// recruiter? 60-second version analytics architect · Amsterdam
// analytics architect Frederic Labadie / Amsterdam

Where what users say meets what they actually do — that is the measurement problem.

Senior analytics architect, 14 years in. I build the tracking, pipelines, and research practice that make product, marketing, and UX teams stop guessing.

14 yrs analytics Telus Digital 11 Amplitude certs MSc Data Analytics HIPAA + consent design EU work-authorised
Open the case library Download CV

alert noise

83% V1 production: 230 alerts to 39 posts across 13–26 April 2026.

analyst hours

38h Recovered per month in a state-regulated retail platform handoff.

BQ pipeline cost

~50% Reduction from repeated-scan collapse in a BigQuery pipeline redesign.

CDP migration CTR

+10.8% YoY CTR lift from Responsys to Bloomreach migration with zero campaign downtime.
Section 01 / selected work three engagements / three different problem shapes / full library in Work

Case / Question discovery

Consumer electronics retailer / 2023–2025 / under NDA

The workaround broke. The data showed the gap. Frontline staff knew why.

I was building a lifecycle segmentation model when an anomaly caught my eye: a cohort returning shortly after purchase to buy specific components. A Sense-Making-informed conversation with floor staff surfaced the cause: the designed system had never supported the informal compatibility workflow they depended on.

Most interesting findings in analytics arrive while you are doing something else. The role of instrumentation is to make the gap visible enough that the question becomes worth asking.

The client subsequently shipped a compatibility configurator. It is now central to the business, with a second iteration planned for 2026.

Read the case
fig. 01 / how the question surfaced discovery chain / not a chart
  1. BuildingLifecycle segmentation model
  2. NoticedSmall returning-cohort pattern
  3. ConversationShop-floor staff / SMM neutral protocol
  4. Root cause foundShadow IT compatibility tool, broken
  5. Client shippedCompatibility configurator

The chart you usually want here does not exist: the engagement is under NDA and the transaction series is confidential. Shown instead: the actual sequence of beats. The proof is the product that shipped.

Case / BI enablement

State-regulated retail platform / 2025–2026

The reporting layer survived the people who built it.

BigQuery, Dataform, and Looker Studio reporting infrastructure replaced recurring manual report pulls, paired with a three-session structured handoff so the client could maintain and extend it independently. When the full team rolled off in June 2025, incoming analysts were operational on day one. The system has run independently since March 2026.

The 38 hours recovered per month is not the point. The point is what those hours were redirected to, and that the redirection survived the people who started it.
Read the case
proof panel / adoption handoff static result / anonymised 
SELECT hours_recovered, sessions, ramp_up_days
FROM   engagement.adoption_handoff
WHERE  client = 'sslp_retail'
hours_recovered38 / mo
sessions3
ramp_up_days0

Case / Mixed-methods research

National home services franchise / Stanley Steemer / 2021–2022

Telemetry locates. Method diagnoses.

Quantitative GA4 telemetry paired with a UserZoom protocol built on Dervin's neutral-questioning. The funnel showed where users dropped off. It could not say what kind of gap they were experiencing.

A winning A/B variant was winning despite a vocabulary problem, not because of clean messaging. Optimising on the quantitative signal alone would have entrenched the wrong message.

This is the case behind the talk "Finding the Hypothesis When You Only Know the Question," and the methodology piece the Sense-Making references point back to.

Read the case
fig. 03 / mixed-methods protocol three findings / three gap types
QuantGA4 / A/B / funnellocates where
QualSMM / UserZoomdiagnoses what
DiagnosisGap-type diagnosisdesign responds
F1 / Problematic gap

Cleaning categorised as a treat, not maintenance: gap is pre-funnel.

F2 / Confusion gap

Promo Code was not the same thing as Online Specials.

F3 / Pre-arrival

The funnel begins upstream of the homepage.

Section 02 / public projects live builds / open source / no brief

Public projects

What I build when no one has briefed me.

Four live independent projects. The code is open, the deployment is mine, and the argument lives in the project itself.

Static React / Vercel / Claude / Redis

Signal Conduit

Analytics-governance workbench for canonical tracking specs, multi-analyst review, platform mapping, and Jira handoff kits.

Live app Source

Quarto / Hugging Face / Vercel

Dervin's SMM

A practitioner's guide to Brenda Dervin's Sense-Making Methodology, paired with a live question rewriter for survey copy.

Live site Source

Next.js / Claude / Supabase + pgvector

Writers Room

Multi-agent collaborative workspace with specialist agents, chained calls, document RAG, and read-only review links.

Live app Source

Next.js / Claude / Spotify API

VibeReader

Book-to-song and song-to-book recommendations based on mood, texture, and thematic resonance rather than genre tags.

Live app Source
Section 03 / approach the working position
Frederic Labadie, analytics architect in Amsterdam
Amsterdam / 2026

Approach

Behaviour is the most honest data source available.

People make their problems visible through their actions. The doctor who builds a workaround. The leadership team that keeps requesting the same report in five different formats because the underlying question has never been answered.

Principle 01

Start with the question.

Before implementation: what decision are you trying to make, and what is the data not doing for you?

Principle 02

Build for adoption.

The job is done when the people around the system can use it independently.

Principle 03

Work across the boundary.

Instrumentation and qualitative method belong together when the measurement layer needs to surface human questions.
Section 04 / stack a scanning shortcut

Stack

What I work with, day to day.

Analytics

GA4 / Adobe Analytics / Amplitude / Segment CDP / Mixpanel evaluation

Data

BigQuery / Snowflake / Databricks / dbt / Dataform / SQL / Python

AI and method

Claude / Hugging Face / Vertex AI prompt design / pgvector / Sense-Making / A/B testing

Full skills and certifications breakdown

Section 05 / contact open to senior roles / Amsterdam

Contact

Open to senior IC roles in Amsterdam.

Best fit: analytics architecture, product telemetry, analytics engineering, and UX telemetry roles where the measurement layer is treated as a research instrument. Currently full-time at Telus Digital; standard notice applies.

Emailfrederic.labadie@gmail.com CVDownload PDF LinkedIn/in/frederic-labadie GitHub/fredericlabadie Recruiters60-second quick scan