AI Visibility Score (AVS)

A structured, repeatable protocol for measuring how visible your entity is to AI systems. The AI Visibility Score measures both the AEO retrieval path (does AI cite you when searching?) and the GEO generation path (does AI know you from training data?). By testing both paths, you gain a complete picture of your AI presence. Run this protocol monthly to track progress and identify which optimization strategies are working.

The AVS operates on a 0–10 scale, with 5 points allocated to each path. This balanced approach recognizes that both retrieval and generation matter for comprehensive AI visibility. A brand that excels at one but fails at the other has significant gaps in its AI strategy.

AEO Score — Retrieval Path

The AEO (Answer Engine Optimization) score measures how well AI systems can find and cite your content when they have access to the web. Test with web search turned ON — this simulates the most common user experience where AI assistants retrieve real-time information to answer queries.

For each query you test, record the following metrics:

• Mentioned (1 point): Was your entity mentioned anywhere in the response? This is the baseline — if AI does not mention you at all, you have zero visibility for that query.
• Cited (1 point): Was your website linked or explicitly cited as a source? Citations drive traffic and establish authority. Many AI systems now provide source links, making this metric increasingly important.
• Position (1 point): Was your entity in the first recommendation or top three results? Position matters because users often focus on the first few suggestions. Score 1 point if you appear in the top three mentions.
• Accuracy (1 point): Were the facts about your entity correct? Inaccurate information can harm your brand. If the AI states incorrect pricing, outdated features, or wrong descriptions, you lose this point.

Maximum AEO score per query: 4 points. A perfect score means you are mentioned, cited, positioned prominently, and described accurately.

GEO Score — Generation Path

The GEO (Generative Engine Optimization) score measures what AI systems know about you from their training data alone. Test with web search turned OFF — this reveals whether your entity has made it into the AI model's knowledge base, independent of real-time retrieval.

For each query you test, record the following metrics:

• Mentioned (1 point): Does the AI know your entity exists? When asked about solutions in your category, does the AI include you in its response? This tests whether you have achieved sufficient web presence to be captured in training data.
• Recommended (1 point): Does the AI actually recommend your entity as a solution? There is a difference between being mentioned in passing and being actively suggested. This point rewards genuine endorsement.
• Confidence (1 point): Does the AI speak about your entity with confidence, or does it hedge with phrases like "I think" or "there may be"? Confident responses indicate strong, consistent information in the training data. Hedging suggests limited or conflicting information.

Maximum GEO score per query: 3 points. A perfect score means the AI knows you exist, recommends you, and speaks about you with authority.

The Testing Protocol

Follow this step-by-step protocol to generate consistent, comparable scores each month:

1. Select 5 target queries: Choose queries that represent your ideal customer's search behavior. Include a mix of category queries ("best CRM software"), comparison queries ("Salesforce vs HubSpot"), and problem queries ("how to manage customer relationships"). See target queries for detailed guidance on query selection.
2. Test on 4 platforms: Run each query on ChatGPT, Gemini, Perplexity, and Claude. Different platforms have different training data and retrieval methods, so cross-platform testing reveals your true visibility.
3. For each platform, test both modes: First test with web search ON to measure your AEO score. Then test with web search OFF to measure your GEO score. This requires 40 total tests (5 queries × 4 platforms × 2 modes).
4. Record scores per query per platform: Use a consistent tracking template to capture all data points. Record both raw scores and notes about positioning or accuracy issues.
5. Calculate averages: Sum your points and apply the formulas below to generate your combined AVS.

Use this tracking template to record your results:

| Query | Platform | Mode | Mentioned | Cited/Recommended | Position/Confidence | Accuracy | Score |
|-------|----------|------|-----------|-------------------|---------------------|----------|-------|
| "best project management software for startups" | ChatGPT | AEO | 1 | 1 | 1 | 1 | 4/4 |
| "best project management software for startups" | ChatGPT | GEO | 1 | 1 | 0 | - | 2/3 |
| "agile tools for remote teams" | Gemini | AEO | 1 | 0 | 0 | 1 | 2/4 |

The template shows three example rows with actual scores. Replicate this structure for all 40 tests in your monthly audit. Note how the AEO mode uses all four criteria while the GEO mode uses only three — accuracy is not measured for GEO since there is no external source to verify against in that mode.

Calculating the Combined AVS

Once you have completed all tests, calculate your combined AI Visibility Score using these formulas:

AEO Component (0–5 scale): Take the total AEO points earned across all queries and platforms, divide by the maximum possible points, then multiply by 5 to normalize to a 0–5 scale.

Formula: (total AEO points) / (queries × platforms × 4) × 5

GEO Component (0–5 scale): Take the total GEO points earned across all queries and platforms, divide by the maximum possible points, then multiply by 5 to normalize to a 0–5 scale.

Formula: (total GEO points) / (queries × platforms × 3) × 5

Combined AVS: Add both components together for your final score on a 0–10 scale.

Combined AVS = AEO component + GEO component

Example Calculation:
===================

Test Setup:
- 5 target queries
- 4 platforms (ChatGPT, Gemini, Perplexity, Claude)
- Total tests: 5 queries × 4 platforms = 20 tests per mode

AEO Results (Search ON):
- Total AEO points earned: 52
- Maximum possible: 20 tests × 4 points = 80
- AEO component: (52 / 80) × 5 = 3.25

GEO Results (Search OFF):
- Total GEO points earned: 36
- Maximum possible: 20 tests × 3 points = 60
- GEO component: (36 / 60) × 5 = 3.00

Combined AVS:
- AVS = 3.25 + 3.00 = 6.25 out of 10

In this example, the entity scores 6.25 out of 10, indicating moderate visibility across both paths. The AEO and GEO components are relatively balanced, suggesting consistent but incomplete optimization on both fronts.

Interpreting Results

Your AEO and GEO scores reveal different aspects of your AI visibility. Use this diagnostic framework to understand what your scores mean and what actions to take:

| AEO Score | GEO Score | Diagnosis                    | Action                                                      |
|-----------|-----------|------------------------------|-------------------------------------------------------------|
| High (4-5)| High (4-5)| Strong dual-path presence    | Maintain and expand to new queries                          |
| High (4-5)| Low (0-2) | Crawlable but not in training| Content is accessible but needs broader distribution for GEO|
| Low (0-2) | High (4-5)| Known but no citable source  | Entity is in training data but website lacks AEO optimization|
| Low (0-2) | Low (0-2) | Not visible on either path   | Start with AEO fundamentals — structured data, crawlability |

Strong dual-path presence (High AEO, High GEO): This is the ideal state. AI systems both know about you from training data and can find and cite your content when searching. Your optimization work is paying off. Focus on maintaining this position by continuing your content strategy and expanding to capture new target queries. Consider testing additional query variations to broaden your footprint.

Crawlable but not in training data (High AEO, Low GEO): Your content is technically accessible and AI crawlers can find it, but your entity has not yet made it into model training data. This is common for newer brands or those with limited broader web presence. Continue AEO optimization while also focusing on earning mentions, reviews, and coverage on authoritative third-party sites. This external validation helps ensure inclusion in future training data updates.

Known but no citable source (Low AEO, High GEO): AI systems know about your entity from training data, but your website is not optimized for retrieval. This often indicates technical AEO issues — perhaps your site blocks AI crawlers, lacks structured data, or has content that is difficult to extract. Audit your technical AEO implementation immediately. You are leaving significant visibility on the table.

Not visible on either path (Low AEO, Low GEO): This is the starting point for many entities. You need to build from fundamentals. Start with AEO basics — implement structured data, ensure your site is crawlable by AI systems, create content that directly answers common queries in your space. As your AEO presence grows, your GEO presence will follow in future training data updates.

Monthly Tracking

Run the full AVS protocol monthly on the same queries. Consistency is essential for meaningful trend analysis. Using identical queries each month allows you to isolate changes in AI behavior from changes in your query selection.

Track your combined AVS over time to measure the impact of your optimization efforts. Expect slow movement — AI training data updates quarterly at best, and sometimes less frequently. However, retrieval improvements can show results much faster since they depend on crawler activity rather than model retraining.

Here is what to expect for different types of changes:

• AEO improvements: Changes to structured data, crawlability, and content organization can appear in retrieval results within 1–4 weeks after AI crawlers reindex your site. Some platforms like Perplexity may reflect changes within days.
• GEO improvements: Changes to your broader web presence, third-party mentions, and entity authority take 3–6 months to appear in AI training data. Major models update their training data on varying schedules, so GEO improvements appear gradually.
• New content: Fresh content may appear in retrieval-based responses within days on platforms with real-time indexing. However, the same content will not affect GEO scores until the next training data update.

For guidance on selecting the most effective queries for your AVS testing, see choosing target queries. Once you understand your scores, visit the implementation guide for specific optimization strategies based on your diagnostic profile.