The client has a basic recommendation engine for recommending vacation resorts to their customers. The client was concerned that the customer experience on the site was unsatisfactory and not really helping members discover experiences that would appeal to them, which in turn was impacting member bookings and experience. They desired an independent assessment of the existing recommendation engine and identify enhancements that would improve customer experience and thus increase bookings.
This was an advisory engagement where we engaged with the client over a period of 4 weeks to deliver an assessment of the current recommendation engine including gaps and remediations/ways of improving.
We started by framing the objectives that the recommendation engine should meet, which included satisfying their general preferences (e.g. “prefer beach vacations”) while also satisfying their current intent (e.g. “right now, I’m looking at ski vacations”). A key challenge was overcoming sparse data (the average member booked less than once a year).
The study delivered recommendations for improvement along 4 dimensions:
Evaluation Metrics – what are the right metrics that align with the stated objectives?
Behavioral Data – Are we capturing and making effective use of customer activity on the website?
Resort Data– Are resort attributes being used to identify and recommend similar resorts in the area?
Algorithm – How does it effectively balance the tradeoffs and handle sparsity? How well does it align with customers to make their choices starting from broad exploratory to narrow focus in how they decide on a resort?
Dr. Rohit Lotlikar is a highly experienced machine learning scientist and brings nearly two decades years of varied experience in building data driven intelligent business solutions.
Dr. Lotlikar has B.Tech. and M.Tech degrees in Electrical Engineering from IIT-Bombay and a PhD in Machine Learning from the University of Cincinnati.
His professional interests and expertise includes investigating and exploiting patterns in data and human behaviour to build data driven decision making/support systems, particularly as applied to understanding customer behaviour, digital marketing and workforce optimization. His experience spans working at startups developing machine-intelligence based products, developing innovative technologies at IBM Research and most recently in management consulting as a director in the analytics practice at Ernst & Young.