The landing page audit we run before touching a single ad

When an account is underperforming, the reflex is to change the media. Different keywords, different audiences, different creative. It feels like progress because the levers are obvious and the platform gives you a new number the next morning.

But if the destination converts at one per cent when it should convert at three, every media improvement is being taxed by two thirds before it reaches the pipeline. Fixing the page is usually cheaper, faster and more durable than fixing the traffic.

What we look at, in order

1. Message match

Read the ad, then read the first screen of the page. If the promise in the ad is not restated in the headline in recognisably the same words, you are paying to create a doubt that the visitor then has to resolve. A surprising share of underperforming accounts send six different ad promises to one generic page.

2. The first screen on a phone

Not the desktop hero — the phone, at the width most of your traffic actually uses. Is there a headline that states the outcome, one sentence saying who it is for, and a visible action? Carousels, autoplay video and a cookie banner covering the call to action are all straightforwardly expensive.

3. Friction in the enquiry flow

Count the required fields. Then ask, per field, whether sales genuinely uses it to qualify or whether it is there because a form builder defaulted to it. Every field you remove above the strictly necessary raises completion, and phone number, company size and “how did you hear about us” are the three most commonly unnecessary.

4. Proof placement

Proof answers a doubt, so it belongs where the doubt occurs — beside the form, not stacked in a testimonial band at the bottom that most visitors never reach. One specific, attributable proof point next to the action outperforms six generic ones further down.

5. Speed, honestly measured

Field data, not a lab score. Largest Contentful Paint above two and a half seconds on mobile is losing you conversions right now, and the cause is nearly always an oversized hero image, a font that blocks render, or a stack of third-party tags loading before the content.

6. What the session recordings show

Watch twenty sessions that reached the form and did not complete it. This is unglamorous and it is consistently the highest-yield hour in the entire audit. People hesitate at a specific field, hunt for a price, or rage-click something that is not a link. You cannot infer any of that from a funnel chart.

Then, and only then, the experiment backlog

The audit produces a list. The list gets scored on expected impact, confidence and effort, and the top items become tests with a stated hypothesis and a pre-agreed success metric. Everything else waits.

A test without a hypothesis is not an experiment. It is a redesign with extra reporting.

Two practical notes. First, respect statistical reality: if your page receives 400 visits a month, you cannot detect a five per cent lift, so change something big or change it without testing and monitor carefully. Second, keep a decision log of every test and its outcome. In eighteen months it will be the most valuable marketing asset you own, because it tells you what does not work for your audience — which is knowledge no agency can sell you.