# AI Visibility Rankings Dashboard

The AI Visibility Rankings dashboard shows the grounding searches (fan-out queries) performed during your AI Visibility Reports. These are the actual searches AI assistants run to gather information before responding to users.

Different assistants ground in different search engines. **ChatGPT and Google Gemini ground in Google**, while **Claude grounds in Brave**. Spyglasses checks each grounding search in the engine the assistant that produced it actually uses, so your ranks reflect what that assistant could have seen.

## What You'll Learn

In this guide, you'll learn:

- What grounding searches are and why they matter
- How to interpret ranking data
- How to use impact scores to prioritize
- How to view competitor rankings

## Understanding Grounding Searches

When you ask an AI assistant a question like "What project management tools work best for remote teams?", the AI doesn't just rely on its training data. It performs searches to gather current information. These searches are called **grounding searches** or **fan-out queries**.

For example, that question might trigger searches like:
- "best project management tools remote teams 2026"
- "project management software features comparison"
- "top rated tools for distributed teams"

**Ranking for these searches is one of the most straightforward ways to improve AI visibility.** If your content appears in the top results for a grounding search (top 30 on Google, top 20 on Brave), you're much more likely to be mentioned in the AI response.

## Rankings Table

The main table shows each grounding search with:

| Column | Description |
|--------|-------------|
| **Search Query** | The exact search AI performed |
| **Query Count** | How many discovery queries triggered this search |
| **Source** | The assistant that produced the grounding search (ChatGPT, Gemini, or Claude) |
| **Rank** | Your position in that assistant's search engine — Google for ChatGPT/Gemini, Brave for Claude (1-30 on Google, 1-20 on Brave, or blank if not ranking) |
| **Impact Score** | Calculated priority for optimization |

### Interpreting Ranks

- **1-10**: Strong position - you're likely to be cited
- **11-20**: Moderate position - may or may not be included
- **21-30**: Weak position - likely not being considered
- **Blank**: Not ranking in the top results - definitely not cited (for this query)

The engine column tells you which index a rank applies to. A search ChatGPT or Gemini performed is checked in Google; a search Claude performed is checked in Brave (which returns up to 20 results, so Brave ranks top out at 20). Use the engine filter to focus on one index at a time.

## Impact Score Explained

The Impact Score helps you prioritize which grounding searches to target. It's calculated based on:

1. **Query frequency**: Searches used in more discovery queries have higher impact
2. **Current rank**: Lower or missing ranks have more improvement potential
3. **Competitor presence**: Searches where competitors rank well are higher priority
4. **Search relevance**: Searches more central to your category matter more

**High impact scores indicate the best optimization opportunities.** Focus your content efforts on these searches first.

### Using Impact Scores

Sort the table by Impact Score to see your highest-priority opportunities. For each:

1. Check if you have existing content targeting this search
2. If yes: Optimize the page for better ranking
3. If no: Create new content specifically targeting this search
4. After ranking improvements, check [recrawl frequency](/docs/dashboards/recrawl-frequency) and when appropriate, run a new report to verify AI now picks up that source

## Viewing Competitor Rankings

Click on any grounding search row to expand it and see which competitors rank for that query.

For each competitor you'll see:
- Their ranking position
- The specific page URL that ranks
- Whether they outrank you

### Competitive Insights

This data shows:
- **Which competitors are winning**: Consistently high-ranking competitors are dominating AI recommendations
- **What content works**: The pages that rank show you what type of content AI values
- **Gap opportunities**: Searches where competitors rank but you don't are priority targets

## Filtering and Sorting

Use filters to focus on specific segments:

- **Has rank / No rank**: Filter to searches where you do or don't rank
- **Impact score range**: Focus on high or low priority searches
- **Query type**: Filter by category (best in category, comparison, etc.)

## Taking Action

Working a grounding search is a loop, not a one-off edit. The same five steps apply whether you're missing from a search entirely or sitting at the bottom of it:

**Pick → score → revise → recrawl → re-measure.**

1. **Pick the search** by impact score, so you're spending effort where a ranking change would actually change an answer.
2. **Score a page against that exact query** in the [Citation Optimizer](/docs/dashboards/citation-optimizer), which runs the page through Spyglasses' model of the ChatGPT citation pipeline gate by gate and tells you *which* gate the page falls out of. That is the step that replaces guessing.
3. **Apply the revision and re-score.** Re-scoring is free and unlimited, so iterate until the readiness banner says the content is publish-ready or has plateaued.
4. **Wait for the recrawl.** A change no crawler has fetched cannot move a rank. Check [recrawl frequency](/docs/dashboards/recrawl-frequency) to see when the engines and AI platforms last pulled that URL.
5. **Re-measure.** Run a new [AI Visibility Report](/docs/dashboards/ai-visibility-reports), or wait for the next weekly grounding-gap run to re-check the search and update its rank band.

### Getting to the Optimizer in One Click

You don't have to copy a query across by hand. On the [Brand Dashboard](/docs/dashboards/brand-dashboard), the **SEO opportunities** drill-down lists every tracked grounding search, and each row carries an **Optimize** button. It opens the Citation Optimizer at Step 1 with that search already selected as a **tracked** (verified) fan-out, carrying its impact score, its gap flag and the competitor pages that rank for it. You go straight to choosing what to score.

The button only appears on searches ChatGPT produced. The optimizer scores the ChatGPT pipeline, so a Gemini- or Claude-only fan-out has no pipeline to score against yet.

### For High-Impact Searches Where You Don't Rank

1. **Expand the row** to see which competitors rank and which of their pages does it. That tells you what the search is actually asking for, in the form the engine already rewards.
2. **Find your closest page.** The optimizer's Source step ranks your own pages by how well their content matches the query, so it will tell you which page is nearest to competing — often not the one you assumed.
3. **If a page is close, score it** against the query. The SERP gate will show the ranking gap as a fail, but the run keeps going, so you still get content guidance and the competitor pages to study.
4. **If nothing is close, score the draft before you publish.** Paste raw markdown into the Source step and work the gates while the page is still cheap to change.
5. **Check for overlap.** If two of your own pages are flagged as overlapping for this query, consolidate or differentiate them first — two half-strength pages competing for the same citation lose to one strong one.
6. **Publish, confirm the recrawl, then re-measure.**

### For Searches Where You Rank Poorly

1. **Identify the page that currently ranks.** That's the page in the race; improving a different one starts you over.
2. **Score that page rather than rewriting it blind.** A page at #24 can be there for very different reasons — thin or unchunkable content, a poor match to what the query is asking, or simply an authority gap on the SERP — and the gate strip names which one.
3. **Fix the failing gate, not the whole page.** The change log ties every edit in a rewrite back to the recommendation that motivated it, so you can see whether you're addressing the constraint or redecorating around it.
4. **Re-score to measure the lift** before you ship anything. This costs nothing and stops you from publishing a revision that didn't actually move the gates.
5. **Ship, wait for the recrawl, and watch the rank band** on the SEO opportunities drill-down, which compares each search against its own previous run rather than against a fixed date.

### Proving the Work Moved Something

Ranking changes are slow and noisy, so pair them with a measurement that isn't. Run an [AI Readiness Site Audit](/docs/dashboards/ai-readiness-audit) on just the pages you're changing, before the work and again after. The per-page, per-factor scoring shows which specific changes moved which factor, at a cadence you control rather than one you wait for.

[Measuring AI Visibility Before and After a Content Refresh](/blog/measuring-ai-visibility-before-after-content-refresh) walks through that comparison in detail — how to scope the audit to only the refreshed pages, how to read the per-factor breakdown, which factors a content change realistically moves and which it doesn't, and how to catch a refresh that quietly regressed a factor.

Finally, **mark the publish date as an [annotation](/docs/dashboards/annotations)**. When the rank band moves three weeks later, the marker on the chart is what connects it back to the work.

## Best Practices

### Target Multiple Grounding Searches Per Content Piece

A single comprehensive guide can rank for multiple related grounding searches. Rather than creating thin pages for each search, build thorough content that covers the topic completely.

### Prioritize High-Query-Count Searches

Searches that appear in multiple discovery queries give you the best leverage. Ranking for one search that's used in 5 queries is more valuable than ranking for 5 searches used once each.

### Monitor Changes Over Time

Rankings fluctuate. After optimizing content, track whether your rankings improve in subsequent reports. Sustained improvement indicates successful optimization.

### Optimize for the Right Engine

ChatGPT and Gemini ground their searches in Google, so Google SEO drives most of your AI visibility. Claude grounds in Brave, so if your audience uses Claude, make sure your content also ranks in Brave. Brave draws heavily on its own independent index, so a page that ranks in Google won't automatically rank in Brave — check both when Claude is important to you.

## Related

- [AI Visibility Reports](/docs/dashboards/ai-visibility-reports) - Run reports to generate ranking data
- [Discovery Queries](/docs/dashboards/discovery-queries) - Manage the queries that trigger grounding searches
- [Brand Dashboard](/docs/dashboards/brand-dashboard) - The SEO opportunities drill-down, with per-search movement and the Optimize hand-off
- [Citation Optimizer](/docs/dashboards/citation-optimizer) - Score a page against one of these searches and rewrite it to close the gap
- [Recrawl Frequency](/docs/dashboards/recrawl-frequency) - Confirm a change has been fetched before you re-measure
- [Measuring AI Visibility Before and After a Content Refresh](/blog/measuring-ai-visibility-before-after-content-refresh) - The before/after audit workflow in full