Wanup had the inventory of thousands of hotels across Europe, competitive prices, a platform with real potential. But 67% of users were abandoning before completing a booking. Not because the hotels weren't good. Because the experience gave them no reason to trust it, and no architecture to bring them back.
The users who did book? They never returned. There was no loyalty layer. No rewards. No relationship. Just a transactional dead end.
Abandonment rate before booking completion.
Source: Funnel analysis · pre-redesign baselineWhen we went deeper into the data, the abandonment wasn't random. It clustered at the same friction points every time: unclear pricing, weak trust signals, no visual differentiation between hotel options, and a checkout flow that asked for too much before delivering any value.
But there was a second problem layered underneath the first. Even when users completed a booking, the platform had no mechanism to bring them back. Every returning user started from zero. No history. No recognition. No reward for loyalty.
"This wasn't a booking problem. It was a trust problem and a loyalty problem."
That reframe changed everything. We weren't redesigning a checkout flow. We were building the infrastructure to turn one-time bookers into a repeat customer base.
I joined as Lead Product Designer on a cross-functional team of 4+ people, working across the full product lifecycle from research and strategy to design system and handoff. The scope was end-to-end: both the user-facing booking experience and the hotelier-facing management interface.
The north star wasn't conversion, it was repeat booking rate. That distinction shaped every decision. A conversion-focused approach would have optimised the checkout. A retention-focused approach had to rethink the entire relationship between the user and the platform. We chose the harder path.
The temptation in a project with this level of abandonment is to go straight to UI fixes. We didn't. Before opening Figma, we ran a structured discovery phase with four research tracks running in parallel.
Deep funnel analysis using Hotjar session recordings and Mixpanel event data. We mapped every drop-off point in the existing flow: when users left, where they hesitated, and what triggered abandonment. The 67% figure wasn't a surprise. The pattern was.
User-side researchWe ran structured interviews with hoteliers using the platform. The supply side had its own friction: slow inventory updates, poor visibility into booking data, no control over how their property was presented to users. A two-sided marketplace fails if either side is underserved.
Two-sided discoveryWe mapped how Booking.com, Hotels.com, and emerging players handled trust signals, loyalty, and repeat booking mechanics. The category had converged on patterns we could challenge, particularly around how loyalty programmes were surfaced to first-time users.
Competitive benchmarkingCross-functional workshops with PM, engineering and business stakeholders to define what entered the first release and what didn't. The loyalty layer was nearly cut. The data from Track 01 (specifically the zero-return-booking problem) kept it in scope.
Stakeholder alignment
The key output from discovery was a prioritised problem map. We had three layers to address simultaneously: immediate conversion blockers (the 67% abandonment), trust architecture (signals that make users commit), and retention infrastructure (the loyalty layer that brings them back). The design system would be the enabling layer underneath all three.
The platform had accumulated years of inconsistency, duplicate components, hardcoded values, no shared language between design and engineering. A two-sided marketplace with separate user and hotelier interfaces made this worse: the same UI problem solved differently in two places, twice.
Most hotel platforms bolt loyalty on as an afterthought, a points counter buried in the account settings. We designed Wanup Rewards as a first-class surface: visible from the search results page, contextualised in the booking flow, and present in the post-booking confirmation.
The session analysis revealed that abandonment wasn't primarily a checkout problem, it was a pre-checkout problem. Users were leaving the hotel search results and the hotel detail pages without ever reaching checkout. They weren't convinced before they got there.
The system had to serve two audiences with fundamentally different mental models: users searching for hotels (goal: confident decision, fast checkout) and hoteliers managing their inventory (goal: control, visibility, efficiency). The same token layer, the same components, configured differently for each context.
Validation wasn't a checkpoint at the end. It was what shaped the final decisions. Several findings directly changed the product.
The north star moved. Users who had never returned to the platform started coming back. The loyalty layer gave them a reason. The redesigned trust signals gave them the confidence to commit the first time.
The repeat booking rate was the metric that mattered most, and the hardest to move. It required the entire architecture: the trust layer to get users to commit the first time, the loyalty layer to give them a reason to return, and the design system to make iteration fast enough to respond to what we learned post-launch.
We spent roughly 70% of research and design effort on the user-facing experience, and 30% on the hotelier management interface. In a two-sided marketplace, that imbalance has consequences. Hoteliers who found the management interface difficult to use updated their inventory less frequently, which hurt the quality of data available to users. In retrospect, I would have pushed for a more equal split from day one, and made the case with the funnel: better hotelier UX → better data quality → lower user abandonment.
Wanup Rewards was scoped relatively late in the discovery phase, it entered the project after the initial IA was already defined. That meant we were retrofitting loyalty into a structure that wasn't designed for it. If I were starting again, I would have treated the loyalty layer as a first principle of the IA, not an additive layer on top of it. The rewards touchpoints we fought to include late in the project would have been natural if they'd been designed in from the start.
The project naturally gravitated toward the pre-booking and booking flows, that's where the conversion problem lived. But repeat booking rate is driven by what happens after the first booking: the confirmation experience, the pre-stay communications, the post-stay feedback loop. We iterated on these, but with less rigour than the core flow. The 3.1× repeat rate tells me there was significant headroom we didn't reach.
Within 120 days, users who had never returned to Wanup started coming back. The loyalty layer gave them a reason to return, and the redesigned trust signals gave them the confidence to commit the first time.
3.1× repeat booking rate was the proof we'd solved the right problem. Not just a checkout. A relationship.
← All case studies