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

# Reading a recommendation

> The brief behind a recommendation: the buyer question, the answers, the shortfall, the proposed work, and the evidence.

Open a card and its **brief** explains, in plain words, why the recommendation exists and what to do about it. Everything on the page traces back to a real answer from your last scan, so you can judge the recommendation on evidence rather than take it on trust.

<Frame>
  <img src="https://mintcdn.com/onsombleai/myRrAp7kiTPZVaHC/images/understanding-results/reading-a-recommendation.png?fit=max&auto=format&n=myRrAp7kiTPZVaHC&q=85&s=6dd7a9c4e9e7515c0dd99801dd78c97a" alt="A recommendation brief showing the buyer question, the AI answer, what is holding you back, the proposed work, and the evidence sources" width="1280" height="850" data-path="images/understanding-results/reading-a-recommendation.png" />
</Frame>

## Where you stand today

At the top, a ladder shows how AI answers currently treat your brand on this question, from **Absent** through **Criticised**, **Mentioned** and **Cited as authority** to **Recommended**. Your position today is marked, and so is the target the recommendation is aiming for.

## The four parts of the brief

<Steps>
  <Step title="Buyer question">
    The question an AI assistant was asked. This is what a potential customer
    might type, not an internal metric.
  </Step>

  <Step title="AI answer">
    What the assistants said about your brand on that question today, summarised
    in plain words.
  </Step>

  <Step title="What's holding you back">
    Why your standing is not higher, and what the answers would need to see to
    move it. When specific factors are missing, they are listed here.
  </Step>

  <Step title="Proposed work and measurable test">
    The action to take, and a short plan the next scans can check: what to
    measure, what to publish, and when to scan again.
  </Step>
</Steps>

## Check the evidence

The brief lists the sources behind your current standing: your own pages that AI answers cite, and the assistant answers themselves. Each source is labelled and quoted, so you can open it and confirm the pattern before you act.

<Steps>
  <Step title="Read the cited sources">
    See which of your pages, and which assistant answers, shaped the standing.
    Open a source to read it in full.
  </Step>

  <Step title="Confirm the pattern is real">
    Check that the gap shows up in more than one answer, and that the claims
    about your brand are accurate.
  </Step>

  <Step title="Verify in Prompt Results">
    Some recommendations from older scans carry less evidence. Use the
    recommendation as a lead and confirm it in [Prompt
    Results](/results/prompt-results) before you prioritise it.
  </Step>
</Steps>

## What's next

<CardGroup cols={2}>
  <Card title="Tracking recommendations" icon="chart-line-up" href="/results/recommendations/tracking">
    Approve a recommendation and follow how your standing moves across scans.
  </Card>

  <Card title="Prompt Results" icon="list-magnifying-glass" href="/results/prompt-results">
    Read the responses and prompts behind a recommendation, engine by engine.
  </Card>
</CardGroup>


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.