Prime Reviews Pro

PrimeReviewsPro is a free Amazon-listing analyzer (histogram-shape distribution analysis, not NLP — we score the math, not the text). We get a trickle of YouTube traffic. The GA4 dashboard this week: - 88 YouTube page_views in 14 days - 87% engaged sessions on /analyze - 0 outbound_click events Zero clicks to Amazon. Zero affiliate revenue. So I started debugging from the result page upward. Three stacked issues: 1. Wrong landing page. The on-video URL banner on our Shorts says "primereviewspro.com" — bare domain. Viewers tap it, land on the brand homepage (90% bounce). The page that actually converts is `/analyze`. Fix: nginx `if ($arg_utm_source = youtube) { return 302 /analyze?$args; }`. Narrow redirect — preserves all UTMs, only fires for YT traffic. 2. Affiliate path required two hops. The result UI's cross-promo cards linked to INTERNAL content pages (`/monitors`, `/printers`). Visitors had to click through to a content page, THEN click an Amazon affiliate link from there. Two hops = catastrophic drop-off. Fix: added a direct-outbound `#amazon-picks` block above the existing cards. Three vetted products, direct `?tag=primereviewsp-20` links, `target="_blank"`. 3. The decoupling insight. While testing the picks reveal, I hit Amazon throttling on our `/api/analyze/free` endpoint. The demo failed silently. Picks were gated on the demo result succeeding — so when the API throttled, the picks stayed hidden. Decoupled them: picks reveal on page-load for YT visitors regardless of demo success. The affiliate path shouldn't depend on a third-party API's mood. Verification cron set for 5/21 (7d post-fix) measuring `auto_demo_triggered + outbound_click + affiliate_click` delta from the 0/0/0 baseline. Builder Q: how do you keep conversion paths resilient when they depend on flaky upstream APIs? Caching, fallback content, decoupled reveal — what worked for you?