Self-Serve AI Marketing Suite & 1,500-URL Content Retro Engine
Internal self-serve AI tooling suite (1,500-URL Content Retro classifier, AI Image & OG generators, 4M+ Job Counter) and agent governance grounded in 49 Google sources.
STAR Engineering Breakdown
01 · Situation & Problem
Non-technical marketers depended on engineering or external SaaS subscriptions for content decay analysis, OpenSearch job counts, keyword research, and branded social/featured image generation.
02 · Technical Execution
Built a one-click Content Retro engine sorting ~1,500 URLs into winners, decaying pages, striking-distance, and dead weight, plus self-serve apps: an AI Image Generator (60+ sessions), OG generator, Keyword Explorer, and a live 4M+ Job Posting Counter.
03 · Cost Engineering & Governance
Replaced a paid featured-image SaaS vendor with a $0 self-hosted renderer after tracing and fixing an upstream text-rendering bug, and cleaned up UTM tags across 136 WordPress posts in a single zero-error REST API batch.
04 · Verified Impact
Administered the 12-user n8n instance and Claude Team org, established agent governance grounded in 49 Google documentation sources, and presented agentic workflows at the company All-Hands.
def classify_url_bucket(m: UrlTelemetry) -> RetroBucket:
if m.clicks_90d >= 150 and m.wow_delta >= -0.05:
return RetroBucket.WINNER
if m.prev_clicks_90d >= 80 and m.wow_delta <= -0.25:
return RetroBucket.DECAYING
if 3.5 <= m.avg_position <= 15.0 and m.impressions_90d >= 500:
return RetroBucket.STRIKING_DISTANCE
return RetroBucket.DEAD_WEIGHTBenchmark Table
| METRIC | MANUAL BASELINE | PROGRAMMATIC ENGINE |
|---|---|---|
| Content Decay & Pruning Audit | Quarterly Manual Spreadsheets | 1-Click Retro (~1,500 URLs Graded) |
| Featured & OG Image Tooling | Paid Third-Party SaaS Vendor | $0 Self-Hosted Renderer + 60+ Runs |
| Sitewide UTM Link Hygiene | Manual Post-by-Post Editing | 136 WP Posts Fixed in 1 API Run |