FranceTravel and aviation2018 – ongoingTUI France
Machine learning scores for display ad targeting
TUI France cut display cost per sale 7.5 times after a machine learning model scored visitors by purchase intent
Problem
TUI France, a subsidiary of the TUI Group and a leading French tour operator, wanted to cut the cost of its ad spend in a competitive travel market. Its CRM data and its Google Marketing Platform data were stored separately and in different formats.
Approach
With the partner Artefact, TUI France joined both datasets in BigQuery using the CRMint tool. Data scientists built a machine learning model on AI Platform that separates website visitors who are browsing from those likely to buy, and retrained it every week. TUI France used the scores to prioritize bids and to build lookalike audiences for display campaigns.
Results
Takeaway
The model was one part of a wider data platform, so the savings cannot be split between the scoring and the data consolidation. TUI France planned a recommendation engine as the next step.
Categories
- Size
- Large enterprise
- Industry
- Travel and aviation
- Business function
- Marketing and sales
- Innovation type
- Process innovation
- AI technology
- Predictive analytics and ML
- Adoption
- Partner co-development
- Evidence
- Company's own claim
Sources
Source summaries
Google Cloud's customer story describes how TUI France and Artefact combined CRM and marketing data in BigQuery and trained a model to predict which visitors would buy. It quotes TUI France's head of digital analytics on display cost per sale falling 7.5 times and customer acquisition cost falling 29%. The page carries no date; the Internet Archive first saved it on November 2, 2020.
Full details are in the original sources.
