Most analytics problems start upstream — at collection, not in the report.
Tracking architecture, GA4, data pipelines and privacy compliance — built so the numbers are right before anyone reports on them. Working with Google Analytics since 2013, agency-side until 2023 and independently since.
Experience
I have worked with Google Analytics since 2013 — more than ten years implementing, migrating and repairing tracking across every version of the platform, including the full migration from Universal Analytics to GA4.
I started as an analytics intern at an agency in São Paulo and became a data analyst there. In 2017 I moved to Dublin, where I spent almost seven years at a digital agency, going from analyst to Director and Partner of Data & Analytics. I left in 2023, took a sabbatical year, and founded BRLabs in 2025.
I also read code. The developers implement, but whoever writes the specification should understand what they are asking for: I can audit how tracking was actually implemented, point to where it is wrong in the code, and give technical support to the team that has to fix it.
Degree in Systems Analysis from the Instituto Federal de São Paulo. Based in Buenos Aires, working with clients in Brazil, Europe and the US, in English, Portuguese and Spanish.
What I do
05
Dashboards that disagree, conversions that vanish, attribution that never adds up — these are almost always symptoms, and teams spend a quarter treating symptoms. Garbage in, garbage out is a foundational principle of computer science, and trying to clean up garbage out when you started with garbage in is what costs time, money and trust. Sublata causa, tollitur effectus: remove the cause and the effect ceases. That is why methodology is not optional at the upstream. When the upstream is done right, every other layer can trust the data.
// the risk
Data protection laws turned undisciplined collection into legal exposure, not just a reporting problem.
// what's coming
AI models decide from what you collect. Sloppy collection corrupts every layer after it — and now the decisions are automatic.
Tracking architecture & GA4
dataLayer design, Google Tag Manager implementation, server-side tagging, event taxonomy and conversion setup. User ID and identity resolution, so that people rather than sessions become the unit of analysis. Plus review of the implementation in the code itself, when your developers want a second pair of eyes.
GA4GTMserver-sideUser ID
Data pipelines & warehouse
Getting data out of the tools that rent it to you and into somewhere you own it. ETL from production databases, BigQuery modelling, scheduled jobs, and reporting built on top of the warehouse instead of on top of an export button.
BigQueryETLLooker Studio
Analytics audit & data integrity
Discrepancy investigation across GA4, ad platforms and the source database. UTM and campaign hygiene, duplicate and missing event detection, and documentation your team can maintain without me.
auditQAdocumentation
Privacy & compliance
LGPD and GDPR applied to the tracking layer, where most of the risk actually sits. Cookie audits, consent mode, CMP selection and configuration, international transfer mechanisms, and a clear record of what you collect and why.
LGPDGDPRconsent mode
Analysis & reporting
Defining the questions the business actually needs answered, then building the report that answers those rather than every metric available. Attribution modelling, cohort and retention analysis, customer acquisition cost and lifetime value — and reading the result: what the numbers say and what to do about it.
attributionLTV / CACcohort analysis
The framework
Martex is a MarTech platform for validating, auditing and monitoring analytics implementations. I designed it, wrote the code and maintain it — the product, the framework and everything under them.
It came out of the consulting work and became two things: an analytics integrity platform, and the upstream data integrity framework behind it. The core move is treating tracking as a contract. Blueprint specifies what should be collected, under which conditions and in what shape — a versioned document you can hand to a developer. Pulse then watches live data against that contract and alerts you when an event stops firing or a value drifts, before your client notices.
It is not a cascade. Validation that fails sends you back — and step five never stops running.
12 tools · free tier for up to 3 clients · 7-day PRO trial
Messages go straight to my inbox. I schedule calls manually rather than through a booking link, so tell me what you need and I'll come back with times.
