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

# Attribute Influence Sources (Beta)

> Identify the websites and platforms that most influence LLM understanding of your category.

This view enables a lightweight exploration based on a set of generated prompts that cover all key attributes per analysis and common user intentions. Each prompt simulates a real user query — without mentioning any brands — and is sent to LLMs to identify the online sources they draw on when forming answers.

By aggregating these sources, the page reveals the **most influential websites and platforms** contributing to how LLMs understand your category landscape.

## How prompts are generated

To ensure diversity and realistic context, prompts are generated across three user-intention types:

| Intent type                  | What it captures                                                              |
| ---------------------------- | ----------------------------------------------------------------------------- |
| **Information request**      | Educational or factual prompts emphasising expert, authoritative sources.     |
| **Media & social retrieval** | Prompts based on news, opinion and social media — capturing public discourse. |
| **Product purchase**         | Queries reflecting buying intent, surfacing retail and commercial sources.    |

This range covers the full spectrum of user behaviour, from informational and social to transactional.

## How the Influence Score works

Each source receives an **Influence Score** based on how frequently it appears and how prominently it is positioned across AI-generated responses. The formula combines two dimensions:

* **Frequency (75%)** — how often a domain appears in responses.
* **Position (25%)** — how high that domain ranks in the list.

The weighting ensures repetition — the most consistent signal of visibility — matters most, while still rewarding sources that appear early in the response.

### Why position matters more at the top

People look more at the top of any list, so the formula uses a curve that gives extra credit to websites in the first few positions — similar to how Google search treats the top of a SERP.

### Why this works well

* **More frequent** = more visible and likely more trusted.
* **Higher position** = more attention from users.
* **Combined**, both produce a strong picture of which websites users are most likely to see and trust in AI results.

## Score interpretation

| Score range  | Reading                                                        |
| ------------ | -------------------------------------------------------------- |
| **50–100**   | Domain is frequently cited and appears in prominent positions. |
| **20–50**    | Moderate citation frequency and/or positioning.                |
| **Below 20** | Limited presence or appears in less visible positions.         |

The scoring system reflects real user behaviour — sources mentioned first carry significantly more weight than those listed later.

## Interface components

<CardGroup cols={3}>
  <Card title="Top Influential Sources" icon="trophy">
    Bar chart of the three most influential domains and their scores — immediate insight into market leaders.
  </Card>

  <Card title="Source Types" icon="chart-pie">
    Pie-chart breakdown of source types (Specialist Blogs, News Media, Retail/Commerce, etc.) — understand the media landscape for your category.
  </Card>

  <Card title="Source Changes" icon="arrow-trend-up">
    Trend visualisation showing how source influence evolves — track shifting AI preferences and emerging influential domains.
  </Card>
</CardGroup>

## Detailed source table

| Column                 | Description                                                                                   |
| ---------------------- | --------------------------------------------------------------------------------------------- |
| **Rank**               | Position in the influence hierarchy.                                                          |
| **Domain**             | Source website.                                                                               |
| **Type**               | Category classification (News Media, Specialist Blog, Branded site, Retail/E-commerce, etc.). |
| **Top Cited URLs**     | Most frequently referenced pages.                                                             |
| **Influence Score**    | Average visibility metric.                                                                    |
| **SearchGPT & Gemini** | Platform-specific performance indicators.                                                     |

<Frame caption="Detailed source table">
  <img src="https://mintcdn.com/jellyfish/MmUm2_YE5Jn6SqTt/images/platform/models/Capture-d_e_cran-2026-05-04-a_-15_27_08-6c2e91a5.png?fit=max&auto=format&n=MmUm2_YE5Jn6SqTt&q=85&s=c6b3d48f416706d54af788b8261f287b" alt="Source table" width="1683" height="972" data-path="images/platform/models/Capture-d_e_cran-2026-05-04-a_-15_27_08-6c2e91a5.png" />
</Frame>

<Note>
  In the **Sources** module, the analysis is **not centered on brand mentions**. The brand and competitor analysis parameters are not used in this section. Data collection prompts rely only on:

  * the analysis category,
  * the analysis attributes,
  * the countries.

  Sources highlights influential domains and URLs based purely on categories and attributes, without referencing any specific brand names.
</Note>

## What's next

<CardGroup cols={2}>
  <Card title="Reddit Feed" icon="reddit" href="/platform/models/reddit-feed-overview">
    Inspect Reddit conversations behind the scores.
  </Card>

  <Card title="Sources & Links (Search)" icon="link" href="/platform/search/sources-links-what-how-purpose">
    See sources and links inside the Search module.
  </Card>
</CardGroup>