Menu

Booking.com about Data Science, Machine Learning and Artificial Intelligence

0 Comment



Thank you very much Pepijn. As one of your million of Booking.com users it is a pleasure to be here with you. What are the main drivers in your opinion for the exponential growth of Booking.com over the last decade? First of all it is important that the most important reason probably is being lucky, being at the right place, at the right time, with the right people and the right product Generally, I think wat we have learned over the past series of years is that 3 main pillars of the success are one, the broad selection, the best pricing, and the most easy to use experience. Thats all on the product and service side. And I think second our culture, which is a culture that enables really rapid innovation on any part of the business. We always call our business as ‘running a marathon’ not a sprint. It is never finished, there is always more to be done. So I think those two things combined are probably what is spinning the fly wheel at Booking.com. Which role do Artificial Intelligence and Machine Learning play in product development itself? So, I would say the two things are quint different. Artificial Intelligence is enabled by Machine Learning and Machine Learning is a sort of a winning statistical model application. So I talk about Machine Learning first. Machine Learning has a big impact on many many things we do. Whether it is how we determine our ranking on the website for the different properties that you will look at. Whether it is how we develop chatbots to deliver services to people, how we do translations automatically, to how do we optimize our bidding on Google search keywords for instance. Essentially is it just better Data Science allows us to make better decisions on every part of the business. Then when it is about A.I. we talk about applications, we have a for instance a chatbot live on Facebook in the USA, that allows people to interact with and make a booking. We also have a chat interface live for delivering services, and here we really have a lot of learning in the implementation. Ok, so related to this topic we see that technology, and Booking.com is regarded as a technology company, and Booking.com and technology in itself the trend is more about automation. But on the other side we also see the human side and the human ‘flavor’ becoming more and more important as part of the future trends. How does Booking.com deals with this difficult balance, which can be regarded as opposites? I like to answer the question on a synthetic level. The thing is what technology, automation and machines are really good at is generating data and testing a hypothesis. What humans are particularly bad at is predicting outcomes. So we really need data to tell us whether and idea will work in practice. As an example, Booking.com does thousands of A/B experiments on our website experience, on our partner experience, on our customer experience, on our employee experience, everywhere we can test, we will test things because we learn faster. What so far computers and technology has been very bad in is creating the hypothesis, is the interpretation. why is a button which now has one white pixel line around it perform better, then that same button without that pixel line? And the hypothesis is: well because it draws the attention of the mind to where there is actionability. And maybe therefore we can guide an user better. Coming up with that hypothesis and interpretation, is incredible hard for computers, Thinking of myself, when I book a hotel I see myself more and more reading the comments and opinions of the rest of the users. My question to you thinking about the future is, will Booking.com become a social website in itself? There is a long term answer and a short term answer. Today we are not a social platform. Even though there is a enormous amount of user generated content on our website, if you look at our experience today. I think only 30% or 40% of the content already comes from existing users. So in that sense we use the crow a lot and it delivers a better experience. At the same time we have not build in social virality. We do not have a social graph of our customers. We know very little of our customers actually. Typically your name and your email address and maybe your phone number. Going forward, if you look at many of the early stage investments that we are making is that Booking.com is moving, not away, but actually expanding into not only doing the accommodation, but also servicing customers with local transport… local things to do…attractions… shopping…food…restaurants …all these topics that travellers actually care about in a destination we are working on. And there we actually see that travellers helping travellers, and locals helping travellers, is a very important element of a foreigner understanding a destination. So going forward I will not be surprised if social becomes more important for our service delivery then it is today, but today there is no virality building. Ok Pepijn, thanks again for your time, it has been a pleasure Thank you very much.

Tags: , , , , , , , , , , , , , ,

Leave a Reply

Your email address will not be published. Required fields are marked *