Michael V.

Most CRO people hand you a deck. Most developers hand you a build. I do both, which means the test I recommend is the test I write, QA, ship and read out. I run experimentation for Gannett across the USA TODAY Network, where the job is the whole loop: research, hypothesis, sample size, build, QA, analysis, and the guardrail metrics that stop a bad result going unnoticed. Before that: AOL/Yahoo, CarGurus, Hunter Fan and MONQ. WHAT I DO Experimentation - A/B and multivariate test design, from hypothesis through power calculation to readout - Test build and QA across browsers and devices, client side or server side - Auditing an existing program and telling you which results you can actually trust - Research, backlog and prioritisation when there is no roadmap yet Analytics and tracking - GA4 and Google Tag Manager, including server side - Conversion tracking, event schemas, and attribution you can defend - BigQuery and SQL, Looker Studio dashboards - Fixing tracking that was set up wrong and is quietly poisoning every decision Build - React, Next.js, TypeScript, Python, Node - Landing pages, full sites, internal tools, integrations - Core Web Vitals and page speed remediation - Accessibility and technical SEO AI and automation - LLM applications, agents, and AI assisted pipelines - Scraping and data extraction - Workflow automation that removes the manual step nobody wants to own HOW I WORK Async by default, with clear written updates. I would rather send you a Loom and a document than book a meeting. I move fast because I use AI heavily in my own workflow, and I am happy to say so. My GitHub is linked on this profile. The public work sample there is a configuration driven form engine in TypeScript: conditional branching, schema linting, and WCAG 2.1 AA accessibility throughout. If you have a specific problem, tell me what you have tried and what the data says. That is usually enough for me to tell you whether I am the right person, before either of us spends money finding out.