Key points from the Data On Duty Masterclass :

Data is worthless if it doesn't generate insights. And insights are worthless... if the data is wrong.

This conviction was the starting point for the Masterclass on April 3, 2025, hosted by Eric Dumain (founder of Data On Duty), alongside two figures from the analytics world: Julien Coquet and Eric Bricier. Both have dual legitimacy: technical expertise and business insight.

The aim: to understand why tracking often fails, how to get out of it, and above all, how to industrialize the production of reliable insights.

From data to insight: what are we really trying to produce?

Before talking about tools, let's ask ourselves the question: what is an insight?
The masterclass gives a pragmatic definition, combining theory and field experience.

An insight is :

• 🔎 An answer to a specific business question
• ✅ A validation (or invalidation) of a hypothesis
• 💡 A chance discovery that changes strategy
• 📊 A key element in a dashboard... provided it's properly interpreted

But all this rests on a fragile pillar: data quality, an unstable foundation in a changing environment.

The promise of reliable insight presupposes consistent data collection over time. In practice, however :

What derails quality:

• Constantly evolving digital environments (techno, UX, app, site...)
• Undocumented production releases
• Untested or partially tested changes
• Tracking tools not aligned with business needs
• And above all: a lack of clear governance

The result? A domino effect, as illustrated in the masterclass:

🔁 A collection bug → discovered too late → corrupted data → flawed analyses → biased decisions → misallocated budgets

And the problem repeats itself, because the foundations are too fragile.

Non-quality: a cost that is often invisible... but very real

The discussion between Eric and Julien highlighted the real cost of faulty tracking:

• 📉 Distorted KPIs
• ❌ Wrongly evaluated campaigns
• 💰 Poorly invested marketing budgets
• 😰 And above all: a silent loss of income

As Eric Bricier summarized:

"We're investing heavily in tracking... without knowing what we're really measuring."

Regulation: a constraint or a catalyst for quality?

RGPD, ePrivacy, Consent Mode, DMA, TDA... there is no shortage of constraints.
But the speakers made a positive point of it:

"Regulation, properly managed, becomes an opportunity to better structure one's data."

Rather than slowing down teams, the RGPD and legal constraints are forcing us to rethink the foundations: clarifying needs, documenting tracking, aligning business, tech and legal. The result: more reliable, more usable data, and a considerably reduced time-to-market. What seemed like a constraint becomes a real performance lever.
In other words: compliance = governance lever.

Marking plans & Datalayer: pillars or Achilles heel?

The masterclass highlighted a paradox:

• The tagging plan and datalayer are designed to structure tracking.
• But in practice, they often become points of fragility.

Julien compares them to Lego: powerful if well designed, catastrophic if not.
Many companies still manage these elements via obsolete Excel files, not aligned with business challenges.

What about server-side technologies?
They're not enough. They require a budget, an adapted architecture, and do not circumvent privacy obligations.

The human limit: overload, dependencies, complexity

One of the highlights of the masterclass was the observation that tracking is still too traditional, despite the use of industrialized tools.

"We've industrialized the build... but maintenance remains manual."

Over-subscribed testers, overworked trainees, DPOs lost in the technicalities: tracking still relies on too few resources, without appropriate tools.

Build, deploy, test: towards an industrial methodology

Eric's side:

A highly technical but effective method:

• Map in Google Sheet/JSON
• Python/Node.js scripts
• Reporting via BigQuery/SQL

In the end, it's not an automated logic, but an artisanal one, with a lot of manual effort and little sharing between teams.

Julien's side:

A focus on cross-functional business needs:

"Marketing, data, IT, agencies: everyone has their own vision. We need a single, shared, living source."

Traditional tools (Excel, Jira, etc.) are neither scalable nor collaborative, and quickly become obsolete.

Budget: what tracking (really) costs

Tracking doesn't just cost in licenses. It costs :

• Human time (QA, acceptance testing, corrections, etc.)
• Maintenance (manual tests, errors, faults)
• Late to the Time-to-Market
• And in lost business opportunities if insights are wrong

Even the most powerful acceptance tools have high installation costs, often incompatible with the short cycles of digital marketing.

🚀 Data On Duty: industrializing tracking, at last

Data On Duty is a concrete response to this situation, designed to meet today's challenges:

It's a collaborative platform 100% dedicated to Digital Analytics, which :

• Automates tracking creation, deployment and validation
• Modular management of marking plans
• Natively integrates privacy (RGPD, Consent Mode...)
• Monitors data in real time
• Drastically reduces the need for manual QA

User feedback is unquestionable:

Julien: "A cockpit for monitoring quality, triggering alerts, and industrializing without complexity."

Eric.B: "100% data-driven, 100% revenue protection, 90% time savings."

What's next? AI to go even further

The platform is already integrating artificial intelligence to :

• Detect data anomalies before they become visible
• Generate route optimization or branding recommendations
• Automatically prioritize value-added corrections

Business AI, focused on performance.