Case Study
The Results
- +270% Booking funnel entries (conversions)
- 6.33x ROI on annual media-driven revenue
- -34% False negatives in purchase prediction
The Problem
Premium holiday packages involve long, considered decisions, with customers returning to the site many times before they book. These multi-session journeys made it hard to tell which visitors were genuinely likely to buy, so investment couldn't focus on the prospects most likely to engage, enter the booking funnel, and convert.
The Solution
-
PM Dragonfly partnered with the client to connect predictive modelling with media activation, turning on-site behaviour into a signal of purchase intent and using it to steer programmatic investment. The work focused on three key areas:
-
1. Predict Purchase Intent
Rebuilt the on-site Quality Score model to identify which visitors were most likely to convert, using behavioural signals from across the wider business portfolio. Launched in the UK, the improved model reduced false negatives, is ready to scale across global markets and can be used as a signal across wider digital marketing.
-
2. Strengthen Measurement Foundations
Improved site tagging and refined Campaign Manager 360 Floodlight events, upskilling media teams along the way. This simplified reporting, removed duplicate counting and excluded customers with active bookings, so spend focused on genuine prospects. New dashboards replaced manual reporting on booking flow entries and other key stages.
-
3. Activate Through Custom Bidding
Developed a new DV360 custom bidding algorithm built on the improved Quality Score and customer data, then A/B tested it against the existing algorithm alongside the client's media agency. The results were clear early on: more engaged traffic, more booking funnel entries and more conversions across channels.
-
- The client now has a scalable, data-driven way to prioritise high-value prospects and invest media where it is most likely to drive bookings.
