Skip to main content

Sources & Links - what, how & purpose

Understand how domains and URLs shape LLM answers.

Apolline Vanneste avatar
Written by Apolline Vanneste
Updated over a month ago

🎯 Why this view exists?

In generative search, the final answer is only part of the story.

LLMs rely on two distinct layers:

  • Domains and URLs used as Sources to build their answer (influence)

  • Domains and URLs shown as Links in the final output (visibility)

These two layers often differ.

The Sources & Links view helps you see and understand this difference, so you can better interpret how answers are constructed across engines.

To go further:


📊 What this view helps you analyse?

The Sources & Links view helps you understand which domains and URLs influence LLM answers, how they are surfaced, and how this varies across queries and engines.

#1 Main table: Comparing Domains and URLs at a glance

ezgif-1c26fb86b10696e9.gif

This view lets you compare domains and URLs across three key dimensions:

1. Role in the answer

The Type column shows whether a domain is:

  • Source → used to build the answer

  • Link → visible in the final output

  • Source & Link → both influencing and visible

This makes it easier to understand how a domain contributes to generative answers.

2. Consistency across engines and queries

By looking at:

  • The Engines column

  • The Results column

  • The Presence rate

You can identify whether a domain appears occasionally or is consistently used across multiple engines and prompts, helping distinguish one-off appearances from recurring influence.

3. Brand mentions

The My Brand Mentions column shows whether your brand is mentioned within the source content itself.

This helps answer questions like:

  • Is my brand discussed even when it’s not visible in the answer?

  • On which domains does my brand appear most often?

#2 Domain details: Understanding content patterns

ezgif-1de34c6f9d334da0.gif

When you click on a domain, you access the list of URLs associated with that source. This view helps you understand what kind of content the LLM relies on from a given domain.

You can observe:

  • Which URLs are used as Sources, Links, or both

  • How often each URL appears

  • Whether brand mentions are present at URL level

  • Their content and structure by accessing those URLs directly from the interface

#3 Results details: linking domains, URLs and queries

Screenshot 2025-12-19 at 15.09.30.png

When clicking on the ‘Results’ number a domain or URL appears on, you access the details of queries by engine on which the item appears as a Source and/or as a Link.

The Results view shows:

  • The search queries (prompts)

  • The role played by the domain or URL for each query

  • Differences by engine

You may notice that:

  • The same query behaves differently across engines

  • A domain is a Source on one engine and a Link on another

  • Some queries rely heavily on Sources without showing Links

This highlights how LLMs interpret intent and assemble answers differently, even for the same topic.


💡 Reading cues: what common patterns can tell you

This section highlights common patterns observed in Sources & Links and how they are often interpreted.

These are signals to read, not rules.

A competitor appears frequently as Source & Link

This pattern is usually observed when content is:

  • trusted by the model to build its response

  • structured in a way that is easy to surface in the final output

It often appears on queries where:

  • the answer requires synthesis

  • the model needs both explanation and examples

This suggests the domain fits well with how LLMs combine reasoning and visibility.

A domain appears mainly as a Source

This typically reflects content that:

  • explains concepts

  • provides background or context

  • supports the model’s reasoning process

This behavior is often seen on queries where:

  • understanding the topic matters more than pointing to a single reference

These domains influence how the answer is constructed, even when they are not visible to users.

A domain appears mainly as a Link

This pattern is usually observed when content is:

  • easy to reference

  • clearly structured (lists, steps, summaries)

It often appears on queries where:

  • the model benefits from pointing users to concrete resources

Here, visibility is more closely tied to format and accessibility than to depth of reasoning.

Differences across engines

When patterns vary by engine, this usually reflects:

  • different source selection logic

  • different preferences for structure, depth or citation

This reinforces the idea that LLM visibility should be interpreted engine by engine, rather than as a single aggregated signal.

The same domain behaves differently depending on the query

You may notice that:

  • a domain is used as a source on some queries

  • and appears as a link on others

This reflects the fact that LLMs adapt how they use content based on the nature of the question, rather than applying a single rule.


How to use this view effectively

Start by identifying which domains consistently act as Sources, even when they are not visible as Links. These domains often shape how answers are constructed and can reveal where influence happens before visibility.

Did this answer your question?