Methodology — how the audit scores

Every Branded Audits report is generated by a deterministic, rule-based audit engine. No LLM is in the scoring loop. No external paid APIs are called. The same URL gets the same score every time. Here's exactly how it works.

1. Sections and weights

The audit is divided into five sections. Each section's score is rolled up into the overall score using these weights:

Section	Weight	Why this weight
SEO	30%	Strongest predictor of organic traffic — the outcome most agencies sell on.
Performance	25%	Core Web Vitals are now a Google ranking factor; page speed directly affects bounce rate.
Accessibility	15%	Compliance + reach. Affects ~15% of users with disabilities.
Security	15%	Trust signal + insurance against breaches that would tank trust.
Mobile	15%	~60% of US web traffic. Mobile-first indexing is Google's default.

These weights total exactly 100%. They are baked into the engine; we don't expose per-tenant weight tuning today.

2. Finding statuses

Within each section, individual checks each produce one of three statuses:

PASS = 100 points
WARN = 50 points
FAIL = 0 points

A section's score is the arithmetic mean of its findings' point values, rounded to the nearest integer. The overall score is the weighted sum of section scores using the table above, again rounded to the nearest integer.

3. Letter grade thresholds

Grade	Score range	Plain-English meaning
A	90–100	This site is in great shape. Lead with strengths in the proposal.
B	80–89	Solid foundation, clear room to grow. Strong agency engagement potential.
C	70–79	Leaving meaningful conversions on the table. Easy wins available.
D	60–69	Multiple high-impact issues. Best case for a paid engagement.
F	0–59	Serious gaps. Recommend a full rebuild or aggressive remediation.

4. Specific checks performed

Performance

TTFB (time to first byte): < 600 ms = pass, 600–1500 = warn, > 1500 = fail.
Page weight (HTML bytes): < 500 KB = pass, 500–2000 = warn, > 2000 = fail.
HTTP compression: gzip or brotli on the response = pass, none = fail.
Cache-Control header: present and not no-store = pass.
Render-blocking scripts in <head>: 0 = pass, 1–2 = warn, 3+ = fail.

SEO

Title tag: present + 30–65 chars (target 50–60).
Meta description: present + 70–165 chars.
H1 tag count: exactly 1 = pass.
Canonical link tag: present = pass.
Open Graph tags: og:title + og:image both present = pass.

Accessibility

Image alt text: every <img> without aria-hidden has an alt attribute.
HTML lang attribute: present.
Form labels: every visible input has either an aria-label, aria-labelledby, or a <label for="..."> pairing.
Skip-to-content link: present.

Security

HTTPS: final URL after redirects starts with https://.
HSTS: Strict-Transport-Security header present.
Content-Security-Policy: header present.
X-Content-Type-Options: present (typically nosniff).
X-Frame-Options: present.
Referrer-Policy: present.
Permissions-Policy: present.
Server header version-leak: warns if header reveals software version (e.g. nginx/1.18.0).

Mobile

Mobile viewport meta tag: name="viewport" with width=device-width.
Responsive images: presence of <picture> elements or srcset attributes.
Apple touch icon: <link rel="apple-touch-icon"> present.

5. What the audit explicitly does NOT do

For credibility, here's what we don't claim to measure:

Site-wide crawls. Single-page audits only. Use Sitebulb or Screaming Frog for crawl-based analysis.
JavaScript execution. The fetch is HTML-only. Single-page apps render only what's in the static HTML.
Real-user metrics. No LCP / CLS / INP from real visitors — that requires Google's CrUX data or instrumentation.
Backlinks. No off-page SEO data. Use Ahrefs or Moz for that.
Keyword rankings. No SERP-position tracking.
Continuous monitoring. On-demand only. The result is a snapshot, not a stream.

6. Report format

Every audit produces a 5-page PDF (or HTML, if the build environment lacks the WeasyPrint native libs):

Cover page — agency logo, brand color gradient, audited URL, overall grade + score, "prepared for [client]" + "prepared by [agency]".
Executive summary — overall verdict, section-score bars, top 5 priorities (failing checks first), top 5 wins (passing checks).
Performance + SEO — every finding with status badge.
Accessibility + Mobile — same format.
Security + Next steps — security findings + a "book a 15-minute call" CTA pointing back at the agency's contact info.

7. Versioning

The scoring rubric is versioned. When checks are added or weights are adjusted, existing audits keep their original scores; new audits use the new rubric. Today's rubric is v1, deployed 2026-04-30.

Try it on one of your prospect's URLs.

The methodology only matters if the result is something you'd send a client. See for yourself.

Run a free audit →