Scoring Methodology
This page explains how MeshWeave computes AEO, GEO, and AAX, how missing inputs are handled, and how to interpret each score as a diagnostic signal rather than a guaranteed outcome.
How to read this page
MeshWeave uses AEO, GEO, and AAX as diagnostic views into AI visibility risk. These scores do not guarantee ranking, traffic, or conversion. They show where AI systems may fail to crawl, interpret, cite, or act on your content.
Shared composite logic
All three scores use a weighted composite with re-normalization. Factors with no score are excluded, the remaining weights are redistributed, and the result is calibrated to compress inflated upper-range scores.
| Rule | Behavior |
|---|---|
| Weighted composite | Σ(score × weight) / Σ(weight) using only factors with non-None
scores |
| Missing factors | Weights are re-normalized across available factors |
| Calibration curve | 100 × (composite / 100)^1.15 compresses the upper range |
| Output | Capped at 100, rounded to 1 decimal place, plus an auto-only composite |
AEO — answer extraction risk
AEO measures how well your content is structured for AI answer engines such as featured snippets, AI Overviews, and voice assistants.
| Factor | Weight | Auto? |
|---|---|---|
| Capture Rate | 30% | No — manual input |
| Schema Implementation | 20% | Yes |
| Content Structure | 20% | Yes |
| Query Match | 15% | No — manual input |
| Voice Rate | 10% | No — manual input |
| Freshness | 5% | Yes |
| Score | Rating |
|---|---|
| 0–25 | Poor |
| 26–45 | Below Average |
| 46–65 | Average |
| 66–85 | Strong |
| 86–100 | Excellent |
GEO — generative discovery risk
GEO measures how well your content is positioned to be discovered, cited, and referenced by LLMs such as ChatGPT, Claude, and Perplexity.
| Factor | Weight | Auto? |
|---|---|---|
| Citation | 30% | No — manual input |
| Topical Authority | 20% | Yes |
| E-E-A-T Signals | 15% | Yes |
| Crawl Access | 15% | Yes |
| Content Depth | 10% | Yes |
| Entity Consistency | 10% | Yes |
| Score | Rating |
|---|---|
| 0–25 | Invisible |
| 26–45 | Emerging |
| 46–65 | Visible |
| 66–85 | Authoritative |
| 86–100 | Dominant |
AAX — agent usability risk
AAX measures how well AI agents can understand your site, assess it, and complete meaningful tasks across key pages.
| Factor | Weight | Auto? |
|---|---|---|
| Homepage Comprehension | 30% | Yes |
| Meta Optimization | 20% | Yes |
| Content Delta | 20% | Yes |
| llms.txt | 15% | Yes |
| Email Validation | 15% | Yes |
| Score | Rating |
|---|---|
| 0–24 | Opaque |
| 25–39 | Unclear |
| 40–59 | Readable |
| 60–79 | Clear |
| 80–100 | Fluent |
Automatic inputs
MeshWeave scores measurable technical signals from crawl data, structured data, content structure, entity consistency, crawl accessibility, and LLM-based page analysis where AAX is enabled.
Manual inputs
| Input | Used in | Purpose |
|---|---|---|
| Capture Rate | AEO | Featured snippets and AI Overviews |
| Query Match | AEO | Natural-language demand alignment |
| Voice Rate | AEO | Voice assistant answer selection |
| Citation | GEO | Observed LLM citation frequency |
Use this methodology as the reference for how the model works, then use your analysis page to see which factors are limiting your site right now.
Talk to an Expert