# Acting on Gaps



In the [loop](/platform/ai-visibility/overview), a gap becomes a campaign, and
the campaign gets re-measured on the same prompts.

## Which Gaps to Take [#which-gaps-to-take]

Rank by decision impact, not by how far behind you are.

| Take first                                                                            | Leave                                                  |
| ------------------------------------------------------------------------------------- | ------------------------------------------------------ |
| Comparison prompts where a competitor is cited and you are not                        | Broad category prompts where six vendors are all named |
| Pricing and evaluation prompts with a clear factual answer                            | Prompts whose answers are opinion pieces               |
| Prompts with high coverage and low visibility: the page exists, the citation does not | Prompts nobody in your pipeline has ever asked         |
| Anything a rep has heard a buyer say out loud                                         | Anything only a tool suggested                         |

## From Gap to Work [#from-gap-to-work]

<Steps>
  <Step title="Scan first, so you have a baseline">
    `trigger_scan` before you publish. See
    [tracked prompts](/platform/ai-visibility/tracked-prompts#scans).
  </Step>

  <Step title="Brief with the prompt and the sources">
    Hand the [content agent](/platform/campaign-agents/content-agent) the prompt
    text, the competitors currently cited, and the
    [cited sources](/platform/ai-visibility/cited-sources).
  </Step>

  <Step title="Write the claim plainly and early">
    Answer engines quote sentences. Put the claim near the top, and write it so
    it still makes sense when it is quoted on its own.
  </Step>

  <Step title="Fix the sources you control in parallel">
    Directory listings and review-site descriptions are yours to edit.
  </Step>

  <Step title="Re-scan more than once">
    Pick the re-scan dates before you publish, so you are not choosing them
    after seeing a number you like.
  </Step>
</Steps>

## What to Expect [#what-to-expect]

**Uneven movement.** The models do not move together. That is what the
[per-model breakdown](/platform/ai-visibility/reading-the-score#per-model-breakdown)
is for.

**Occasional regressions.** Answers change on their own. Read trends, not
points.

<Info>
  How long a published page takes to show up in an answer engine is not something
  we can promise, and it is not published anywhere. Measure it on your own prompt
  set: baseline, publish, re-scan, and record what you saw. Your FDE can tell you
  what other teams have seen.
</Info>

## Closing the Loop on Revenue [#closing-the-loop-on-revenue]

Visibility is a leading indicator. Two things connect it to revenue:

* **[Traffic Intelligence](/platform/traffic/analytics)** attributes AI referrals
  by engine, so you can compare how ChatGPT visitors convert against organic
  ones.
* **[The conversion funnel](/platform/traffic/conversion-funnel)** carries them
  through to qualified leads and meetings.

<Tip>
  [SAM](/platform/sam/overview&#x29; reaches the visibility data, so
  &#x2A;"are we showing up in AI answers for \[prompt]?"* works from Slack, and
  [a schedule](/platform/sam/schedules) can post it weekly.
</Tip>