Most landing page audits start in the wrong place. Someone opens a browser, scrolls the page once, mutters something about the hero image, and calls it done. What they've produced is an opinion, not an audit. A real audit works from a fixed set of criteria applied consistently, so you're not just catching the things that happen to annoy you on a given Tuesday.
Ferguson's audit data across 149 pages makes the case for systematic checking pretty bluntly: 82% of pages fail on objection coverage, meaning the vast majority of landing pages never address the doubts a visitor is already carrying when they arrive. That's not a design problem or a copy problem in isolation - it's a structural gap that only shows up when you're looking for it deliberately. The checklist below is built around the failure patterns that actually show up at scale, not the ones that make for tidy blog posts.
The 5-second test: what your page communicates before anyone reads it
Across the pages audited so far, 41% fail the 5-second test - which means a visitor landing on those pages cannot tell, within five seconds, what the product does and who it's for. That's a brutal number when you consider that most paid traffic decisions are made in roughly that window. The 5-second test isn't about whether your headline is clever. It's about whether a stranger, with no context, can orient themselves fast enough to stay.
When auditing for this, cover three things specifically. First, read only the headline and subheadline - does the product category come through without the supporting copy? Second, look at the hero image or illustration: does it reinforce the message or is it decorative? Third, check whether the primary CTA label adds any information, or whether it just says 'Get started' into the void. A CTA that reads 'Start your free trial' tells you something. One that reads 'Let's go' tells you nothing.
CTA structure: the single-primary-CTA problem most pages ignore
56% of pages in Ferguson's dataset fail the single primary CTA check. This one surprises people, because the instinct when you're building a page is to give visitors options - a demo button, a trial button, a 'learn more' link, maybe a chat widget. The logic feels generous. In practice, it fragments attention and forces a micro-decision at exactly the moment you want momentum.
The audit question here is not 'do we have one button?' but 'is there one action that clearly dominates the page?' Secondary links are fine. A ghost button next to a filled one is fine. What kills conversion is two equally weighted CTAs competing for the same visual space, leaving the visitor to figure out which one applies to them. If your page has a 'Book a demo' button and a 'Start free trial' button at the same size, same weight, same position in the hierarchy, you've created a decision where there should be a direction.
Check every section of the page, not just the hero. CTAs accumulate as pages get longer, and what starts as a clean structure often becomes cluttered by the time someone adds a mid-page module six months after launch.
Objection coverage and specific claims: the two checks most pages fail
Objection coverage and specific claims describe the same underlying failure from two angles. Objection coverage - failing on 82% of pages - is about whether the page anticipates and addresses the doubts a visitor brings with them. Specific claims, failing on 55% of pages, is about whether the page gives visitors anything concrete to believe. Together, they describe a page that makes assertions without evidence and never acknowledges that the visitor might be sceptical.
For objection coverage, the audit process is to list the three to five most common reasons a qualified visitor would not convert, then check whether the page addresses each one. Common objections for SaaS pages include: 'This will take too long to set up', 'I'm not sure it works for my use case', 'I don't know if I can trust this company', and 'I don't know what happens after I sign up'. Most pages address none of these directly. Some gesture at them with a FAQ section buried below the fold. Very few build the answers into the main copy where the objection is most likely to surface.
Specific claims is a simpler check but requires honesty. Go through every benefit statement on the page and ask: is this verifiable, or is it just an adjective? 'Powerful reporting' is an adjective. 'Reports that update in real time, with export to CSV in one click' is a claim. 'Saves you time' is an adjective. 'Customers report saving an average of four hours per week on manual reconciliation' is a claim - and yes, it requires you to have actually collected that data, which is part of the point. Vague benefit language is often a symptom of not knowing your customer well enough to be specific, and visitors sense that.
Social proof: not just whether you have it, but where and what kind
Nearly half the pages in Ferguson's current dataset - 46% - fail the social proof check. Some of those pages have no testimonials at all. But a meaningful portion have testimonials that don't do the job: quotes that are enthusiastic but vague ('Great product, highly recommend!'), logos without any context about what those companies actually use the product for, or star ratings with no accompanying text.
The audit criteria for social proof should cover four things. Whether proof exists at all. Whether it's specific enough to be credible - a testimonial that names a concrete outcome is worth ten that describe feelings. Whether it appears near the points of friction on the page, not just in a dedicated section. And whether the sources are identifiable: a name, a job title, a company. Anonymous quotes are almost worthless because visitors can't assess whether the person is like them.
Also check whether the proof matches the audience the page is targeting. A page aimed at enterprise buyers that shows testimonials from solo founders is actively unhelpful. The visitor's implicit question is 'does this work for someone like me?' and the wrong proof answers 'no'.
Post-CTA clarity: what happens after someone clicks
64% of pages fail post-CTA clarity - the second highest failure rate in the dataset. This is the check that asks: does the visitor know what will happen when they click the button? Not in a legal sense, but in a practical one. Will they be asked for a credit card? Will they get an email? Will they be taken to a form, a calendar, a dashboard? The uncertainty around what comes next is a genuine conversion barrier, and it's almost free to fix.
The audit here is straightforward. Read the CTA label and the surrounding copy. If you removed everything except those words, would a visitor know what the next thirty seconds of their life looks like? 'Start free trial - no credit card required' passes. 'Book a demo - pick a time that works for you' passes. 'Get started' does not pass. Neither does 'Request access' with no indication of what access means or how long it takes.
It costs nothing to add three words of reassurance beneath a button, and the failure rate suggests almost nobody does it. A three-word addition beneath a button is the lowest-effort fix with the highest reach in the dataset.
Page load speed and mobile: the technical checks that quietly drain conversion
41% of pages in Ferguson's audit data fail the page load speed check. Load speed is one of those areas where people nod along and then don't actually measure it, partly because the tooling feels technical and partly because the page loads fine on their MacBook on a fast connection. The audit discipline here is to test on a throttled connection and on a mid-range mobile device, not on the hardware you use every day.
For a checklist audit, the load speed check should flag: images that haven't been compressed or converted to modern formats, render-blocking scripts in the page head, and third-party embeds - chat widgets, video players, analytics tags - that load synchronously. You don't need to fix all of these in one pass. But you do need to know they're there, because a page that takes four seconds to show anything above the fold is losing visitors before they've seen a single word of copy.
Mobile deserves its own line in the checklist rather than being treated as a subset of load speed. Check that the primary CTA is reachable without scrolling on a standard phone screen. Check that tap targets are large enough to hit without zooming. Check that any social proof or trust signals visible on desktop haven't been hidden or collapsed on mobile in a way that removes them from the experience entirely. These are different failure modes from desktop, and they need to be checked separately.
Running the checklist without it becoming a bureaucratic exercise
The temptation after running these checks is to hand a long findings document to someone else — which is where the work usually stops. A findings document without a sequenced action list typically produces one meeting and no changes.
A more useful structure: after completing the checklist, identify the single failure that is most likely to be costing conversion right now, based on where it sits in the visitor journey. A failed 5-second test affects every visitor. A failed post-CTA clarity check affects only visitors who reach the CTA. Both matter, but they don't matter equally — so prioritise the 5-second test failure over the post-CTA wording, then sequence the rest by reach. The diagnosis framework in our earlier piece covers how to sequence fixes when you have multiple failures competing for attention.
If you're running audits for clients or across multiple pages, the Ferguson benchmark data gives you a useful reference point for what failure rates look like across the broader population - so you can tell a client whether their objection coverage problem is typical or unusually severe.
The checklist items above aren't exhaustive - there are legitimate audit criteria around visual hierarchy, headline specificity, and pricing page structure that deserve their own treatment. But in Ferguson's data, objection coverage is the check that pages most consistently fail on a first audit pass — and the one most likely to still be unaddressed after the findings have been reviewed.