How the audit works
Every sentence in your report traces back to something a program actually checked on your website. This page explains what gets measured, what does not, and the rules that keep the difference honest.
Software measures. Writing explains. The two never swap jobs, and that separation is the reason a finding in your report can be checked rather than taken on faith.
Measured
"phone_links": []
"response_time_ms": 6210
"has_viewport_meta": trueFacts, not opinions. Whether a tappable phone link exists is not a judgment call, so no judgment is involved.
Explained
Your phone number is on the page as text, so tapping it on a phone does nothing. Someone standing outside your door has to copy it by hand or go back to Google — and that search is where they find someone else.
Plain English, written for you rather than for a developer. It can explain what was found. It cannot add anything that was not.
Six categories, on every scan. Everything here is read from your live site the way a visitor's browser would see it — nothing is installed, and nothing changes on your end.
Google's own Lighthouse scoring, run against your site on a simulated phone. Load time, layout stability, and how long someone waits before anything useful appears.
Whether the page is built to resize for a phone at all, text small enough to force pinch-zooming, images missing descriptions, and a missing tab icon.
Page title, description, headings, and the structured business markup that tells Google your hours, address, and phone number are what they claim to be.
Tap-to-call, visible hours and address, and whether ordering or booking is something a visitor can start without hunting for it.
What the site is built on, and signals a visitor reads as staleness — a copyright year several years behind being the most common.
A spot-check of the links on your homepage, to catch pages that have quietly stopped existing. Read the rule below on this one — it is stricter than it sounds.
An audit that overstates its findings is worse than no audit, because you would spend real money fixing something that was never broken. These are the rules that stop that.
Rule one
Links get sorted into two piles, and only one of them ever reaches you. A link is called broken when your server actually answered and said the page is gone. When the connection simply fails without any answer — which small hosts do at random, several times a day — it goes in a separate pile marked unverified, and it stays out of your report.
The lazy version of this check would report both as broken and look more thorough for it. It would also mean telling you a working page is dead, which is exactly the kind of thing you would pay to fix and find nothing wrong with.
Rule two
Some checks depend on outside services, and sometimes one is unavailable when your scan runs. When that happens the result is recorded as unmeasured, and the report is required to stay silent about it rather than assume the worst.
This is not theoretical. An early version treated an unmeasured value as an absent one and produced a finding about a security setting that was working fine the whole time. Silence on an unknown is the only honest option.
Rule three
No report quotes a number, estimates hours, or tells you what a problem is costing you. It cannot, honestly — nothing visible from outside a website reveals your average ticket, how many calls you get in a week, or how many of them turn into customers.
Anyone who hands you a free scan with a lost-revenue figure attached made that figure up. The audit's job is to show you the problem clearly enough to decide whether it is worth a conversation.
Worth being direct about the edges, since every audit tool has them and most do not say so.
The audit is free and there is nothing attached to it. If nothing in it is worth acting on, that is a useful answer too.
Get your free audit