A Real CRO Audit Example: What's Actually in the Document

A Real CRO Audit Example: What's Actually in the Document

Most people buying a conversion audit for the first time have no idea what shows up in their inbox. A checklist? A Loom video? A hundred-page deck nobody reads?

So here's a real one, published with the client's permission. In August 2026 Mike Young bought the $99 audit for GIS Case Analysis, his independent GIS consulting practice. He maps and analyzes spatial evidence for criminal defense attorneys: location, timing, route, and witness visibility. Three-page site. He came to me because his bounce rate was high and he couldn't tell why.

Every finding and structural choice below is exactly what went in his document. I've left out his contact details and a couple of small typos, which are his to fix rather than mine to publish.

What I look at before writing anything

The analysis runs against the live site, not a description of it. Full-page screenshot captures at desktop width and at 390px for mobile, page source read directly, and performance measured through PageSpeed Insights. Every page, both widths.

That matters more than it sounds. Plenty of audits get written off a quick look and a summary. If you're going to tell someone their site has no contact form, you should have checked the page source and confirmed there are zero form elements on it, which is what I did here. Findings that can be verified should be verified.

1. Executive summary, with the issues already ranked

The first thing in the document is the whole answer. Not background, not methodology. Six paragraphs of what's happening on this site and why, then a numbered list of the priority issues in order of conversion impact.

Mike's site got eight. Number one was that there was no working way to contact him. Number eight was a set of small credibility details: a couple of typos in buttons, a default favicon, and a shopping cart icon sitting in the nav of a consulting site that doesn't sell anything online.

The order is the product. Finding eight problems is easy. Knowing that the first one is worth more than the other seven combined is the part you're paying for, and it's the part a scanning tool cannot do.

2. What's working, and should not be touched

This section exists because audits create anxiety. Hand someone a list of everything wrong with their site and a fair number of them respond by changing everything, including the parts that were carrying the business.

On Mike's site the demonstration exhibits were genuinely excellent. He publishes simulated cases to work around case confidentiality, properly cartographed and legended, with timestamped events. One puts an accused-stated route and the prosecution's route side by side with a cell sector overlay, which is exactly what a defense attorney needs to see. That was the strongest asset on the site and every recommendation I made was about getting more people to look at it.

His niche was sharp, his pricing was transparent in a category that hides rates, and his technical SEO scored 100 out of 100 with accessibility at 97. All of that got written down explicitly as do not change.

An audit that only lists problems is a worse product. You need to know where the floor is.

3. Who the site is actually talking to

Before judging any page you have to know who's landing on it and what they're deciding. Mike's traffic was almost entirely cold outbound, so his visitor was a defense attorney checking whether a cold email was worth answering, under time pressure, needing someone who would survive cross-examination.

That single fact reordered the entire audit. A visitor arriving from a search has already decided you're real. A visitor arriving from cold outreach has not, and needs proof of a human being before anything else on the page gets read. Same site, completely different diagnosis depending on where the traffic comes from.

4. A persuasion scorecard

Seven rows, each marked present, partial, or absent, with a sentence on what the evidence actually is: authority, social proof, urgency, commitment, liking, scarcity, reciprocity.

Three came back absent on Mike's site. No credentials, certifications, years of experience, testimony history, or photograph anywhere. No testimonials or references of any kind. And no published turnaround times, which for someone selling expert work against a trial date quietly disqualifies him before a conversation starts.

The scorecard is useful because it's structured. It catches the thing that isn't there, which is much harder to notice than the thing that's wrong.

5. The journey table

This is the center of the document. Seven steps from landing to contact. For each one: what the visitor is trying to do, what the site currently does, and a verdict of works, partial, or breaks.

Laid out that way, the shape of the problem becomes obvious. Mike's site handled the hard parts. It proved the work was good and it published honest pricing, which most of his competitors won't do. Then it broke at step four, where the attorney asks whether this person survives cross-examination, and broke completely at step seven, where they try to get in touch.

The line I ended that section on was the summary of the whole engagement: the journey succeeds at the two hardest steps and fails at the three cheapest. Credentials, turnaround times, and a contact form are an afternoon of work between them.

That's the sentence he needed. Not "your site has problems." Your site is closer than you think, and the gap is smaller than you fear.

6. Competitive landscape and wireframes

A short read on who else the visitor is evaluating and what those sites do differently, then three wireframe plates: current homepage structure against recommended, the recommended layout in detail, and the mobile first screen before and after.

Wireframes go in because written recommendations get misread. "Move the credential line up" means five different things to five people. A drawing means one.

7. Eight priority fixes, each with an impact rating

The bulk of the document. Each fix is structured the same way: what's wrong now, why it matters in money or lost inquiries, and the recommended approach broken into what to do today, this week, and after.

Fix one on Mike's site was the contact path, tagged critical. The footer link on all three pages was malformed, with his email address, phone number and URL concatenated into a single string. It displayed correctly, so it looked fine. Clicking it opened a blank message to an invalid recipient. Anyone who had tried to reach him that way never got through, and he had no way of knowing.

I verified it in the page source: zero form elements across all three pages. No contact page either, and on mobile the only contact button collapsed into the hamburger menu, leaving no visible way to reach him without scrolling roughly five screens to a broken link.

The fix was fifteen minutes of splitting one link into two, then an hour or two adding a real form with the four qualifying fields that would let him scope an inquiry without a round trip.

That's the whole game. The most expensive problem on the site was also one of the cheapest to fix, and he had been living with it while wondering why his bounce rate was high.

8. Performance, a checklist, and the recommended structure

PageSpeed findings with the specific culprits, since mobile was loading its main content in 7.3 seconds. Then a checklist for tracking the work. Then a closing section with a recommended page structure: eight positions, each with the question it answers and why it sits where it does.

The structure section explains its own reasoning, including where it departs from the usual advice. Proof normally goes before price, so the number gets judged against work someone has already valued. For Mike I recommended a single price anchor early anyway, because his rates were unusually low for expert work and functioned as a positive surprise worth spending up front. Full rate card still came after the exhibits.

Rules of thumb are a starting point. The audit is where you find out which ones don't apply to you.

The part I'd want you to take away

Mike asked me about his bounce rate. He assumed the problem was his messaging.

Messaging was the one thing already working. Someone landing on that site knew within a second what it was and who it was for. What was missing was any signal that a real qualified person was behind it, and any working way to act once convinced.

This is the normal outcome. People are usually wrong about their own bottleneck, and not because they aren't smart. They're too close to it. You cannot see your own site the way a stranger does, and the thing you're anxious about is rarely the thing costing you money.

Four critical fixes on his site came to about ninety minutes of work.

Here's what he said afterward:

"The report I received was loaded with insight and improvements. I dropped it into ChatGPT and it was able to walk me through making all the changes myself."

Mike Young, GIS Case Analysis

That second sentence is the one I care about. The document is specific enough that a non-technical owner can hand it to an AI assistant and execute it without hiring anyone. That's the intended outcome, not an accident.

What you'd get

Same document, same structure, for your site. Full conversion analysis, every issue written up with what's wrong and what to do instead, everything ranked so the first thing you read is the first thing to fix, delivered in one business day for $99.

If you're still deciding whether you need this, the free version of the checklist is here and it will find real problems on its own. If you want to know how this compares to a five-figure agency engagement, that comparison is here.

Otherwise, the audit is here.