Audit Methodology
Transparency is core to HiveAudit. This page documents our technical approach, scoring logic, and the standards we use to evaluate websites.
Last Updated: June 30, 2025
How We Audit
Technical Execution
Crawl Behavior: Our engine performs a static HTML analysis. It identifies itself as HiveAuditBot/1.0.
JavaScript: Currently, we do not execute client-side JavaScript. We audit the source as delivered by the server.
Redirects: We follow up to 5 consecutive redirects before timing out.
Timeouts: Scans time out after 8 seconds of inactivity.
Scope & Limits
Submission: Most checks (SEO, A11y) are performed on the specific URL submitted.
Discovery: Sitemap checks discover and ping up to 50 internal URLs.
Crawl Depth: Discovery scans are limited to a depth of 3 levels.
Rate Limiting: Scans are limited to 5 per minute per unique IP address.
Category Methodologies
SEO
Key Checks
- Title presence & length
- Meta description quality
- H1-H6 hierarchy
- Canonical tags
- Alt attributes
Standard
Google Search Central
Accessibility
Key Checks
- Image Alt text
- Form labels
- Lang attributes
- ARIA role validity
Standard
WCAG 2.2 Level AA
Security
Key Checks
- HSTS configuration
- CSP
- X-Frame-Options
- SSL status
Standard
OWASP Secure Headers
Technical
Key Checks
- Viewport config
- Robots.txt validity
- Sitemap structure
- Tag detection
Standard
W3C Best Practices
Scoring Logic
Scores are calculated using a weighted average of passed checks. Severity influences weight:
- Critical: -20 points (e.g., missing Lang attribute)
- High: -10 points (e.g., missing Page Title)
- Moderate: -5 points (e.g., missing Meta Description)
- Opportunity: 0 points (informational only)
Note: Scores are indicators of technical health, not guarantees of ranking.
Known Limitations
What We Do Not Check:
- Deep server-side code or databases
- Client-side JavaScript performance
- Visual design quality
- Content plagiarism
- Login-protected pages
False Positives:
Automated tools occasionally misinterpret complex implementations. If you believe a finding is incorrect, please contact support with the URL and evidence.