Methodology
How the Startup Security Score works
The Startup Security Score is a 0–100 rating of a company's externally visible security. It combines six passive checks. Every finding has a severity, a specific fix and the SOC 2 control it relates to. This page documents exactly what we check and how we score it.
Reviewed by Jay Dabhi, CISM · Cybersecurity Analyst
Updated
1. Email authentication (SPF, DKIM, DMARC)
SOC 2 CC6.6- How we check
- We query your public DNS for an SPF record on the root domain, a DMARC record at _dmarc.<domain>, and DKIM keys on common selectors.
- Why it matters
- Missing or permissive email authentication lets attackers send email that appears to come from your domain. It's the most common signal enterprise reviewers check first.
- What we flag
- No SPF record, multiple SPF records, or an SPF record ending in +all or ?all
- No DMARC record, or DMARC set to p=none (monitor only)
- No DKIM key on common selectors (low severity; custom selectors may be a false positive)
2. TLS and HTTPS
SOC 2 CC6.7- How we check
- We open one TLS connection to your website, read the certificate and negotiated protocol, and check whether plain HTTP redirects to HTTPS.
- Why it matters
- Invalid or expiring certificates break trust and availability. Outdated protocols and missing redirects expose visitors to interception.
- What we flag
- Invalid certificate (critical)
- Certificate expiring within 14 days
- TLS 1.0 or 1.1 negotiated
- HTTP that doesn't redirect to HTTPS
3. Browser security headers
SOC 2 CC6.7 / CC8.1- How we check
- We read the response headers of a single normal request to your homepage.
- Why it matters
- Headers such as HSTS, Content-Security-Policy and frame protection are cheap defenses against downgrade attacks, cross-site scripting and clickjacking.
- What we flag
- Missing Strict-Transport-Security
- Missing Content-Security-Policy
- No X-Frame-Options or CSP frame-ancestors
- Missing X-Content-Type-Options
4. Exposed subdomains
SOC 2 CC6.1- How we check
- We read public Certificate Transparency logs (via crt.sh) for certificates issued to your domain. We send nothing to your servers for this check.
- Why it matters
- Staging, admin and internal tools are often less hardened than production and are a common entry point for attackers.
- What we flag
- Subdomains named like dev, staging, test, admin, internal, jenkins, grafana or vpn
- An unusually large number of public hosts (informational)
5. Leaked secrets in public code
SOC 2 CC6.1- How we check
- We search public GitHub code for your domain next to words such as password, api_key and secret. We report counts and repository names only, never the matched values.
- Why it matters
- Credentials committed to public repositories are found by attackers within minutes.
- What we flag
- Public code mentioning your domain alongside secret keywords
6. Disclosure hygiene
SOC 2 CC2.3 / CC7.1- How we check
- We check response headers for software version numbers and fetch /.well-known/security.txt.
- Why it matters
- Version numbers help attackers pick exploits. A security.txt file (RFC 9116) tells researchers how to report vulnerabilities to you.
- What we flag
- Server or X-Powered-By headers containing version numbers
- No security.txt with a Contact line
7. Scoring and grades
Every scan starts at 100. Each finding subtracts points by severity, and the score never goes below 0. Informational findings are shown but don't affect the score.
| Severity | Points |
|---|---|
| Critical | −25 |
| High | −12 |
| Medium | −6 |
| Low | −2 |
| Info | 0 |
| Grade | Score |
|---|---|
| A | 90–100 |
| B | 75–89 |
| C | 60–74 |
| D | 40–59 |
| F | 0–39 |
8. What "passive" means
We only read public information: DNS records, certificate transparency logs and public GitHub code. We also make the same two requests any visitor's browser would make (your homepage and security.txt). We do not port-scan, fuzz, brute-force or send exploit payloads. Active testing is a penetration test. That happens only in PentestLive after the domain owner verifies ownership, or in a scoped engagement with Bluefire Redteam.
Limits: a passive scan cannot see authentication flaws, broken access control, injection bugs or cloud misconfigurations. Those need a penetration test.
See your own score
Free, about a minute, no signup.