How a lean CRO team runs continuous experimentation across three brands and twenty locales
Van de Velde built a self-service testing programme across Primadonna, Marie Jo and Sarda, while decreasing load on development.
“I can run a lot of tests at the same time, in parallel to the development team. Tests on the PLPs used to require quite a bit of development. Now we do them ourselves, which also drives down the cost. ”
The Challenge
As a premium lingerie and swimwear group with multiple brands across international markets, Van de Velde sees CRO as an important lever for continuously improving the customer experience.
Primadonna, Marie Jo and Sarda are very distinct brands, each with its own website, audience and mix of markets. Those markets run across roughly twenty locales, each with its own language, its own content, and sometimes its own commercial rules. That is a large surface area for a small team.
Every experiment that touched the front end had to be built by a central development team, whose roadmap was already well filled with the group's bigger projects.
Due to this dependence, the CRO team's ideas had to be shortlisted. The tests that lost out were rarely the weak ones. They were often the small ones. Copy changes, content tiles, CTA experiments. Things that were never quite worth a development slot of their own.
The result was a set of high-traffic, high-stakes websites that were not being systematically improved, and a backlog of ideas that were never able to be validated.
“The simple tests, changing a title or a piece of copy or a CTA. That was never a technical problem. It just never happened, because it was such a hassle. That was one of the biggest wins for me.” - Imogen Creytens, Digital Content & CRO Specialist
The Solution
The CRO team builds its own tests.
Van de Velde implemented Dalton across all three brands. Their development team handled the integration per brand. From that point on, individual experiments no longer needed them. The CRO team now builds, reviews, and launches its own experiments across homepages, collection pages, and product detail pages, by prompting in the visual editor or accepting Dalton proposals.
More tests, running at once
Within weeks of going live, the team had four experiments running on Primadonna and four on Sarda. The rigour didn't change. A test still runs for four weeks before it's called, and never less than two, exactly as it always did. What changed is how many can run inside that window.
Ideas the team wouldn't have had
Dalton scans each brand's pages and proposes variants that are ready to review, giving the CRO team additional inspiration. The content cards on the collection pages started that way. They weren't on the roadmap. The suggestion appeared, and the team decided it was worth trying.
"The suggestions are definitely valuable. I've pulled a few of them out, including things I hadn't thought of myself. And I'm very happy with how fast you can set a test up." - Imogen Creytens, Digital Content & CRO Specialist
Fewer debates, more data
A disagreement about a headline, a layout or a piece of content stops being a matter of whose opinion prevails and becomes a question with an answer, four weeks out.
That turns out to be as valuable inside the organisation as it is on the page. The CRO team is now able to foster a true experimentation culture and bring data into discussions that might otherwise be based on opinions.
Three brands, one workflow
The three brands are very different. Primadonna is built around fit. Marie Jo leads on design. Sarda is younger and more fashion-forward. The same page type calls for a different hypothesis on each site, and one generic playbook is not going to work.
Each brand keeps its own account, its own pages, and its own tone of voice. The workflow to build, review and launch is identical across all three.
One test, twenty locales
A result in one brand or market can immediately become a hypothesis everywhere else. Rather than rebuilding from zero, the team can copy a test to other locales, or group markets together in one experiment.
The Results
Beyond the individual wins, Van de Velde has a testing programme it runs in-house with velocity and at scale.
Two examples of validated experiments on collection pages, both measured over tens of thousands of live sessions.
+12.9% Purchases
From bringing verified customer review content into the product grid.
90% confidence · 46,700 sessions · Primadonna’s collection page
+24.9% Purchases
From surfacing bestseller signals on products already selling fast.
92% confidence · 54,500 sessions · Sarda collection page
Why It Worked
Van de Velde already believed in experimentation. They have a dedicated CRO team, a testing discipline they hold to, and they treat CRO as a lever to improve the customer experience and learn rather than a side project.
The barrier was dependency. Every test had to go through a development team with its own priorities, so ideas got cut before they could be run.
Dalton made experiments fast and simple enough to build that a lean team can now cover three brands and twenty locales on its own. That is what makes a continuous programme possible at this scale, maximizing the impact of the CRO team, rather than being limited by development slots.