Ask any AI to audit your landing page. It will tell you to add social proof.
That sentence is true of every page ever built, which is what makes it worthless on yours. Generic CRO advice is survivorship bias: what worked in someone else’s context, abstracted until every detail that made it true is gone. cro-audit scores all 8 dimensions or fails, and a linter rejects any finding that does not name a specific element on your page and cite a real benchmark with a source.
For founders and operators with a page that gets traffic and does not convert — who have already read the checklists and want to know what is wrong with this page.
opvs-skills install @di-atomic/cro-auditThen say: “audit my landing page” or “why isn’t my pricing page converting?”
Why a tidy list of best practices is not an audit
1. Add social proof to build trust 2. Make your CTA stand out 3. Move the CTA above the fold 4. Reduce form friction 5. Improve page speed (true of every page ever made, unfalsifiable about yours, and nothing to disagree with)
You cannot argue with any of it, so you cannot act on any of it. Worse, item 4 is how a founder got 340% more signups and 60% less revenue.
FINDING #1 social_proof HIGH
element: hero, above the fold
fix: testimonials block OR
sys-proof-recent-sales-toast
benchmark: 15-25% named+photo+outcome
vs 2-5% generic unattributed
source: Digital Applied, 2026
effort: low → ICE rank 1
Names the element, names the fix, carries the range, cites who measured it. You can disagree with this one, which is the entire point.
A finding you cannot argue with is a finding that says nothing.
What you actually get
All 8 dimensions, every time
Value-Prop, CTA, Social Proof, Urgency, Trust & Security, Form Friction, Mobile, Page Speed. verify-audit-coverage.mjs demands 8 of 8. If a dimension cannot be assessed it has to say so and say why — silence is not a pass.
The platitude is a build failure
lint-no-best-practice-platitudes.mjs fails the run on any generic recommendation, and grade-evidence-density.mjs requires at least 80% of findings to name a page element and cite a benchmark. A rule the agent can talk itself out of is not a rule.
No dark patterns, no legal exposure
Fake urgency, fake scarcity and invented social proof get flagged as risks to remove, never recommended. On REACH, CLP, SDS, biocide, health and finance pages it will not add an unqualified outcome claim to lift conversion — existing absolutes are flagged as exposure.
Two findings that broke the checklist for me
One page, start to ranked list
> audit my pricing page
route → recipes/audit-pricing-page.md
score 8 dimensions ........ 8/8 assessed
verify-audit-coverage ..... PASS 8/8
lint-no-best-practice ..... PASS 0 platitudes
grade-evidence-density .... PASS 91% cited
TOP_LEAKS (ranked by severity x lift / effort)
1 social_proof HIGH hero has no proof above fold
15-25% vs 2-5% (Digital Applied 2026)
2 form_friction HIGH 8 fields on the trial form
16 → 8 fields lifted 120% (Baymard)
3 urgency RISK countdown timer resets on reload
→ remove. this is a dark pattern.
> finding #2 recommends removing friction
paired warning: name the downstream metric.
measure signups AND trial-to-paid, not just signups.
alternatives: $1 paid trial / money-back / freemium-with-limits
Finding 3 is a risk to remove, not a win to keep. And the paired warning on finding 2 exists because a single-metric win hid a 60% revenue loss for a real founder.
Where it stops
sys-* components, copy-engine for the words, section-designer for the layout. It also will not promise you a lift. Every finding is a hypothesis with a direction and a cited range attached, because on your traffic it might be the one that goes the other way.I built this by dogfooding the same stack I run for clients.
cro-audit is one skill in the system behind Di-Atomic — the marketing & compliance agency that runs cognitoAI, SpiderIQ and OPVS. If you want the pages themselves diagnosed, fixed and tested rather than a list of best practices — in any of our seven languages, including regulated categories like REACH and CLP where a conversion claim is a legal artifact — that’s the day job. Let’s talk.
Book a 30-min call with Di-AtomicJust want the skill? Install @di-atomic/cro-audit free.