Most landing page audits don't move the number. They produce a 40-slide deck with screenshots, a few highlighted headlines, and a vague list of "opportunities". The brand spends a fortnight implementing the cheaper recommendations, sees no measurable lift, and concludes that CRO is a soft science.
It isn't. The audits are just bad.
A landing page audit is a diagnostic instrument. If yours doesn't tell you what's leaking, how much it's leaking, and what to fix first, it's not an audit. It's an opinion in a Notion doc.
What a landing page audit framework actually needs to do
Three jobs. That's it.
First, the framework has to surface every meaningful friction layer on the page, not just the headline and the button colour. Second, it has to score what it finds, so the team can prioritise. Third, it has to produce a test backlog, not a wishlist. If a recommendation can't be turned into a brief, it's noise.
Most frameworks fail on the second job. They list 30 "opportunities" with no weighting, and the team ends up shipping the easiest ones rather than the highest-leverage ones. That's how brands spend six months testing button copy while the hero is promising the wrong outcome to the wrong audience.
The framework below is what the PMD team uses on cold-traffic landing pages for £5M to £200M Shopify brands. It's opinionated, it has a strict order, and every layer maps to a measurable category of leak. Run it cleanly and you'll have a scored test backlog by lunchtime.
The PMD eight-layer audit, in order
The order matters. Each layer feeds the next. If layer one is broken, fixing layer six is rearranging deckchairs on a page that was already underwater the moment it loaded.
Layer 1: Promise-match
Open the ad that's driving the traffic. Read it. Now open the landing page. Is the page delivering on the specific promise the ad made: same outcome, same mechanism, same audience? Or is the ad selling "finally sleep through the night" while the landing page leads with "shop our award-winning bedding collection"?
This is the single most common leak we find. Cold traffic doesn't recover from a broken promise-match. The visitor's brain registers the mismatch in roughly 1.5 seconds and they're gone. No amount of clever copy lower on the page can save it.
Layer 2: Hero and above-the-fold load
The hero has to do three things in three seconds: state the outcome the visitor wants, prove you can deliver it, and tell them what to do next. If a stranger can't tell you what your product does, who it's for, and why they should care from the hero alone, the page is leaking before the visitor scrolls.
Run the five-second test the way we do internally. Screenshot the hero, show it to someone outside the brand for five seconds, take it away, and ask them what the product is and who it's for. If they can't answer, the hero needs rebuilding, not optimising.
Layer 3: Belief stack
Every objection a cold visitor holds (will it work for me, is this real, what's the catch, why this brand, what if it doesn't work) has to be addressed in roughly the order it arises in their head. The belief stack is the sequence of proof points the page uses to close those objections.
Audit it by listing every objection the avatar holds, then walking the page top to bottom and noting which sections close which objections. Gaps in the stack are gaps in conversion. Most pages we audit close objections one and two beautifully, then forget objection four exists entirely.
Layer 4: Visual hierarchy and scan path
Cold visitors don't read. They scan. Defocus your eyes on the page and note what your eye lands on, in what order. If the scan path doesn't tell the story, the copy never gets read.
The worst offenders are pages where the eye lands on a stock photo, then a button, then jumps to a testimonial, then back up to a sub-header. There's no narrative. The visitor leaves because nothing pulled them down the page in a coherent line.
Layer 5: Friction inventory
List every action the page asks the visitor to take or evaluate, in order. Every dropdown, every variant selector, every "choose your bundle", every modal, every discount code field. Each one is a place to lose them. Some of that friction is necessary; most of it is leftover from a designer's instinct to make the page feel premium.
This layer is where most audits stop being theoretical and start producing testable hypotheses. A typical mid-market Shopify product page carries two to four units of unnecessary friction. Each one is a test.
CRO Obsessed
Stop guessing which layer is leaking. Start auditing the one that actually moves your profit.
PMD is a full-funnel CRO and profit-optimisation agency. We help subscription and high-LTV Shopify brands fix the funnel that sits between their ad spend and their profit.
Book a 30-min call with Paddy →Layer 6: Mobile-first pressure test
Audit on mobile first. Always. Most teams design on desktop and check mobile last; cold traffic from Meta and TikTok is comfortably over 80% mobile, so the desktop view is the secondary surface, not the primary one. Open the page on a real device, not a Chrome emulator, and ask: does the hero work on a 5.8-inch screen? Is the CTA reachable with a thumb? Is the price visible without scrolling past three blocks of body copy?
Pair this layer with our mobile checkout teardown for the full mobile profit picture. The LP and the checkout are one continuous mobile surface, and treating them as separate problems is how brands lose 20% of conversions to thumb-fight friction nobody on the team has actually felt.
Layer 7: Conversion path and offer clarity
Trace the path from page-view to confirmed order. How many clicks, decisions, and form fields between the visitor saying "yes" and the order processing? Every step is attrition. The cleanest converting pages have a single CTA, a single offer, and a default state that doesn't require the visitor to choose anything.
The other half of this layer is offer clarity. A visitor should be able to answer four questions on the page: what am I getting, for how much, when, and what happens if I don't like it? If those answers aren't visible without scrolling or hunting, you have an offer problem dressed up as a conversion problem.
Layer 8: Speed, stability, and tracking trust
The unglamorous one. If the page takes 4.2 seconds to render on a mobile 4G connection, you've lost a meaningful share of cold visitors before they see the hero. If your tracking is double-firing or under-attributing, every test result will be wrong and you'll iterate in the wrong direction for months.
Audit Largest Contentful Paint on a throttled mobile network. Audit your GA4 and Shopify analytics for parity. If the two disagree by more than ten percent, fix tracking before testing anything else; otherwise the audit's recommendations will be measured against a broken ruler.
How to score and prioritise what the audit finds
Every leak you log gets three scores from one to five: severity (how much of the conversion gap does this explain), confidence (how sure are we this is the cause), and effort (how hard is the fix and the test). Multiply severity by confidence, divide by effort. Sort descending. That's your test backlog.
This isn't ICE or PIE with the labels filed off. The difference is the discipline: every leak has to map back to one of the eight layers, which forces the team to think structurally rather than reaching for the obvious.
Most audits we inherit from previous agencies look like this: "23 quick wins", almost all of which are layer 2 (hero copy) and layer 4 (visual hierarchy), because those are the easiest leaks to spot. The actual leaks were on layer 1 (promise-match) and layer 5 (friction inventory). The team had been optimising the wrong surface for nine months.
"A scored backlog forces the team to think structurally, rather than reaching for whatever's easiest to spot."
A real audit: how this framework caught a £25,535 monthly win
One of the public PMD case studies, the £25,535 cart-threshold split test, came directly out of a layer 5 finding. The audit flagged that the cart drawer was asking visitors to evaluate too many things at once: free-shipping threshold, bundle upgrade, subscription option, and a discount code field, all stacked above the checkout button.
The hypothesis that fell out of the audit wasn't "redesign the cart". It was much more specific: "restructure the free-shipping threshold so it does one job, which is nudging AOV." One split test, one win, £25,535 of additional monthly revenue, sustained.
That's the value of a structured audit. The recommendation isn't "try things until something works". It's "the leak is on layer 5, here's the specific friction, here's the test that closes it." The same discipline produced the rebuild that helped Rory's Travel Club grow over 650% after its Shopify migration, where the audit's biggest finding was upstream of the page entirely: the theme itself was the friction.
The counter-view: when a landing page audit is the wrong tool
Not every problem is a CRO problem. We've audited pages converting at 4.8% on cold traffic where the brand still wasn't profitable. The page was fine. The offer was wrong, the LTV was thin, and the CPMs were eating the margin.
Three situations where an audit isn't the answer.
First, when the offer itself doesn't have product-market fit. A great audit on a page selling something nobody wants will produce a marginally less-bad conversion rate, and a team congratulating itself for a 6% lift on a fundamentally unprofitable funnel. The real fix is upstream.
Second, when the traffic source is the problem. If your Meta creative is selling a different product than the page delivers, the audit will keep telling you the page is broken. The page might be fine. The creative is misleading and the bounce is rational.
Third, when the brand needs a rebuild, not an audit. If the theme is a tangle of conflicting apps and unaddressed tech debt, the audit will produce 40 recommendations the dev team can't implement cleanly. Sometimes you commission a proper Shopify build first, then audit the new surface.
Your Monday-morning landing page audit checklist
Block 90 minutes on Monday. Open the page on your phone first, your laptop second. Walk through the eight layers in order. Score each leak as you go: severity, confidence, effort. Don't stop to fix anything. The audit comes first; the test backlog comes out of the audit, not alongside it.
At each layer, the question to ask is the same: what is the visitor's brain doing right now, and is the page helping it or fighting it?
If you do this honestly, you'll have a prioritised test backlog by lunch. Most audits we run produce 18 to 25 logged leaks, of which five to eight are high-priority tests that ship within the first 30 days. That's a real backlog, not a wishlist.
If you want a wider audit that covers the whole funnel rather than a single page, the broader Shopify CRO audit framework covers store-wide diagnostics. If you're auditing the upstream creative-to-page handoff, the cold-traffic landing page playbook and the notes on writing a winning advertorial are the companion reads.
For ongoing work, our profit-optimisation engagement wraps audits, testing and shipped fixes into one retained programme. The PMD CRO learning hub is the open library if you'd rather build the capability in-house. Either route, if you want a second pair of eyes on a single live page before you commit to anything bigger, book a 30-minute call with Paddy McLarnon and we'll teardown one of your pages live on the call.
FAQs
How long should a landing page audit actually take?
A proper page-level audit using this framework takes 60 to 90 minutes if you know the brand and the offer. The output is a scored backlog of 18 to 25 leaks, with five to eight high-priority tests. Anything taking longer is usually the auditor padding the deliverable.
What's the difference between a landing page audit and a full CRO audit?
A landing page audit is page-level. A CRO audit is funnel-level. It covers the LP, the PDP, the cart, the checkout, the post-purchase flow, and the retention loop. Use an LP audit when you already know the leak is on a specific page. Use a full CRO audit when you don't know where the leak is.
Can the framework be run on a Shopify product page, not just a dedicated LP?
Yes. The framework works on any page that has to convert traffic: PDPs, advertorials, collection pages, lead-gen pages. The layers are the same; the specific friction patterns shift. Shopify PDPs tend to leak heaviest on layers 4 (visual hierarchy) and 5 (friction inventory), because the default theme behaviour stacks variant selectors, swatches, and apps in ways that fight cold visitors.
Should the audit be done by an in-house team or by an external agency?
In-house teams know the brand and the data. External teams see the page without the cognitive bias of having built it. The strongest setup is in-house running the framework monthly, with an external audit once a quarter to challenge assumptions and surface what the team has stopped seeing. Either way, the framework is the same.
What tools do I actually need to run this audit?
Almost none. A real phone for mobile testing, a screen recording tool, your analytics (GA4 plus Shopify), a session replay tool such as Microsoft Clarity, Hotjar, or Lucky Orange, and your ad account to pull the upstream creative. The most useful tool on the list is a stopwatch for the five-second hero test.
How often should we re-audit a page that's already converting well?
Quarterly at minimum, and any time the upstream creative changes meaningfully. A page that converted well in Q1 against one set of ads can quietly leak ten or fifteen percent of its conversion rate by Q3 because the creative shifted and the promise-match drifted with it. Re-running layer 1 alone is a 20-minute exercise that frequently saves the quarter.
Full-funnel CRO. Profit obsessed.
Want this on your store?
We help subscription and high-LTV Shopify brands turn cold traffic into post-click profit. Strategy, copy, design, development, and CRO under one roof.