I'm Raúl Fernández, and I make sure the data arrives the way it should
Sounds simple until you try it: browsers blocking cookies, consent banners that consent to rather more than they should, attribution lost along the way. That's where I live: +25 million events a day that have to land complete, deduplicated and compliant.
Sounds simple until you try it: browsers blocking cookies, consent banners that over-consent, attribution lost along the way. That's where I live: +25 million events a day.
I've led teams of up to 13, yes. But my place is still the codebase, not the 1:1 calendar.
Let's talk RésuméEqual parts technology and letters. That split explains how I approach everything, at work and outside it: through logic and imagination
ABOUT
I've spent 17 years in digital, and the last 8 almost entirely inside one layer: the bottom one, where data is born. Agnostic data layers, cross-domain and cross-device identity, custom middleware so client and server can talk without going through anyone else's collector, MCP servers that let AI operate directly against tag management, analytics platforms and legislative databases. So no, I haven't been sitting still.
And what I've learned in that time is that the data is never actually missing: what's missing is the architecture meant to govern it around the needs of the business. Take attribution: the classic kind is device-scoped — the chain of touchpoints lives in a cookie and breaks the moment someone moves from phone to desktop. I moved it server-side and tied it to the identity engine, so it attributes to people instead of browsers. The funny part is that the touchpoints had been there all along: they were just hanging off the wrong thing.
The other half of the job isn't done by me: it's done by the team without me. That's what I built the MCP servers and multi-agent systems for, automating more than 90% of the department's daily work; a self-service layer so clients manage their own tracking from the dashboard without opening a single ticket; and agnostic data layers that standardise events across every vertical, so switching tools no longer means rewriting everything. The department no longer runs on operational chores: that time goes into understanding the data.
I've led teams — at Jakala I took mine from 5 to 13 people — yes. But my place is still the codebase, not the 1:1 calendar.
Consultancy and agency work took me through consumer electronics, automotive, leisure and entertainment, banking and pharma, to name a few. Always with the same brief: quality data as a competitive advantage.
Implementations at global scale
«Accuracy in analytics is an ideal you almost never reach. Acknowledging that limit is the first step towards building data you can actually decide on, without distorting reality.»
WORK
My most recent work (you know how confidentiality goes, ahem). The first three are really one thing: an identity engine, with attribution and consent built on top of it.
Edge-first identity engine
Identity resolution running at the edge, resilient to ITP/Safari and GDPR-compliant. Syncs anonymous identities with canonical ones through dynamic consent logic, without a single third-party cookie.
User-level server-side attribution
Standard attribution is device-scoped: the touchpoint chain lives in a cookie and breaks when ITP expires it, when the user moves from phone to desktop, or when they cross domains. I moved the touchpoint ledger server-side and tied it to the identity engine: every touchpoint reconciles against a user —deterministically by userID, probabilistically by cookie when there isn't one— rather than against a browser. Attribution stops expiring.
Enterprise cross-domain CMP
Built on the same identity engine: the consent decision stops living in a per-domain cookie and is bound to the user server-side. What someone accepts or rejects on one domain is honoured across the rest, with no third-party cookies and under today's partitioned storage. Centralised policy and automated GDPR compliance across every business vertical.
MCP servers for analytics and legislation
MCP (Model Context Protocol) servers that let an AI system operate directly against tag management platforms, digital analytics and legislative databases. Unified architecture, published open source.
Multi-agent automation system
An ecosystem of specialised agents with swappable skills for development, auditing, documentation and analysis, wired into the MCP servers and AI notebooks. It eats more than 90% of the department's daily work.
Universal identification system
A hybrid system combining deterministic and probabilistic methods with advanced fingerprinting, server- and client-side. Cross-domain and cross-device tracking by triangulating three identification variables.
Custom client ↔ server middleware
Proprietary middleware for direct client-server communication without going through GA4's standard collect endpoints. Full control over transmission to GTM server-side and over latency, sidestepping the protocol's limits.
Agnostic data architecture
A unified data layer across every business vertical: events and parameters standardised independently of the platform. Switching analytics tools no longer means rewriting the whole tracking setup.
Self-service analytics platform
A self-service layer inside the corporate CMS so each team manages its own analytics and marketing IDs, client-side and server-side, with automatic validation. No ticket required.
Advanced auditing tool
A platform-agnostic analytics auditing tool, focused on the data layer, its performance, and control over user consent and cookies (Consent Mode v2).
Chrome extensions for GTM
Two public extensions on the Chrome Web Store: GTM DecodeEZ, for AI-assisted client-side GTM analysis, and sGTM PreviewEZ, for debugging server-side GTM. Over 100 active users between them.
Advanced GTM templates
Custom templates for client-side and server-side GTM: GA4 multi-domain setups, Google Ads enhanced, bespoke sGTM clients and specialised tags for specific enterprise needs.
The main clients I work, or have worked, with:
FEVER
I standardise data through an agnostic layer, optimise the tagging systems and answer for data quality at the world's leading live-entertainment discovery platform.
SAMSUNG
Nearly a decade as my main client, covering Spain and Portugal. I started on content and ended up as Tagging Manager and Technical Project Lead, running tracking for the operation across two markets.
LEROY MERLIN
More than two years responsible for making sure the data didn't go missing and, above all, that it meant something — at one of Spain's largest retail e-commerce operations. PhD with honours in data persistence.
SANTANDER
Technical consultancy (often rather technical) on data collection, digital analytics and data engineering for Spain's largest banking group. Seasoned query slinger.
MSD
Responsible for the digital production team — 7 people — for 6 months, as Project Manager at one of the big pharma companies.
K-GLOBAL
A group with three furniture e-commerce platforms. We audited their analytics, proposed a data layer populated mostly from the back end, and left the whole thing spotless.
SPAIN BUSINESS SCHOOL
Lecturer on the Digital Analytics master's since 2021. I teach Google Tag Manager and the fundamentals of data collection and technical analytics — which is exactly what nobody else teaches.
CONTACT
If your product has a problem in the layer where data is born —tracking that doesn't arrive, identity that doesn't add up, consent quietly eating half your users—, that's a conversation I'm interested in. Write to me.
And if what you're bringing is a role, tell me about the team and the scope before the title. Audits, workshops or a one-off question work too. I don't bite.
MADRID, SPAIN · CET/CEST