The 48-hour site audit: what I check and what I skip

This is the checklist I use to frame a site review. Run the first pass yourself today. If you want a human fit check, request a qualified triage; any custom teardown is scoped and priced after fit is confirmed.

Start with the site grader for an instant score, then use this checklist to understand what the score means.

Layer 1: Performance (does it load before they leave?)

Target: meaningful content in under 1.4 seconds on a mid-range phone over 4G. Tool scores help, but load behavior decides.

• Largest Contentful Paint. What is the hero made of? Unoptimized hero images and render-blocking scripts are the usual culprits. Next/image with AVIF/WebP and prioritized hero loading fixes most cases.
• Interaction delay. Heavy client JavaScript, chat widgets, and animation libraries that block the main thread. Defer non-critical scripts; keep the critical path lean.
• Layout shift. Images without dimensions, late-injected banners, web-font swaps. Reserve space, preload key fonts.
• Payload and caching. Megabyte bundles from page builders are common. Tree-shake, code-split, serve static edge-cached assets.

I check lab numbers plus a throttled mobile pass. Lab-measured means controlled conditions, not a promise about every visitor's network. Anyone quoting field guarantees from a lab test is overselling.

Layer 2: SEO (can anyone find it?)

Skipped by most rebuilds, expensive to retrofit.

• Titles and meta descriptions. Unique per page, intent-matched, under truncation limits. No duplicates, no "Home | Home."
• Headings and internal links. One H1 per page, logical H2 flow, descriptive anchors. Orphan pages with zero internal links do not rank.
• Sitemap and robots. Valid XML sitemap submitted, robots allowing the right paths, canonical tags where duplicates exist.
• Open Graph and social cards. Every key page needs a share-ready title, description, and image.
• Structured basics. Correct language tag, viewport, semantic landmarks. No fake review markup: I never ship aggregateRating without verifiable third-party ratings.
• Indexability blockers. Accidental noindex tags, client-rendered-only content, blocked asset paths.

Quick self-check: view source on your homepage. If the title, description, and canonical are missing or duplicated, that is day-one work in the $100–$500 Quick Fix range.

Layer 3: Accessibility (can everyone use it?)

Accessibility is conversion work. Everyone uses sites one-handed, sunlit, or aging at some point.

• Contrast. Body text at 4.5:1 or better. Muted gray on white fails constantly.
• Keyboard path. Tab through the page. Can you reach the menu, form, and CTA with visible focus?
• Labels and names. Every input has a real label, every icon button an accessible name. Placeholders are not labels.
• Alt text and headings. Informative images described, logical heading order, no empty links.

You do not need perfection to start. Fix contrast, labels, and keyboard reach first; those three cover most lost conversions.

Layer 4: Funnel (does it ask for the right thing?)

Fast, findable, and accessible pages still fail when the ask is muddy.

• One primary action per view. "Book the call" or "start checkout," not five equal buttons. Secondary actions styled secondary.
• Value before form. Headline states outcome, subhead states mechanism, proof sits within one scroll. Asking for details before establishing value inverts the trade.
• Form friction. Every field costs completions. Name, contact, one context field, then stop. Multi-step only when the logic earns it.
• Trust near the ask. Verifiable reviews, real project proof, clear warranty and reply-time expectations. My model: 7 featured concept builds with Lighthouse mobile scores measured 2026-10-15 and no client affiliation implied, plus 4 verified Freelancer.com orders and 3 named references (2 employment, 1 collaborator).
• Mobile and follow-up. CTA reachable by thumb, confirmation state after submit, and a notification path that cannot silently fail.

Browse work for funnel references across storefronts and dashboards, keeping in mind those are self-initiated concept builds for scope illustration.

How I rank the fixes

Tools list dozens of issues with equal alarm. I rank by expected revenue per unit effort:

1. Broken or blocking first. Forms that do not submit, checkout errors, noindex on money pages.
2. High-traffic friction next. Slow hero, confusing CTA, or trust gap on the page that already gets visits.
3. Compounding SEO after. Titles, internal links, sitemap hygiene that pay monthly.
4. Polish last. Animations, decorative upgrades, nice-to-have sections.

Each audit item ships with severity, effort size, and the page it affects. Keep the list even if we never work together; it sequences any developer's time better.

What a qualified custom triage covers (and does not)

Covers: the four layers above on your homepage plus one key conversion page, ranked fixes, effort sizes, and a fixed-price path for the top items.

Does not cover: full-app security review, penetration testing, or legal compliance certification. If the audit surfaces those needs, I say so and point you to the right specialist.

The bottom line

Performance decides if they stay. SEO decides if they arrive. Accessibility decides if they can act. Funnel decides if they do. Check in that order, fix by revenue per effort.

---

Want a second set of eyes? Request a qualified triage and I’ll reply with the fit and next step, or run the instant grader right now for a first score in under a minute.