Skip to content
RAY BEHARRY

Agent 46 of 71 · Conversion Optimization

Landing Page Strategist

Make every landing page do one job for one buyer at one stage. Audit page structure against the Tension → Turn → Proof → Next Step anatomy, check that the hero matches the positioning canon, move named proof up to where doubt peaks, and cut sections that do not move the visitor toward a single next step.

Agent contract

Use to audit or plan a landing page: section order, message hierarchy, proof placement, CTA path and friction points, scored against the approved positioning and stage. Use before a redesign or a paid campaign points traffic at a page.

When it runs

  • A campaign, launch or paid channel is about to send traffic to a new or existing page.
  • Funnel Analysis Agent reports a page whose bounce or scroll-drop sits well below the site baseline.
  • Positioning canon changed and live pages still carry the old category or promise.
  • A stakeholder asks why a page is not converting or requests a redesign.

What it needs

InputRequiredFrom
Page URL or page copy exportYesYou
Approved positioning and stage emphasisYesPositioning Canon
Proof items with ProofIDs and expiryYesProof Vault
Target ICP and personaYesICP Library
Page-level engagement (sessions, scroll depth, exits) for the last 28 daysOptionalGA4 (integration)

How it works

  1. Declare the page's single job: the persona, the buying stage and the one conversion action it exists to produce.
  2. Map every section top to bottom and label it Tension, Turn, Proof, Next Step or 'no role'; sections with no role are cut candidates.
  3. Compare hero headline and subhead to the approved positioning canon entry; flag category, problem or benefit drift with the POS record ID.
  4. Locate each claim on the page and check it against proof-vault; decision-stage claims without a valid, unexpired ProofID are marked blocking.
  5. List friction points: competing CTAs, form fields beyond what routing needs, unexplained jargon, proof placed below the fold of the objection it answers.
  6. If GA4 data is supplied, annotate sections with scroll reach and exit share for the stated date range; otherwise mark behaviour notes as hypothesis.
  7. Prioritise fixes by expected impact × effort and name which fixes deserve an A/B test versus a straight correction.
  8. Write the audit with the landing-page-audit template and hand testable fixes to the experiment designer.

What it delivers

Page job statement, section map with anatomy labels, canon-drift findings, claim-to-ProofID table, friction list, prioritised fix list split into 'just fix' and 'test first'.

Output contract: landing-page-audit

Quality bar

  • The page job names exactly one persona, one stage and one primary conversion action.
  • Every flagged claim cites the POS or PRF record it conflicts with or lacks.
  • Behavioural observations carry a GA4 date range, or are labelled hypothesis.
  • No more than seven prioritised fixes; each states the friction it removes.

Hands off

When a person steps in

  • Escalate to the page owner if a live claim is unproven or its ProofID has expired; recommend unpublishing the claim now.
  • Escalate if regulatory or legal language on the page appears to have been altered.

Example

Request

“Audit the /audit-automation demo page before we put LinkedIn spend behind it.”

Result

Page job: Controller, consideration, book demo. 11 sections mapped, 3 with no role. Hero says 'finance platform' vs canon POS-0007 'continuous audit readiness'. 2 claims missing ProofIDs (blocking). Form trimmed from 9 to 5 fields recommended. 4 'just fix' items, 2 'test first' items handed to the A/B Experiment Designer.

Illustrative example from the package documentation, not a client result.