Dates that stay put. More direct bookings.

Strong interest, repeated searches

Company wanted more guests to complete bookings on its own website. Visitors were checking availability, but mobile Safari journeys showed repeated date searches and weaker completion. The Director of Direct Revenue initially considered a more prominent direct-booking incentive. The Digital Product Manager wanted to understand why guests were restarting a task they had already completed.

The break was between two parts of the journey

FullSession funnels narrowed the investigation to the move from the hotel website into the booking engine. Within the permitted booking-engine capture, session replays showed that a rate-plan change sometimes replaced the guest's selected dates with defaults. Lift AI elevated the issue on the strength of the journey it affected and its estimated commercial impact.

The team reproduced the handoff defect when a guest returned from a room detail panel and chose a different rate. The displayed price then referred to the wrong dates. Some visitors corrected the dates; others restarted or left. Guests also legitimately compare stays, so the team treated repeated searches as a symptom to test, not a count of errors.

Preserve intent instead of adding an incentive

The booking-engine vendor corrected date persistence and added a visible stay summary before room selection. The Digital Product Manager verified the behavior across the hotel site and the separately instrumented booking domain. Payment pages remained outside replay capture. The work took two weeks, including vendor release coordination; FullSession identified and validated the problem while the vendor shipped the fix.

A concurrent comparison during the same trading period

Company ran a 28-day visitor-randomized test for mobile Safari visitors who reached availability results. The control arm contained 14,937 eligible visitors and the treatment arm 15,062. Confirmed bookings within seven days increased from 1,302 in control to 1,435 in treatment: 8.72% to 9.53%, or 9.3% relative improvement.

Measured against the control rate, the treatment cohort produced about 122 more bookings. Fourteen-day cancellations were 12.0% in control (156 of 1,302) and 11.8% in treatment (169 of 1,435). Average booked stay value was £410 in control and £408 in treatment. The team used booking conversion as its primary outcome.

How booking completion was measured

Both variants ran across the same properties, dates and rate inventory. Persistent experiment assignment was carried into the booking engine, and server-side confirmations supplied the outcome count. The last entrants received seven days to convert and a further 14 days for the cancellation check. The result applies to the eligible mobile Safari journey. Completed stays and final revenue were not yet available. Rates and relative improvement are rounded.

What changed in the revenue meeting

Company rolled out the fix and added date persistence to its booking-engine regression checklist. Lift AI's opportunity estimate had put the issue on the agenda; the experiment supplied the measured result. The Director of Direct Revenue gained a practical alternative to offering another incentive: remove the obstacle that prevents an interested guest from completing the stay they selected.