Booking.com · 2020-2021 · UX Designer

Retiring the tablet stack

Tablet ran on its own legacy site: separate bugs, legal exposure, and traffic down 8.5% year over year. I self-started the project to deprecate that codebase and run tablet on the mobile stack. I built the business case, aligned stakeholders, recruited the engineers, and led the migration under a strict no-tablet-only-components rule.

Impact

  • Conversion moved from −2.5% to −0.83% against control
  • Room selection +17%; map usage +161%
  • Tablet codebase fully deprecated, freeing four teams' roadmaps
  • Outstanding legal compliance issues resolved as part of the migration

Problem

Maintaining a separate tablet stack slowed delivery and increased risk. Tablet traffic was down 8.5% year over year. Early tests showed the mobile layout did not translate well to larger screens. Conversion on tablet lagged and key interactions dropped, so a straight swap risked making outcomes worse.

Tablet migration home page before and after

Solution

Replace the tablet site with the existing mobile experience, then close gaps with targeted experiments. I partnered with engineering to add analytics for filterable events like room selection and map interactions. We ran multi-variant A/B tests and shipped in small iterations, tuning layout density, tap targets, and control placement for larger screens. We kept everything within the mobile design system to eliminate rework and avoid new maintenance debt.

Tablet migration search results before and after

Impact

Conversion moved from a −2.5% drop to −0.83% versus control, while core behaviors improved. Room selection increased 17% and map usage rose 161%. The tablet codebase was fully deprecated, freeing four teams’ roadmaps, and all future mobile updates now apply to tablet by default. Legal compliance issues were resolved as part of the migration. We measured success through experiment results and new analytics events tied to conversion, room selection, and map usage, and we documented the trade-offs so the business could make an informed call on rollout.

Tablet migration detail page before and after

Learnings

Progress can mean removing what is broken, not adding more. Targeted cleanup reduces risk, lowers costs, and improves outcomes.

← All work