Case Study | Data Analytics | Hospitality Industry
Problem Statement
A chain of luxury hotels and agency catering to customers around the world was struggling with low return on investment (ROI) from its digital marketing campaigns, achieving only a 1.5x ROI compared to an industry benchmark of 3x. The marketing team lacked insights into which channels and campaigns were driving bookings.
Solution
A Travel agency implemented a data analytics solution focused on marketing attribution and campaign optimization.
Technical Approach
- Data Collection and ETL Process
- Extracted data from digital marketing platforms (Google Ads, Meta Ads), website analytics (Google Analytics), and booking systems (MySQL database).
- Built an ETL pipeline using Apache Airflow to ingest, clean, and transform data, including handling missing data with median imputation for numerical metrics.
- Stored processed data in a data lake using Google BigQuery for scalable querying.
- Attribution Modeling
- Built a multi-touch attribution model to assign credit to various marketing touchpoints in the customer journey.
- Calculated conversion paths and attribution weights for channels such as search ads, social media, email, and organic search.
- Predictive Analytics
- Developed a model to predict the likelihood of booking, incorporating features such as customer demographics, channel engagement, and seasonality.
- Tuned hyperparameters and achieved an accuracy of 88% on the test set.
- Campaign Optimization
- Created a real-time dashboard using Looker Studio to monitor campaign performance metrics (e.g., cost per acquisition, conversion rate) and recommend budget reallocations.
- Built an optimization algorithm to maximize ROI subject to budget constraints.
- Personalization and Deployment
- Used insights from the predictive model to tailor marketing messages to high-value customer segments, integrated with the marketing platform via APIs.
- Deployed the predictive model as a REST API using Flask, hosted on Google Cloud Run for scalability.
Tech Stack
- Data Storage: Google BigQuery (data lake)
- Data Processing: Apache Airflow (ETL)
- Programming: Python, R
- Libraries: pandas, numpy, lightgbm, PuLP, ChannelAttribution (R)
- Visualization: Looker Studio
- Deployment: Flask (API), Google Cloud Run
- Integration: Adobe Campaign API
Implementation
- Conducted a 2-month pilot across three high-traffic properties before scaling to the entire chain.
- Trained the marketing team on using the dashboard and interpreting attribution insights.
Impact
- Efficiency: Reduced time spent on manual campaign analysis by 60%, allowing the team to focus on strategy and creative development.
- Productivity: Marketing campaign execution speed increased by 35% due to automated budget recommendations and performance tracking.
- Decision-Making: The leadership team gained visibility into high-performing channels, enabling data-driven decisions on budget allocation and channel strategy.
- Overall Growth: Marketing ROI improved to 3.2x, bookings increased by 18%, and revenue grew by 14% within a year.