No alert fires when a visitor gives up on a slow page, skims past a vague headline, or abandons an eleven-field form. Those failures are silent, which is why most owners never see them. The scan exists to record them against fixed thresholds, in ninety seconds.
90s
end to end
40+
checks run
6
categories
0
signups to see it
Nothing to install and nothing to connect. The scanner requests the pages your server already serves, measures them against fixed thresholds, and returns one number.
01 — Pull
read-only · nothing installed
02 — Analyze
03 — Deliver
Audit report
68/100
Needs worktop fix: consolidate competing CTAs
40+ checks in · one score out
Engine 01
Google Lighthouse loads your site on a throttled mobile connection and records largest contentful paint, layout shift, page weight, and server response time. Field data from real Chrome users takes priority; lab values fill the gaps. Twelve checks, each threshold fixed in advance.
→ feeds: performance · media
Engine 02
A parser fetches your homepage and contact page and reads the raw HTML. Title length, schema blocks, tap-to-call links, form field counts, navigation depth: twenty-one checks on what the markup states before any styling is applied.
→ feeds: seo · conversion mechanics · trust
Engine 03
A language model scores a screenshot and your copy against anchored rubrics. The prompt spells out what a 2 looks like and what an 8 looks like. Temperature is set to zero, so the same site returns the same verdict. Its share of the composite is capped at 25 percent.
→ feeds: messaging · hero · trust placement
One number from 0 to 100. It is a weighted mean of six category scores, and each category is a weighted mean of its checks: pass counts 100, warn counts 50, fail counts 0. Machine-measured checks carry roughly 75 percent of the total. Nothing is estimated. Any check that cannot be measured is excluded, and the remaining weights renormalize.
90 and above, Excellent. 75 to 89, Good. 60 to 74, Needs Work. Below 60, At Risk. The bands are fixed, and the same URL scanned twice in one day returns the same number to the digit. That consistency is the point.
Conversion and CTAs carries 25 points, the heaviest weight, because a fast site that never asks for the call still loses the lead. Performance 20. Trust and Contact 20. SEO Foundations 15. Media 10. Messaging 10. The weights follow revenue impact.
Every check carries a fixed impact rating from 1 to 5 and an effort rating of low, medium, or high. Failing and warning checks sort by impact first, then effort. The top five become your fix list: what was found, why it costs you leads, and what the fix involves.
If your site blocks the scanner, the report says so. If an engine fails, its categories read “could not be measured” rather than a padded score. A number you cannot trust is worse than no number.
Composite allocation
Σ = 100 · machine-measured ≈ 75% · ai capped at 25%
“Welcome to our website” in place of the service, the audience, and the outcome. The visitor decides in one glance, and that glance is spent.
The number renders as plain text with no tel: attribute. On the phone in the visitor's hand, calling you means copy and paste. Most will not bother.
When every action carries the same visual weight, the visitor chooses none of them. One dominant action per screen.
Testimonials sit in a carousel at the bottom of the page while the CTA stands alone at the top. Proof works when it sits beside the button.
Every extra field costs completions. Five inputs is the ceiling before completion rates fall.
Each of these maps to a check the scan runs automatically. No judgment calls and no exceptions.
An audit that ends with a PDF changes nothing. The scan is the instrument. This is the procedure that follows it.
Step 01
Ninety seconds, more than forty checks, zero opinions. The scan records what is measurable: where the site is slow, where it is silent, and where it asks too much of a visitor.
Step 02
Every finding carries a fixed impact and effort rating. The top five become your fix list, the shortest path from the current score to the one the site should hold.
Step 03
Twenty minutes on a call. We walk the full report, then cover what the scan cannot reach: traffic sources, keyword rankings, and competitor gaps.
Step 04
Hero rewrites, CTA consolidation, trust placement, speed work. Fixes ship in impact order. Small fixes ship in days.
Step 05
Same URL, same checks, same thresholds. The difference between the first score and the new one is the work, measured by the tool that found it.
The free scan covers what a machine can measure. The diagnostic call covers the rest: traffic, keywords, and the competitive picture, read by a person.
20 minutes · a person walks the report with you