HighCategory
Largest image loads last on mobile
Visual audit.
A visual audit is a black box audit of your whole website, whether it is a store, a brochure site or a SaaS product: I see exactly what a customer, a search engine or an AI assistant sees, and nothing more. Every page template is crawled and measured on a laptop and a phone, and your competitors are measured the same way on the same day. No logins, no accounts, no source code, nothing asked of your team.
AI-generated
MediumSearch
Filters reset on the back button
LowProduct page
Two titles compete for one query
Example findings
The audit ladder
Read in order: each audit takes more access than the one before it. All prices exclude VAT.
What it delivers
Every template, as a customer sees it
What it covers
- Every page template crawled, on a laptop and a phone
- Speed measured separately for each template, against Google’s thresholds
- Navigation, listing depth, filtering and site search
- Product pages against what a buyer needs to decide
- The basket, checkout and the customer account area
- Legal, delivery, returns and trust pages
- Structured data, and whether AI assistants can read and cite you
- Two competitors measured the same way, on the same day
- Every finding ranked: what it is, what it costs, how to fix it, how long and how much
What it does not
- How many sales each finding cost: that needs your analytics, in the backend audit
- Anything behind a login, or the admin
- Why the code behaves as it does
- The servers, hosting and database
- Carrying out the fixes, quoted separately once you have the report
AI-generated
Fixed
£950
Visual audit · black box, no logins
What arrives: a report, not a dashboard
- A one-page verdict: what was found and what matters most
- The store scored across nine areas, from findability to AI readiness
- Lighthouse and Core Web Vitals, template by template
- Every finding ranked, each with the page it was seen on and a screenshot
- How you compare with two competitors, figure against figure
- What I would do first, and in what order
- Sent password protected, through a link that works once
Not covered
- A sales figure against each finding: that needs the backend audit
- A task list for a developer: the fixes are quoted separately
Why every template is measured
One average hides the slow page
Two pages from the same store, timed the same way on the same day. The diamond is how long a customer on a phone waited for the main picture to appear. The hairline is where Google’s own verdict turns from ‘needs improvement’ to ‘poor’.
4.3s
Google’s poor line 4s
0s8s
Method · Largest contentful paint, timed with Google’s free testing tool set to a mid range Android phone on a normal mobile signal.
17.5s
Google’s poor line 4s
0s20s
Method · The same tool, the same settings, the same day. This was the page where a customer decided whether to hire the retailer.
A sitewide score would have averaged the two and called the store slow-ish. Measured template by template, the problem has an address. Timed for a UK retailer with a catalogue of several thousand products, issued September 2026. The client is not named.