Scope: what this audit covers and what it does not
The scorecard rates one landing page, as a visitor from a specific traffic source sees it. It covers whether the page continues the promise of the ad, email or link that brought the visitor, how clearly the offer and headline state what is on offer, how the page is ordered, what proof supports the claims, how much effort the form or checkout asks for, how fast and usable the page is on a phone, whether people using assistive technology can complete it, and whether the conversion is recorded. Score each important traffic source separately if they carry different promises.
The scorecard is not a design guide. The building blocks of a page are explained in the anatomy of a landing page guide, form design in the landing page forms guide, and the tick-box pass before launch in the landing page checklist. Whole-site technical health, SEO and analytics set-up have their own scorecards in the audit library. To look beyond one page, an expert review of the conversion journey walks every step from the advert to the confirmation and logs each issue with a severity and a confidence rating instead of a score. The audit also does not judge price or product; if the offer is uncompetitive, the page can only present it clearly. When a low score leads to a rebuild rather than fixes, the replacement page is specified in a landing page brief, which turns each failed criterion into a requirement for the designer.
Before you score: access and evidence
Open the page as the visitor would: from the actual ad or email, on a mid-range phone and on a desktop, in a private window. Save screenshots of the ad and the page side by side. Pull the page's recent traffic, its conversion rate by device and source, and its form or checkout drop-off if the analytics set-up captures field or step events. The conversion rate calculator is useful for putting CVRShare of visits or clicks that convert. by segment on the same basis before comparing.
Gather any qualitative evidence available: session recordings, heatmaps, on-page survey answers or sales team feedback about what prospects ask. Run the page through a field-data speed report and an automated accessibility checker, keeping in mind that automated checkers catch only part of what matters. Finally, complete the form or checkout yourself once, end to end.
Criteria and scoring
Rate each row missing, partial or solid. Message match and tracking are gates.
| Criterion | What good looks like | Score guidance |
|---|---|---|
| Message match (gate) | Headline and first screen repeat the promise, offer and wording of the source | Missing: generic page / Partial: related but different promise / Solid: clear continuation |
| Offer clarity | A visitor can say what they get, for whom and what happens next within seconds | Missing: unclear offer / Partial: clear only after scrolling / Solid: immediate |
| Headline and subhead | Specific benefit, plain language, supports the offer | Missing: slogan or vague / Partial: specific but weak / Solid: specific and relevant |
| Primary call to action | One main action, visible early and repeated where the reader decides | Missing: hidden or competing actions / Partial: visible but vague / Solid: clear and consistent |
| Structure and hierarchy | Sections follow the questions a visitor has, in order; scannable | Missing: wall of content / Partial: some order / Solid: logical and scannable |
| Proof | Reviews, cases, logos or data relevant to this audience, close to the claim | Missing: none / Partial: generic or far from claims / Solid: specific and placed well |
| Risk reducers | Guarantees, pricing clarity, privacy note and contact options where needed | Missing: none / Partial: some / Solid: addresses likely doubts |
| Distractions | Navigation, exits and secondary offers kept to what helps | Missing: full site navigation and many exits / Partial: some trimmed / Solid: focused |
| Form friction | Only necessary fields; clear labels, errors and progress | Missing: long or confusing form / Partial: some extra fields / Solid: minimal and clear |
| Speed | Field data shows the page loads and responds quickly | Missing: slow on mobile / Partial: borderline / Solid: fast |
| Mobile usability | Content, form and action easy to use with a thumb; nothing cut off | Missing: broken on phone / Partial: awkward / Solid: built for mobile |
| Accessibility | Labels, contrast, focus order and keyboard use support everyone | Missing: form unusable without a mouse / Partial: some failures / Solid: no known barriers |
| Conversion tracking (gate) | The conversion event fires once per real conversion, with source data intact | Missing: not tracked / Partial: duplicated or loses source / Solid: verified |
For hierarchy, proof and form friction, the usability research collected in Nielsen Norman Group's articles is a sound reference point. For accessibility, score against the WCAG success criteria as a guide to what to check, not as a legal opinion. Speed is read through Core Web VitalsGoogle's user experience metrics. field data where the page has enough traffic to report it.
How to score and prioritise
Score the gates first. A missing message match means visitors arrive expecting something the page does not offer, and every other improvement works against that headwind; the message match guide explains how to rebuild the first screen. A missing tracking score means you cannot measure anything that follows, so it is fixed before any other change ships.
Then sort the remaining weak rows into two piles. Blockers are things that stop a willing visitor converting: a broken form on mobile, an inaccessible field, a call to action below a long section on a phone. Fix those directly. Everything else is a hypothesis: the evidence suggests it may help, but there is room to be wrong. Rank hypotheses by the traffic the page receives, the size of the drop-off they address and how much effort they need, and put the top ones in the test backlog as candidates for an A/B TestControlled comparison of two variants..
Reading the result: what to do with each band
Red means a gate scored missing. Stop sending additional paid traffic to the page until the promise matches the source and the conversion is tracked. Both fixes are usually quick, and both change how every other row should be read.
Amber means the gates hold but several rows are partial or missing. Fix the blockers within days, then work through the hypothesis list as tests or as considered changes if traffic is too low to test. Rescore after each change has had enough traffic to judge.
Green means the page is in good shape. Use it as the template for new pages and keep testing bigger ideas, such as a different offer or format, rather than small tweaks.
From audit to action plan
Turn each blocker into a ticket with the finding, the screenshot and the fix. Turn each hypothesis into a test card: the observation, the change, the expected effect on behaviour and the metric that will judge it. Share the band and the top blockers with whoever owns the traffic source, because a mismatch is often fixed as easily in the ad as on the page.
A boiler-service lead page
A heating engineer runs search ads promising a "fixed-price boiler service from £79". The landing page is the general services page, opening with a company history. Message match scores missing and the band is red. Mobile scores partial: the quote form sits below six screens of text on a phone. Form friction scores partial with eleven fields, including the boiler's serial number. The fixes are immediate: a dedicated page that opens with the fixed price and booking form, cut to five fields. Proof and risk reducers go into the test backlog. Over the next month the page's conversion rate moves from 2.1% to 5.4% on similar traffic, and the first test compares two placements of customer reviews.
Common mistakes
- Scoring the page without the ad beside it. Message match is a comparison. Without the source in view, the gate cannot be scored.
- Testing blockers. A broken mobile form does not need an experiment. Fix it and test the genuinely uncertain changes.
- Auditing on a desktop only. Much paid traffic arrives on phones. Score the device the visitors actually use.
- Relying on automated accessibility scans. They miss keyboard traps and confusing labels. Use the keyboard yourself.
- Mixing traffic sources in one score. An email visitor and a search visitor carry different promises. Score each important source separately.
- Letting the team score alone. People who built the page no longer see its gaps. Add a scorer who has never seen it before.
