Workflows
CRO Pipeline
Audit web pages for conversion optimization and generate prioritized A/B test experiments.
CRO Pipeline
The CRO Pipeline workflow audits web pages for conversion optimization issues and generates prioritized A/B test experiments with specific implementation guidance.
Input Schema
{
url: string; // URL to audit
pageType?: 'landing' | 'homepage' | 'pricing' | 'feature' | 'signup' | 'checkout';
conversionGoal?: string; // e.g., "increase signups by 15%"
}Pipeline Steps
Step 1: Page Extraction
The page content is fetched via the Web Extract tool (with SSRF protection) and analyzed for structure, content, and technical elements.
Step 2: CRO Audit
The CRO Specialist agent audits the page across 8 dimensions:
| Dimension | Score | Assessment |
|---|---|---|
| Headline Clarity | 0–100 | Value prop communicated in under 3 seconds? |
| CTA Visibility | 0–100 | Above fold, single clear action, action-oriented copy? |
| Social Proof | 0–100 | Specific testimonials near decision points? |
| Urgency/Scarcity | 0–100 | Genuine or fabricated? |
| Trust Signals | 0–100 | Security badges, guarantees, press mentions? |
| Form Friction | 0–100 | Number of fields, validation, progressive disclosure? |
| Mobile Responsiveness | 0–100 | Touch targets, thumb zones, font sizes? |
| Above-Fold Content | 0–100 | First viewport tells the complete story? |
Step 3: Issue Identification
Each issue is categorized:
{
severity: string; // 'critical' | 'high' | 'medium' | 'low'
area: string; // Which CRO dimension
detail: string; // What's wrong
fix: string; // How to fix it
}Step 4: A/B Test Design
For each significant issue, an experiment is designed:
{
name: string; // Descriptive test name
hypothesis: string; // "If we [change], then [metric] will [improve]"
variant: string; // What the test version looks like
control: string; // Current state
primaryMetric: string; // What to measure
sampleSize: number; // Minimum for statistical significance
}Step 5: Quick Wins
Identifies fixes that can be implemented immediately without A/B testing — low-risk, high-impact changes like fixing broken CTAs, adding missing trust signals, or reducing form fields.
Output Schema
{
url: string;
overallScore: number; // Weighted average 0–100
dimensionScores: Record<string, number>; // Score per dimension
issues: Issue[]; // All identified issues
abTests: ABTest[]; // Designed experiments
quickWins: string[]; // Immediate fixes
}