About This Project

SEO Experiment Lab

A showcase project: a working prototype of an AI-powered tool for SEO experimentation in financial content. It combines structured hypothesis frameworks, AI-powered analysis, and data-driven measurement to bring rigor to on-site SEO work in a YMYL category.

Why I Built This

SEO product managers at financial content publishers face a unique challenge: organic traffic is under pressure from AI search (Google AI Overviews, ChatGPT, Perplexity), and YMYL content demands the highest E-E-A-T standards. Yet most SEO work happens ad hoc, without structured experimentation or rigorous measurement.

This tool addresses that gap. It provides a structured workflow, from content analysis through experiment planning to statistical measurement, specifically designed for financial content. The HBML hypothesis framework, E-E-A-T scoring, GEO readiness assessment, and financial-content-specific considerations are all tailored to the decisions a PM at a personal-finance publisher actually has to make.

What This Demonstrates

Content SEO Analyzer

SEO domain expertise, E-E-A-T fluency for YMYL content

Experiment Planner (HBML)

Structured experimentation, hypothesis-driven product work

Page Preview

Programmatic SEO, schema markup, technical fluency

A/B Test Dashboard

Data-driven decisions, statistical rigor

GEO Readiness Scoring

AI search strategy, forward-looking product thinking

Decision Log + Export

Cross-functional communication, stakeholder reporting

Tech Stack

Next.js 16App Router, Server ComponentsTypeScriptStrict mode, Zod validationTailwind CSS 4CSS-based config + shadcn/uiVercel AI SDKClaude streaming with structured outputRechartsDashboard visualizationsPostHogProduct analyticsVitest + PlaywrightUnit tests + E2E + a11yGitHub ActionsCI pipeline

System Visuals

Three diagrams that capture how the tool fits together: system architecture, the four-stage analyze-plan-execute-measure workflow, and the analytics flow.

System Architecture
System architecture: Next.js App Router with /analyze, /plan, /preview, /dashboard routes, calling Vercel AI SDK streamObject with Zod schemas, backed by Claude and PostHog. Experiment state lives in localStorage.
Analyze, Plan, Execute, Measure
Four-stage workflow: Analyze with the Content SEO Analyzer, Plan with the Experiment Planner using HBML and ICE scoring, Execute with the Programmatic Page Preview, Measure with the A/B Test Dashboard.
Analytics Flow
PostHog analytics flow: user interactions on each tool fire structured events that feed funnels (analyze to plan to dashboard) and dashboards tracking adoption, time-to-experiment, and decision-log usage.

Built by Marcelo Otero

Product manager and builder. I made SEO Experiment Lab to show how I take a product from problem framing to a shipped, instrumented prototype, applying SEO domain expertise, AI integration, and product thinking end to end.

LinkedIn