Fewer dead ends: personalised recommendations for Visit Wales, built on its own data

The challenge

People come to Visit Wales looking for inspiration, but they want very different things. Families might want child-friendly activities. Solo travellers might want adventure, or a quiet place in the countryside. If the site can’t respond to those differences, visitors lose interest quickly.

Visit Wales already collected rich data about how people used the site. The question was how to turn that data into recommendations that helped each visitor, without taking control away from the editors who run the site.

Our approach

We started with the data. Using Drupal as the platform, we channelled behavioural information from Google Tag Manager and Google Analytics into BigQuery, which gave us a clearer picture of how people moved through the site.

Patterns emerged quickly. Families often explored accommodation and activities together. People looking for events followed distinct paths from inspiration to booking. And new visitors behaved differently from returning ones.

Working with iCrossing, we used those patterns to build a machine learning model that recommends the next best step for each visitor.

What we built

• Recommendations based on what a visitor is looking at. Browse family-friendly accommodation, and the site suggests family-friendly activities nearby.
• Recommendations for new visitors, guided by the journeys of similar past visitors.
• Recommendations at the bottom of product pages, for events, places to stay and things to do, so a page that used to be a dead end now leads somewhere.
• Integration through Drupal components, so the recommendations fit the site’s design and editors keep control of content and publishing.

The outcome

Event pages saw the clearest change after the recommendations went live. In August 2025, they recorded [5,000 / nearly 6,000] more visits than in August 2024, a 30% increase. The system now makes around 43,000 recommendations a month.

Return visits rose too. The site’s DAU/MAU ratio, which shows how many monthly users also come back daily, rose from 20.7% in August 2024 to 21.6% in August 2025.

This is the start of more personalisation work with the Welsh Government. The next phase brings the same approach to article content, so visitors are guided to their next story or travel guide.