Improving booking-funnel conversion: measure before you rebuild

When conversion drops in peak season, the reflex is to look for a culprit. A central, visible, slightly older component makes an ideal one. At Homair, it was a drawer (a sliding panel inherited from the design system) used to present the mobile homes.

Rather than rebuild it on a hunch, I set up a multi-segment quantitative analysis on Contentsquare with a Product Owner, to measure its real usage before judging it. The data cleared it: exposure, scroll depth and click-through were all excellent. The friction sat elsewhere, and the investigation moved back onto the right leads instead of sacrificing a structural design choice.

The lesson is simple: in a booking funnel, never rebuild a component on a presumption. A data culture exists first and foremost to avoid breaking what works. (The full approach is in the case study The drawer case.)

A funnel that converts guides, instead of offering the most options. Total freedom tires people out and breeds doubt; a good constraint reassures.

When we designed the multi-campsite packs (building a stay across 2 to 4 nearby campsites, when only 1–2% of bookings involved more than one site), I proposed a key trade-off: tie the number of campsites to the length of stay rather than leaving it to free choice (2 campsites for 4 nights, up to 4 for 16 nights). That guardrail prevents unrealistic combinations while preserving the sense of personalisation. The product was taken up as a reference by the English Eurocamp team for their own market.

The lesson: every field you leave open in a funnel is a decision you delegate to the user. Take some of it back onto yourself, and you reduce mental load instead of conversion. (Read: Turning a commercial weak spot into a strategic product.)

In any booking funnel, some friction can’t be removed. Adding these packs to the cart took several seconds of server-side computation, an unavoidable latency, and a classic place to lose bookings.

Rather than hide it behind a generic spinner, I made it a brand moment: a Rive animation that changes with the number of campsites in the pack, filling the wait and rewarding the choice just made. That idea kicked off a broader digital branding effort I still lead today.

The lesson: an unavoidable wait isn’t necessarily a loss. Well staged, it becomes a signal of quality, and sets your booking funnel apart from all the interchangeable checkouts in the industry.

The deepest cause of an under-converting funnel is rarely a component. It’s a catalogue logic that has never been confronted with real users. Many booking sites follow marketing and inventory logic, without passing through the journey of the person actually booking.

On the full site redesign, in a company where user research wasn’t an established practice, I carried that shift: translating an art direction into a workable design system, then, on my own initiative, running a series of post-launch usability tests (19 participants, 6 scenarios) to finally confront the funnel with reality. The result: a usability score of 77/100 and a direction validated by the users tested, not just by internal intuition.

The lesson: measure the whole funnel against real journeys, not page by page against marketing targets. (Full context: Being the thread of continuity in a many-handed redesign.)

• Measure before you rebuild. Analyse a component suspected of hurting conversion instead of condemning it: data often protects a good design choice.
• Arbitrate constraints. Every option left open is mental load delegated to the user; a good constraint guides and reassures.
• Take care of unavoidable friction. A well-staged wait becomes a brand moment rather than a drop-off point.
• Break out of catalogue logic. Confront the whole funnel with real users; usability is measured, not assumed.

Booking-funnel conversion comes down to a discipline: observe, measure, arbitrate, then act once you know where the friction is.