Landing Page Conversion Review

A landing page has one job: move a specific visitor from a specific source toward one action. Most reviews fail because they critique the page in a vacuum. This skill anchors the teardown in traffic context, walks the page the way a skeptical first-time visitor does, and outputs scored findings ranked by expected impact — not by how easy they were to notice.

When to use this skill

A page converts below expectation and nobody can say why
Pre-launch review of a new landing page before traffic gets spent on it
An ad or email drives clicks that never become signups, and the page is the suspect
Periodic teardown of the highest-traffic pages, quarterly or after a repositioning
Skip it for a homepage serving six audiences; review that one audience at a time

Workflow

Establish context before judging anything. Where does the traffic come from, what did that ad or link promise, who is the visitor, and what single action counts as conversion? A perfect page for the wrong source is a broken system; message mismatch is its own finding class.
Run the five-second impression. From the top of the page only: what is this, who is it for, why should I care, what do they want me to do? Any unanswered question is a headline problem, and headline problems dominate everything below them.
Audit the argument section by section. The page should make one cumulative case: claim, how it works, evidence, objection handling, action. Note where the argument skips a step — claim straight to pricing — or repeats itself without adding new proof.
Inventory the proof. Count the specific evidence: named outcomes, numbers, recognizable customer contexts, demonstrations. Generic praise quotes count as zero. Score proof against the size of the claim; a bold claim over thin proof reads as a lie, not as confidence.
Walk the friction path. Every field, click, decision, and unanswered doubt between arrival and conversion. Classify each as necessary, reducible, or removable — then check load behavior and the mobile rendering, where a third of the friction usually lives.
Check the ask. One primary action, visible without scrolling, repeated after each major section, labeled with what happens next ("See your report" beats "Submit"). Two equal-weight actions split the page's one job in half.
Score and rank. Apply the rubric below, then order findings by expected conversion impact and tag each with effort: copy tweak, layout change, or structural rebuild.

Scoring rubric

Score each dimension one to five: message match with source, headline clarity, argument completeness, proof density, friction (five means minimal), and call-to-action focus. Report the two lowest scores as the headline findings — fixing a two beats polishing a four.

Output format

A one-line verdict, the six rubric scores, then findings in three groups: fix first (high impact, low effort), fix next (high impact, high effort), and fine for now. Each finding states what was observed, why it costs conversions, and the specific proposed change — with rewritten copy included whenever copy is the fix.

Quality bar

Every finding cites something observable on the page; "feels cluttered" is taste, not a finding
Proposed fixes are concrete enough to implement without a follow-up meeting
The review names what the page does well; teams ignore teardowns that read as demolition
No recommendation contradicts the traffic context established in step one
The verdict answers the client's real question: is this page the problem, or is the offer?

Worked example: one finding, written properly

Observation: the primary button says "Submit" and first appears 2.5 screens down; the ad that drives 80 percent of traffic promises "see your compliance score in 60 seconds." Why it costs conversions: visitors arriving on that promise cannot see any route to the score without scrolling past two sections that do not mention it. Proposed change: move the form above the fold, retitle the button "Get my compliance score", and echo the ad's 60-second promise beside it. Impact: high. Effort: layout change. That is the shape every finding should take — observed, costed, prescribed.