Discover more of Wales with personalised recommendations

The Challenge

The site received millions of visits, but too many journeys ended prematurely. Families exploring accommodation rarely discovered nearby child-friendly activities. Solo travellers seeking adventure were served the same generic recommendations as coach tour groups. Visit Wales held years of behavioural data, but it sat unused, a missed opportunity to guide each visitor to what they’d genuinely value.

Research from Accenture shows 91% of consumers are more likely to engage with brands that provide relevant recommendations. Yet most destination sites still serve one-size-fits-all content.

The strategic question: how do we turn the data we already collect into journeys that feel personal, without overwhelming editors or compromising editorial control?

Our Approach

This required more than technical implementation — it needed genuine partnership. Box UK embedded with Visit Wales and their agency iCrossing, working as one cross-functional team rather than as separate vendor and client.

In practice, that meant:

• Joint workshops to map user journeys and define what ‘relevant’ meant for different visitor types
• Collaborative iteration on the recommendation logic, testing assumptions against real behavioural data
• Knowledge transfer throughout, ensuring Visit Wales could evolve the system long after go-live

The technical strategy followed from this foundation: channel behavioural signals from Google Analytics into BigQuery, identify patterns in how different visitors moved through the site, then build a machine learning model to predict the ‘next best step’ for each user.

What We Built

The system works quietly in the background, adapting to each visitor in real-time.

Browse family-friendly accommodation, and the site suggests family activities nearby, not adventure sports or romantic getaways. Arrive as a first-time visitor, and recommendations draw on the journeys of thousands who came before you. Return for a second visit, and the system picks up where you left off.

Recommendations appear naturally at the bottom of every product page: events, places to stay, things to do. What was once a dead end becomes an invitation to keep exploring.

For Visit Wales’ content team, nothing changed visibly. The recommendation engine integrates through existing Drupal components, preserving editorial control while giving every visitor a more tailored experience. Editors curate and publish as always; the system handles personalisation behind the scenes.

Delivery

We launched the recommendation engine in early 2025, navigating the complexity of integrating machine learning into a live, high-traffic site without disrupting existing user journeys or editorial workflows.

The constraints were real: legacy data structures that needed cleaning, integration points that required careful orchestration between Drupal, Google Analytics, and BigQuery, and the need to keep the system transparent enough for Visit Wales’ team to understand and maintain.

Throughout, knowledge transfer was built in, not as an afterthought, but as part of the working model. Visit Wales now owns the logic and can evolve recommendations as their strategy shifts, without dependency on external support for every change.

The Outcome

The impact was immediate and measurable. In August 2025, Event pages recorded 5,000 additional visits compared with the same month in 2024, a 30% year-on-year increase! The system now delivers 43,000 personalised recommendations every month.

But the real shift is in loyalty. Google’s DAU/MAU metric, the ratio of daily active users to monthly active users, rose from 20.7% to 21.6%. That single percentage point represents thousands more people choosing to return, finding enough value to come back again and again.

For Visit Wales, this means stronger engagement with potential visitors, more completed journeys from inspiration to booking, and better data to demonstrate value to Welsh Government stakeholders and tourism businesses across Wales. Personalisation has moved from a ‘nice to have’ to a competitive advantage.

What this means for your organisation

This story will resonate with destination marketing organisations, government digital teams, membership bodies, and cultural institutions, any organisation with diverse audiences and rich behavioural data they’re not yet using to guide individual journeys.

The common thread: you know your users are different, but your site treats them the same. You collect data, but it doesn’t shape what people see. You want to personalise, but you’re wary of black-box solutions that leave you dependent on vendors.

If that sounds familiar, let’s talk about how we can turn your data into journeys that feel personal, while keeping you in control.