Challenge
The homepage search tool was the primary starting point for product discovery on the Fred Olsen Cruise Lines site, but its design was holding users back rather than helping them. The interface relied on two sequential dropdowns — one to choose which filter to search by (e.g. destination, ship, date), and a second, visually identical dropdown to choose the specific value within that filter. This forced users through an unintuitive, single-path decision tree with no way to combine filters.
Session recordings and analytics showed the impact clearly: a high proportion of searches were being submitted empty, with users abandoning the flow before navigating to a product page. The tool wasn’t surfacing the filter options guests actually wanted, and there was no way to explore combinations (e.g. destination + date range) in one step.
Process
Research and Analysis
- Analytics review: I identified where in the search flow users were dropping off, and how many searches were being submitted with no filters applied.
- Session recordings: I reviewed real users interacting with the dual-dropdown pattern to understand where confusion originated.
Design
- I reworked the tool to surface the most popular filter options upfront in a single, intuitive interface, allowing options to be stacked and combined rather than chosen one at a time. I added live result counts and clear indications of non-combinable options, so users could see the impact of each filter as they applied it.
Validation & Iteration
- I validated the initial concepts through on-site surveys triggered after search interactions, confirming the new pattern reduced confusion and supported combining filters naturally, and iterated where needed.
Delivery
I shipped this as an A/B test through the Experimentation programme I rolled out and lead, allowing performance to be measured against the existing design with real traffic before a full rollout decision.
Solution
A redesigned search tool that:
Results
+20%
Active users reaching product pages
+16%
Search submission rate
–15%
Exit rate from search results
–75%
Reduction in unfiltered searches (21% → 5%)
–38%
Reduction in filter usage on results, indicating higher relevancy
+£1.9m
Modelled incremental annual revenue
Key takeaways
- A high volume of empty searches is rarely a “user error” problem — it’s usually a sign the interface is asking users to make decisions in the wrong order.
- Letting users combine filters rather than forcing a single linear choice significantly increased both engagement and result relevance.
- Running the change as a controlled A/B test through the Experimentation programme gave a robust, business-ready case for rollout, rather than relying on qualitative signal alone.
