307 checks. 90 seconds. One structured response.
Everything that happens between POST and 200 OK — and why it produces a score you can trust.
THE PIPELINE
Three steps. One structured response.
One POST request. 307 checks fire in sequence. Structured JSON returns.
curl -X POST \ https://api.weavn.app/v1/scan \ -H "Authorization: Bearer wdoc_live_••••" \ -d '{"url": "https://your-site.com"}'
Build against this schema once. Every URL returns identical structure.
27 diagnostic categories.
Every scan evaluates the full conversion surface of a page — not just the obvious elements. Checks are gated by site type: a SaaS site gets SaaS checks. An ecommerce site gets ecommerce checks. Universal checks run on everything.
POST to response. Eight steps.
URL SUBMISSION
Request received
You POST a URL with optional parameters: fields to return, finding_depth (brief or full), async mode, page paths for multi-page scans, and site_type override. The API validates the URL and queues the scan.
POST /api/v1/scan{ "url": "https://your-site.com", "finding_depth": "full", "async": false}
BROWSER RENDERING
Full page render
Browserless Pro renders the full page using headless Chrome with stealth mode enabled. JavaScript executes completely before analysis begins. We see exactly what a real visitor sees — not the raw HTML.
SITE CLASSIFICATION
Intelligence before checks
Before a single check runs, the scanner classifies the site across four dimensions: site type (SaaS, ecommerce, service, B2B, creator, local), price point tier, buyer complexity (transactional, consultative, enterprise), and traffic temperature. This classification gates which checks apply and how severity is weighted.
{ "site_type": "saas", "buyer_complexity": "consultative", "score_profile": "saas_consultative"}
CHECK EXECUTION
307 checks, correctly applied
Universal checks run on every site. Vertical checks run only when relevant — ecommerce checks never fire on a SaaS scan, SaaS checks never fire on a service site. Page-type checks apply only to the matching page. This means every finding is relevant to what's actually being scanned.
AI ANALYSIS
Sonnet evaluates the page
Claude Sonnet reads the rendered page content and evaluates each applicable check. Every finding must reference specific visible content — never fabricated. The model identifies what's present, what's absent, and what's present but incorrectly placed.
WEIGHTED SCORING
Score calibrated to site type
Seven dimension scores are computed and weighted by the site's classification profile. Trust signals weight higher for consultative buyers. Checkout friction weights higher for ecommerce. The overall score reflects what actually matters for that site's conversion context — not a generic average.
{ "score": 61, "score_profile": "saas_consultative", "dimensions": { "trust_signals": 37, "conversion_architecture": 58, "message_clarity": 63 }}
BENCHMARKING
Positioned against real sites
Each dimension score is compared against accumulated corpus data from sites in the same vertical. A trust score of 37 on a SaaS site returns "Bottom 15% of SaaS sites" — not a generic percentile. Benchmarks improve as corpus volume grows.
{ "dimension_benchmarks": { "trust_signals": { "score": 37, "average": 61, "percentile_label": "Bottom 15% of SaaS sites" } }}
OUTPUT ASSEMBLY
Structured response returned
Findings are sorted by priority_rank (P1 fix this week, P2 fix this month, P3 when you can), each with fix_effort and impact_tier. Strengths surface what's working. The full response is a complete conversion diagnostic in one structured JSON object.
{ "findings_summary": { "p1_count": 3, "p2_count": 8, "p3_count": 12 }, "strengths": [...], "findings": [...]}
RESPONSE SCHEMA
One response object.
ACCESS
Two ways to use the same engine.
The scan engine and its output are identical regardless of how you access it.
DASHBOARD
Results without writing code.
Scan any site from the dashboard. See your score, ranked findings, AI-rewritten copy, and how you benchmark against your vertical. Track over time. Send white-label reports to clients.
API
Build conversion intelligence into anything.
POST any URL, get structured JSON. Integrate the scan engine into your product, pipeline, or automation. Batch endpoint, async mode, webhooks. No dashboard required.
USING THE DASHBOARD?
See what a scan returns in plain English.
Score, ranked fixes, rewritten copy — no code required.
How it works →BUILDING WITH THE API?
Full endpoint reference and schema.
Plans, rate limits, async mode, batch endpoint.
API docs →