---
title: "Methodology"
description: "How the AEO score is measured — recommendation presence, citation share, narrative accuracy and more."
canonical: "https://aeo.express/methodology"
dateModified: "2026-08-24"
---

# Methodology

Every score comes from real prompts run against real answer engines, parsed by a versioned, deterministic scoring configuration. Payment plays no part anywhere in this process.

## Score components and weights

| Component | Weight | What it measures |
| --- | --- | --- |
| Recommendation presence | 30% | How often leading answer engines recommend your company for relevant buyer questions. |
| Recommendation position | 20% | Where you land among competitors when you are recommended. |
| Citation share | 15% | How often your own domain is cited as a source, versus third parties. |
| Competitive share of voice | 15% | Your presence relative to named competitors across the same prompt set. |
| Narrative and entity accuracy | 10% | Whether models describe your product, category and claims correctly. |
| Content and citation readiness | 10% | Whether your public content gives models enough to cite confidently. |

## Confidence

Confidence is reported separately from the score itself, and reflects how many prompts and engines produced eligible observations. A low sample never lowers the score — it lowers confidence in that score.

## What can never affect this score

No purchase, contribution, supporter placement or founder number is readable by the scoring pipeline. The database roles enforcing scoring have no grant on billing data at all.

## Sitemap

- [Markdown sitemap](https://aeo.express/sitemap.md) — every public page, grouped.
- [llms.txt](https://aeo.express/llms.txt) — the same list in llmstxt.org form.
- [XML sitemap](https://aeo.express/sitemap.xml) — for classic crawlers.