meeting | Track the same signals next week                        |

## What commonly blocks teams?

| Problem                                       | Cause                                                        | Fix                                                      |
| --------------------------------------------- | ------------------------------------------------------------ | -------------------------------------------------------- |
| Scores are visible but next steps are unclear | Metrics were not translated into execution units             | Start from Strategy Map GAP types and priority clusters. |
| It is hard to explain to customers            | Technical signals were not converted into business language  | Add cause, impact, and next action to every report item. |
| Results do not change immediately             | AI answer surfaces need time to recrawl and reassess sources | Track the same prompt set and metrics weekly.            |

## One-line definition

> Fan-out is the path from `root question → sub-questions → source candidates → validation loop → final answer/citation`.

| Stage             | What AI systems do                                                                                                            | What to do in HaloX                                                               |
| ----------------- | ----------------------------------------------------------------------------------------------------------------------------- | --------------------------------------------------------------------------------- |
| Root question     | Interpret the original user question.                                                                                         | Add representative questions to a strategic prompt set.                           |
| Sub-questions     | Split into definition<br />comparison<br />buying criteria<br />location<br />risk<br />and review questions.                 | Use Strategy Map to inspect clusters and gap types.                               |
| Source candidates | Look for evidence across official sites<br />newsrooms<br />blogs<br />reviews<br />video<br />communities<br />and articles. | Use Citation Tracking to separate Source from Citation.                           |
| Validation loop   | Compare consistency<br />authority<br />freshness<br />and structure across sources.                                          | Use Site Audit and Content Factory trust reports.                                 |
| Answer/citation   | Mention a brand or cite a link in the final answer.                                                                           | Track citation rate, question share, and repeated competitor exposure in reports. |

## Why this came up in meetings

Customer and partner meetings kept repeating the same point: a single brand page is not enough. You need to follow the sub-questions and source paths that AI systems use.

| Meeting requirement                                                                                            | Fan-out interpretation                                                                                                              | Connected HaloX features                       |
| -------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------- |
| Financial/platform teams want non-brand conversion questions, brand search and non-brand conversion questions. | “Before signing up,” “comparison,” “how to,” and “best” questions fan out from the root intent.                                     | Prompt Analysis, Strategy Map, Content Factory |
| Enterprise communications teams want leadership<br />issue<br />newsroom<br />and owned-media monitoring.      | Brand-definition questions fan out into leadership<br />issues<br />owned media<br />and external articles.                         | Citation Tracking, Site Audit, Weekly Reports  |
| PR/brand teams ask about Naver, Google, and AI-channel differences.                                            | The same question can lead to Naver blogs<br />newsrooms<br />YouTube<br />articles<br />LinkedIn<br />and other sources.           | Citation Tracking, Agency GEO operations       |
| B2B SaaS teams ask about GEO after SEO drops.                                                                  | Recover the core search topic first<br />then expand into problem<br />comparison<br />and adoption questions.                      | Site Audit, Strategy Map, Content Factory      |
| Local businesses ask about local GEO.                                                                          | Questions split into “location + recommendation,” service comparison<br />before-visit questions<br />and English/global questions. | Local GEO guide, Prompt Analysis               |

## How to analyze fan-out in HaloX

<Steps>
  <Step title="1. Split one representative question into question groups">
    Separate brand questions from non-brand, comparison, buying and location questions. Explain prompts as “core questions AI receives.”
  </Step>

  <Step title="2. Inspect sub-question clusters in Strategy Map">
    Fan-out questions usually spread across several keyword clusters. Review search demand, AIO/GEO/SEO state, gap type and urgency together.
  </Step>

  <Step title="3. Check source paths with Citation Tracking">
    Separate whether your brand is only mentioned, used as a source candidate, or explicitly cited as a link. Repeated competitor citations are high-priority content gaps.
  </Step>

  <Step title="4. Verify whether AI systems can read the content">
    Even good content may fail when robots rules, CDN behavior, schema gaps or JavaScript dependence block discovery.
  </Step>

  <Step title="5. Turn each question into answer assets in Content Factory">
    Build definitions, comparison tables, FAQ pages and evidence blocks with internal links.
  </Step>
</Steps>

## Example fan-out prompt sets

| Industry / objective | Root question                        | Fan-out sub-questions                                                                                       | Needed content                                                                |
| -------------------- | ------------------------------------ | ----------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------- |
| Financial platform   | “Best crypto exchange”               | How to trade<br />fee comparison<br />security<br />before-signup checks<br />app usability                 | Comparison page<br />beginner guide<br />FAQ<br />security/policy page        |
| B2B SaaS             | “Best CRM marketing automation tool” | Definition<br />adoption criteria<br />alternatives<br />case studies<br />pricing/operations               | Problem guide<br />comparison content<br />checklist<br />case page           |
| PR/brand             | “What kind of company is this?”      | Leadership<br />issues<br />business structure<br />official newsroom<br />external articles                | Fact sheet<br />newsroom hub<br />FAQ<br />issue explainer                    |
| Local business       | “Best clinic in Gangnam”             | Location<br />service criteria<br />before-visit questions<br />reviews/trust<br />foreign-language support | Branch landing page<br />local FAQ<br />service comparison<br />English guide |
| Agency proposal      | “How do you diagnose GEO?”           | Diagnosis items<br />score interpretation<br />execution priority<br />report outputs                       | One-page diagnosis, operating loop, report sample                             |

## Why citation reason matters

A brand appearing in an AI answer is not enough. The role inside the answer matters.

| State               | Plain-language explanation                                              | Next action                                                                  |
| ------------------- | ----------------------------------------------------------------------- | ---------------------------------------------------------------------------- |
| Mention             | “The name appears, but we do not know whether it was used as evidence.” | Check brand accuracy and competitor co-mentions.                             |
| Source              | “The page became an evidence candidate.”                                | Strengthen structure<br />references<br />freshness<br />and internal links. |
| Citation            | “The page appeared as a cited link/source.”                             | Track whether it repeats across the same question group.                     |
| Competitor citation | “A competitor was selected as evidence.”                                | Compare their page format with your content gap.                             |
| No reliable source  | “AI relies on generic summaries instead of official evidence.”          | Add FAQ<br />fact sheets<br />entity hubs<br />schema<br />and bot access.   |

## Convert questions into content

| Content element            | Why it matters                                    | Example                                                                                         |
| -------------------------- | ------------------------------------------------- | ----------------------------------------------------------------------------------------------- |
| First-paragraph definition | Gives AI a clear answer sentence.                 | “GEO is an operating method for managing brand mentions, sources, and citations in AI answers.” |
| Question-led H2/H3         | Matches the sub-query shape.                      | “Why is my AI citation rate low?”                                                               |
| Comparison table           | Supports recommendation and comparison questions. | Criteria table across options.                                                                  |
| FAQ                        | Captures real customer objections.                | “Should we do GEO before SEO is fixed?”                                                         |
| Evidence/source links      | Increases citation readiness.                     | Official docs<br />newsroom<br />data<br />policy<br />examples.                                |
| Internal links             | Connects the fan-out path.                        | Definition → comparison → playbook → report.                                                    |

## Talk track for customers

> “AI does not answer a question as a single keyword lookup. It expands the question into related sub-questions and source checks. HaloX therefore monitors prompt sets, source paths, citations and content actions as one operating loop.”

## Related docs

* [Customer-meeting GEO scenarios](/en/guides/customer-meeting-scenarios)
* [Prompt Analysis](/en/features/prompt-analysis)
* [Strategy Map](/en/features/strategy-map)
* [Citation Tracking](/en/features/citation-tracking)
* [Content Factory](/en/features/content-factory)
* [PR and brand AI citations](/en/guides/pr-brand-ai-citations)
* [B2B SaaS GEO operations](/en/guides/b2b-saas-geo)