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

# About the Action Center

> What to do next to improve your AI visibility, and whether doing it worked.

The **Action Center** tells you what to do next, and then tells you whether doing it worked.

Most AI visibility tools stop at the first half. RankRaft AI holds your AI answers *and* your server logs, so after you act it can show whether an AI crawler came back to the page and whether the answers moved, against prompts you didn't touch.

## How actions are made

* **From rules, not from a chatbot.** Every card comes from a fixed rule applied to your own answers and crawler checks. No AI writes suggestions, and every card shows the count that triggered it.
* **One prompt per card.** Each content action targets one question, so the rest of your prompts stay untouched and can serve as a comparison.
* **One metric per card.** Each action declares a single target metric (mention rate, citation rate or prompt coverage) and only ever reports movement in that metric. Several actions can't all claim the same improvement.
* **Frozen wording.** A card's text is written when it's created and never rewritten, so it still says what it said when you acted on it.

Technical checks run from your first day, so even a new trial account has real cards. Content actions appear once your first scan has answers to compare.

## Reading a card

Each card shows:

* A category: **Technical**, **Content** or **Off-site**.
* **Impact** (High, Medium or Low) and **Effort** (Small, Medium or Large).
* The engines it should move.
* A **Why:** line with the evidence, for example *Named in 0 of 14 answers; Acme named in 9*.

Cards are ranked by expected effect on your score: the target metric's weight in the Visibility Score, times impact, adjusted for effort and how many prompts and engines it reaches. Prerequisites, such as confirming competitors or connecting your site, are always pinned on top.

## Working through actions

**1. Open a card**

The detail page has three sections: **What we found** (the evidence, traceable to specific prompts), **What to do**, and **What happened**.

**2. Do the work**

RankRaft AI never publishes anything to your site. You decide what ships.

**3. Click I've done this**

We record a baseline for the card's prompts at that moment and tell you when the result is due: *Recorded. We will have a result on 12 Nov.*

**4. Watch for the crawler**

Done cards show whether an AI crawler has re-read the page, for example *GPTBot re-read /pricing on 3 Nov*, or *No AI crawler has re-read /pricing yet — which is usually why nothing has moved.*

**5. Read the result**

When the measurement window closes, the card shows a verdict. See [Measuring results](/analytics/action-center/measuring-results).

**Hide** removes a card you don't want to act on. **Undo** moves a done card back to your list.

A problem that stops appearing in your data is marked stale rather than deleted, so your history stays intact.

> [!NOTE]
>
> Suggestions come from your own AI answers and crawler checks. Doing one can help, but no action guarantees AI visibility.


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