Selected work
Experience at scale,
Deep focus on impact.
Zoé van Havre, Principal
Case study · Aviation · Enterprise experimentation program
From scattered testing to a company-wide learning engine.
LATAM Airlines had the ambition to run on evidence. It had teams, tools, and traffic. What it did not have was a system that turned all of that into trusted, comparable learning. That is what got built.
The situation
A large aviation organization with hundreds of people spread across dozens of product and commercial teams. Experiments were happening, but unevenly. Standards varied by team. Results were not always trusted, not always comparable, and not always connected to the decisions they were meant to inform. The volume of testing understated the potential, and the potential was significant.
What I built
The work was not to run more tests. It was building the system underneath the tests so that more of them would be worth trusting.
- Measurement standards and metric definitions that held up across teams, so a result in one part of the business meant the same thing as a result in another.
- Statistical methods and quality bars that raised the credibility of what got shipped and what got believed.
- Decision processes that connected experiment outcomes to the choices leadership was actually making.
- The enablement, training, and operating rhythm that let 35+ teams run the system themselves rather than depend on a central bottleneck.
The outcome
- Experiment volume grew to nearly 1,000 experiments per year.
- The program spanned 700+ professionals across 35+ teams.
- More importantly than the count: the results teams produced became trusted and comparable, which is what lets a number turn into a decision.
What leadership could do afterward that they couldn't before
- See which teams and initiatives were producing measured impact, not just activity.
- Compare results across the organization on a common standard.
- Make investment and prioritization calls with evidence in hand.
- Rely on teams to keep learning independently, at a pace the organization set rather than one a central team could sustain.
Built inside one of the most rigorous experimentation cultures in the world.
Before consulting, eight years at Booking.com, an organization that runs experimentation at a scale most companies never see: on the order of 25,000 experiments a year. Progression from data scientist to manager, with methods like covariate adjustment and sequential testing shipped into production at that scale, and published research to match. The standards that shape my client work were set in an environment where being wrong was expensive and measurement was not optional.
Make the signal
clearer.
Bring me the question your organization cannot answer with confidence yet.