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

# Themes & Narratives

> Explore the recurring stories AI platforms tell about your brand, grouped by topic.

A **narrative** is a single recurring story AI tells about your brand, pulled from real responses: "praised for ease of use", "seen as expensive for small teams". A **theme** groups related narratives under a shared topic, such as Pricing or Customer Support. Themes give you the headlines; narratives give you the detail.

## The Themes Tab

Each theme is shown as a card with:

| Element             | Shows                                                          |
| ------------------- | -------------------------------------------------------------- |
| Sentiment Score     | The theme's 0-100 score, with its label                        |
| Sentiment breakdown | The split of negative, neutral, and positive mentions          |
| Narratives          | How many narratives sit under the theme                        |
| % of responses      | The share of AI responses where the theme appeared, with trend |
| Surfaced By         | Which AI platforms raised the theme                            |

Search to find a specific theme, or sort by **% of responses**, **Narrative count**, or **Most negative** to surface problems first.

## Theme Detail

Click a theme to open it. The header charts daily mentions by sentiment and summarises the key numbers for the period. Below it, three tabs:

* **Overview** - the narratives within the theme and how they moved, the sentiment split per AI model (useful when one platform frames a topic differently to the rest), the most cited sources, and when the theme was first and last seen
* **Responses** - the AI responses behind the theme, with the relevant excerpts
* **Sources** - the web pages cited in those responses

## The Narratives Tab

The narrative explorer lists every narrative with filters down the side:

* **Search** - find narratives by wording
* **Sentiment** - positive or negative
* **Momentum** - filter by how narratives are moving (see below)
* **Themes** - limit to one or more themes

Sort by **% of responses**, **Largest movement**, **Latest**, or **Most negative**, and switch between card and compact views.

## Momentum

Every narrative carries a momentum badge comparing the selected period to the previous one:

| Badge    | Meaning                                                        |
| -------- | -------------------------------------------------------------- |
| Emerging | Brand-new this period, with no presence in the previous window |
| Surging  | Share of responses grew sharply versus the previous period     |
| Steady   | Roughly flat versus the previous period                        |
| Cooling  | Share of responses fell sharply but the story is still present |
| Dormant  | Present last period, gone this one                             |

<Note>
  Momentum measures movement, not sentiment. A surging narrative can be praise or a growing problem, so check its sentiment too.
</Note>

## Narrative Detail

Click a narrative to see the full picture:

* **Model Agreement** - how often each AI model stated this narrative. Broad agreement means the story is embedded across platforms, not a one-off
* **Related narratives** - other stories from the same theme
* **Responses** - every response linked to the narrative, with the extracted excerpt
* **Sources** - the pages cited in the responses where the narrative appeared. They aren't tied to the narrative itself, but they show what AI was drawing on in the same answers
* **Lifecycle** - when the narrative was first and last seen

## Export

Click **Export** on the Themes or Narratives tab for an Excel download matching your current filters. Theme and narrative detail pages also export their responses and sources.

## What to Look For

* **Negative narratives with high % of responses** - these shape buyer perception the most, address them first
* **Broad model agreement on inaccuracies** - if every platform repeats the same wrong claim, correct it at the source
* **Sources around negative narratives** - the pages cited alongside them often point to where PR or content effort will pay off