Conversion audit

Ranking is only half of it. A page that arrives at the top of the results and then fails to tell a visitor what to do next has cost you the same money as a page nobody found — you simply paid it further down the funnel.

Docket runs 9 checks on the pages that have to convert. Not one of them tells you whether your offer is good, and none of them can replace testing. What they do is catch the mechanical failures: the page that ranks for a promise it does not repeat, the contact form nobody can complete, the phone number that will not dial on the device most people are holding.

What Docket looks at

• Published email addresses that cannot receive mail cvr.dead_contact
• Form friction cvr.form_friction
• Funnel coverage cvr.funnel
• Title / headline consistency cvr.message_match
• Calls to action cvr.no_cta
• Pricing transparency cvr.pricing
• Social proof cvr.social_proof
• Phone links that will not dial cvr.unusable_phone
• Above-the-fold value proposition cvr.value_prop

Three findings from one real site

Auditing a US coffee retailer on 2026-08-14 produced three conversion findings, and they are a fair illustration of what this lane is and is not good for.

"2 key pages have no clear call to action", at HIGH. This is the mechanical kind and it is worth exactly what it says: two pages whose job is to move somebody forward offered nothing to click. That is checkable, and it is checkable without any opinion about the business.

"No social proof on any conversion page", at MEDIUM. Also mechanical — no reviews, testimonials, counts or badges appeared in the markup Docket read. Note the scope: in the markup Docket read. A site that loads its reviews from a widget after the page renders has social proof that this check cannot see, which is why it sits at MEDIUM and not higher.

"The site only serves buyers who are already ready", at LOW. This is the interesting one. Every page was built to sell to somebody who has already decided; nothing addressed a person still working out whether they want the thing at all. That is a real gap and it is also a judgement, which is why it is reported at LOW and phrased as an observation rather than an instruction. You may be running that strategy deliberately.

Message match is the one people fix first

The check that most often changes something is cvr.message_match: whether the promise in your title tag — the words somebody read in the search results and clicked — is repeated in the headline they land on.

When they differ, the visitor's first act on your page is to check whether they are in the right place. Some of them decide they are not. Nothing on the page is broken, the bounce looks like ordinary disinterest, and the cause is one sentence written by a different person at a different time from the one above it.

It is also the cheapest thing on this list to fix. You are editing one line.

The phone number that will not dial

Two checks here exist because of failures that are invisible on a desktop and total on a phone: a number printed as text with no tel: link, and an email address on a domain that cannot receive mail. Both look perfectly fine to whoever published them, because whoever published them never tried to use them from the device their customers use. The mail one is worth its own page — an address on a domain with no MX record bounces to the sender, so you never learn that anybody tried.

What this lane deliberately does not do

It does not score your copy, judge your design, or predict a conversion rate. It cannot tell you whether your price is right, whether your offer is compelling, or whether the person who left would have bought at a different headline. Those are answered by testing against real visitors, and any tool that claims to answer them from your HTML is selling you its own opinion with a number attached.

What it can do is make sure the mechanical things are not silently costing you the traffic you already earned.

Download Docket

Common questions

What is a conversion audit?

A check of the pages that have to turn a visitor into an enquiry or a sale, looking for mechanical failures: no call to action, a headline that does not match the title people clicked, a phone number that will not dial, contact details that cannot receive anything.

Can a tool tell me why my landing page does not convert?

Only partly, and honestly only the mechanical part. Whether the offer is compelling or the price is right can only be answered by testing against real visitors. A tool that gives you a conversion score from your HTML is reporting its own opinion with a number attached.

What is message match and why does it matter?

Whether the promise in your title tag — the words somebody read in the search results before clicking — is repeated in the headline they land on. When it is not, the visitor's first act is to work out whether they are in the right place, and some decide they are not. It is usually a one-line fix.

Does Docket see reviews loaded by JavaScript?

Not by default. The social proof check reports what appeared in the HTML it read, which is why it is reported at medium rather than higher — a site whose reviews arrive from a widget after render has social proof this check cannot see.