Scaling self-service campaign creation at Booking.com
2023 to 2024, Booking.com
I led design across Booking.com’s merchandising platform, connecting campaign patterns, creation tools, analytics and guidance to make self-service work at scale. During my time on the team, adoption doubled to 172 monthly internal users. By the end of 2024, the platform supported 405 campaigns across 82 use cases, with our PM reporting €80M from campaigns run through it.
The Black Friday campaign, from platform configuration to the experience travellers see.
Making campaign creation truly self-service
A campaign often started with a Slack message. Teams needed help with platform access, patterns, placements, content and measurement. Some teams also needed help trying new creative, like video, cinemagraphs, modified or entirely new patterns.
I worked on the patterns, tools and guidance together so teams could find and use the capabilities or next best steps without waiting for someone on our team.
Shipping 13 reusable patterns from real campaign needs
New patterns could only ship when they had a real campaign use case. I used each brief as an opportunity to meet a team’s immediate needs while building something others could reuse. Through co-creation and experimentation, the library grew to 13 patterns, including heroes, carousels, campaign cards, banners and alerts. Detailed guidance helped teams create consistent, accessible, on-brand campaigns without a designer checking every variation.
Turning team requests into reusable patterns through workshops
I ran workshops with marketers, designers, PMs and engineers to understand new campaign needs. I brought proposals to discuss trade-offs, then developed directions for experimentation that could also become reusable patterns.
The Genius signature adapts an existing campaign pattern for loyalty-specific content.
Extending a shared Property Card for campaigns
As an example, I worked with Homescreen designers on a transactional card that accommodated a range of property attributes and features. We added a recommendation-specific footnote to further extend the pattern to our merchandising use cases.
A shared Property Card brings together property details, pricing and contextual information such as a traveller’s last-viewed stay.
Making simple campaigns fast and complex campaigns manageable
Across more than 20 product releases, I turned feedback into platform improvements: tabular views, search, filtering, multi-region and platform configuration, ownership and notifications.
Our users ranged from engineers launching a single banner to marketers running hero campaigns to product teams managing complex campaigns. Progressive disclosure and strong defaults supported this range of needs; a simple campaign could be configured in three steps.
Audience and experiment settings sit alongside a live preview within the campaign creation flow.
Helping marketers understand campaign performance
Marketing teams were used to tools such as Meta and Google Ads, while our analytics assumed technical knowledge. I redesigned the experience to support how marketers assessed campaign performance.
I introduced a funnel from audience to clickthrough, hourly and daily comparisons, and controls for isolating relevant metrics as well as comparative metrics. With clearer labels and contextual explanations for unavailable comparisons, teams could understand the data without relying on an engineer.
Bringing five sources of guidance into one hub
I centralised five sources of truth into a shared documentation hub covering platform access, patterns, content guidance, experimentation, help and escalations. I built relationships with teams responsible for different surfaces to understand what worked on their pages. Technical teams needed to maintain and contribute to the pattern library, so I designed a reviewable, version-controlled structure aligned with the design systems team’s documentation standards.
Rebuilding stakeholder trust through visible progress
Some stakeholders remained frustrated as the platform improved. I wrote monthly updates reaching around 1,000 people, connecting their feedback to shipped changes and inviting further input. These updates helped rebuild relationships and find new campaign partners.
Building user advocacy across engineering
I partnered with our researcher to interview platform users and record them struggling with key tasks. We showed the recordings to engineers and asked them to identify the takeaways themselves, rather than presenting synthesised findings. Hearing users’ frustrations directly helped engineers understand the problems behind the design roadmap and become advocates for addressing them.
We were so impressed by the user advocacy that we documented how to run the sessions in a guide used by 90+ teams.
Proposing an architecture to reduce days of setup to one hour
In one example, a complex Genius campaign required 80 instances, each configured, submitted and monitored separately. Teams reported days of manual work, including repeatedly uploading assets and cross-referencing complex segmentation and campaign information in Excel.
I proposed a shared campaign architecture so teams could configure placements together rather than manage each instance separately. They could organise placements into nested groups, share assets and settings, assign experiments and review performance at each level.
Based on the manual work reported to us, we estimated that the proposed workflow could reduce these setups to around one hour.
Unshipped proposal: Genius placements across platforms are organised within one campaign, with group context and change history visible.
I worked closely with our writer to clarify complex interactions, including creating nested groups, expanding content across different patterns and duplicating campaign instances. We surfaced explanations alongside the decisions they supported.
The team approved the proposal. However it did not ship: after I left the team, the PM and engineering manager also departed, and the project moved into maintenance mode.
Unshipped proposal: creators choose whether to share content across placements and organise them into a group.
Outcomes and adoption
By the end of 2024, the platform supported 405 campaigns across 82 use cases. Our PM reported €80M from campaigns run through it. Platform adoption doubled during my time on the team, reaching 172 monthly internal users.
I saw less routine hand-holding, stronger stakeholder relationships and more mature guidance as the platform expanded. These were shared platform outcomes, supported by work across the team.