
Best Western Hotels & Resorts® France doubles its experimentation velocity and frees up its CRO team with PBX
Industry
Travel
key features
Prompt Based Experimentation

2×
more tests launched
20 min vs 4h
to launch a test
up to 90%
time saved
TL;DR: Best Western® Hotels & Resorts France x Kameleoon: Experimentation without technical bottlenecks
With PBX, Best Western® Hotels & Resorts France has reduced its reliance on development teams and significantly accelerated its experimentation programme.
Key takeaways:
- Tests launched in 20 minutes to 4 hours (down from several days)
- Twice as many experiments run with the same team
- Up to €2,000 saved per test
- A more mature test-and-learn culture
The challenge: A strong CRO strategy held back by execution
Best Western® Hotels & Resorts France, a cooperative of more than 330 independent hotels across France, operates in a highly competitive ecommerce market. The digital team, led by Ecaterina Pisica, Ecommerce Manager, oversees acquisition, SEO/GEO, CRO and UX, alongside Carole Portal, UX Project Manager.
Although the team had a mature experimentation programme, one challenge remained: execution.
Every test followed a long and resource-intensive process: brief → IT request → development → QA → deployment.
Because every experiment depended on development resources, the process created a major engineering bottleneck and slowed the rollout of new ideas.
The result was delays of several days, sometimes weeks, while a growing backlog of promising experiments remained untouched.
{{quote-1}}
The turning point: PBX unlocks team autonomy
PBX (Prompt-Based Experimentation) was introduced to remove this bottleneck.
The platform allows teams to describe an experiment in natural language and instantly turn it into a ready-to-launch test, including targeting.
The transformation happened step by step:
- Aligning the team around a shared experimentation vision
- Building a clear roadmap despite limited resources
- Using Kameleoon to accelerate execution
- Prioritising experiments with the ICE framework
{{quote-2}}
To illustrate the impact on both experimentation speed and team autonomy, here are four examples implemented on the hotel's booking website.
{{use-case-1}}
Replacing the full address with the Paris district
The challenge: For hotels in Paris, the arrondissement is often more meaningful to visitors than the full street address.
The solution: Replace the address with the arrondissement using a simple prompt.
The result:
- Test launched in 20 minutes (previously 1 to 2 days)
- Improved page readability
- Approx €400 saved in external development costs
{{metric-block-1}}
{{use-case-2}}
Adding navigation anchors below the hero gallery
The challenge: Visitors had to scroll through long hotel pages before reaching key information.
The solution : Add navigation anchor links directly beneath the hero gallery.
The result :
- Test launched in just 2 hours
- No significant performance impact
- Around €500 saved
Even though the experiment did not improve performance, the team was able to learn immediately and move on to the next idea without tying up development resources.
{{metric-block-2}}
{{quote-3}}
{{use-case-3}}
Reducing the size of decimal prices
The challenge: Small visual details can influence how users perceive pricing.
The solution: Slightly reduce the size of the decimal digits displayed on hotel pricing.
The result: +8% increase in average order value
{{quote-4}}
{{use-case-4}}
Promoting last-minute offers on the homepage
The challenge: Last-minute offers were underused despite their potential to improve occupancy.
The solution: Create a dynamic homepage block displaying personalised discounts based on hotel and booking dates.
{{quote-5}}
The result:
- Test launched in 4 hours
- Multiple variations tested rapidly
- Approx €2,000 saved
When the data disproves a promising idea
Not every experiment delivers positive results, and that's exactly what makes a mature experimentation programme effective.
Best Western® Hotels & Resorts France experienced this with an AI-generated audio guide added to hotel pages. The goal was to enrich the customer experience with an audio presentation of each property.
The concept seemed engaging and innovative.
The results told a different story.
Exit rates increased by 28%.
Using Contentsquare analytics and session replays, the team quickly identified user friction and ended the experiment before it could negatively affect performance.
It is a clear reminder that intuition alone is never enough. Data should always guide decision making.
The golden rule: structure before you automate
For Ecaterina Pisica, the message is clear:
Don't automate chaos. Si vos données et vos process ne sont pas propres, l'IA ne fera qu'amplifier le désordre. Nettoyez d'abord, structurez et automatisez ensuite.
PBX accelerates execution, but success ultimately depends on the strength of the experimentation strategy behind it.
Moving faster without a clear direction simply means moving faster in the wrong direction.
Conclusion
With PBX, Best Western® Hotels & Resorts France has reached a new level of CRO maturity.
Instead of being constrained by its experimentation backlog, the team is now in control of it. Ideas move from concept to live experiment within hours. More tests lead to more learning, and more learning drives continuous optimisation.
More importantly, the organisation has embraced a genuine experimentation culture, where failed tests are seen as valuable opportunities to learn rather than setbacks.
The transformation is about more than speed.
It is about giving teams the freedom to experiment more, learn faster and build sustainable, measurable growth.

Prompt-Based Experimentation

What we learned from 1,000 prompt-based experiments
Too many ideas were stuck in the backlog. We had the ideas, but not the speed to execute them. With a small team, external development resources and an ambitious roadmap, many tests simply had to wait.

PBX removed our development bottlenecks and made collaboration much smoother. We can now experiment faster, learn continuously and create real business impact.

Even when a test doesn't work, we haven't wasted time or development resources. We learn and move on.

We were very sceptical about this test because it seemed so subtle. Yet it worked.

Variations are optimised much faster, mock-ups become live experiments almost instantly, and we can explore several ideas in parallel.

20 min
of preparation
0
technical resources
~400 €
savings
20 min
of preparation
0
technical resources
~400 €
savings
Power experiments with user insights
Connect Mixpanel to Kameleoon for real-time insights and cohort-based experimentation.
