Price Recommendations - Driving Revenue Strategy Using Machine Learning
Pricing in hotels can result in a lot of optimisation given that there is limited inventory to sell each day. This session focuses on how Treebo developed an automated machine learning based pricing engine within 2 months and scaled it up in next 6 months to recommend real-time prices for 400 hotels. This resulted in ~26% improvement in booked revenue, 30 days in advance.
If you want to run a profitable hotel chain, price is what you can play the most with to optimise revenue and profits.
Treebo hotels is a technology powered hotels startup that ties up with hotel partners with a value proposition of higher revenue for them and a better quality for the guests. They wanted to develop a system which could optimise revenue for every hotel by recommending optimal prices. A system that could automatically keep changing prices throughout the day, based on multiple factors like competitor prices, seasonality, web traffic, time remaining to sell, etc.
Pricing for hotels is more complex than pricing for a regular item like a shoe because it involves a booking date and a stay date. So on a particular day, you can set different prices for different future stay dates. Doing this manually is sub-optimal and prone to errors. So, you would need a system that can make changes multiple times a day for multiple future dates.
While existing products can be helpful in kickstarting an automated pricing system, they would come with some constraints. This session will throw light on the pros and cons of building your own system rather than subscribing to a third party product.
Shikhar will talk about how they came up with a minimum viable product in just two months and scaled it up to complete pricing for 400 hotels within the next 6 months. This helped the company in booking 26% more revenue, at least 30 days in advance which is wonderful for the business. They’ll also cover the challenges faced like sourcing of market data, building data pipelines and convincing business teams to rely heavily on their solution and how they dealt with them.
Intended audience: Product managers, Data scientists, Revenue Managers, Business leaders
Key takeaways: How to set up your own pricing engine, how to quickly scale-up a data science project for the benefit of business
Brief understanding about the business
Previous state and desired state
What did we build?
The journey of building the new system
Challenges faced while building and how we tackled them
Success Metrics and Impact
Shikhar is Product Manager for Pricing at Treebo hotels. In last 3 years at Treebo, he has played multiple roles including business and operations.
Previous to that, he was part of Flipkart. There, he was part of the monetisation business and played key role in scaling up that business. He was part of the team that launched a completely automated new product called Product Listing Ads.
Before that, he was an analyst at Mu Sigma, a major player in providing analytical services to fortune 500 clients. He worked on a couple of projects there involving logistic regression modelling and customer loyalty analytics.