Andy Owen-Jones - bd4travel

We started bd4travel after working in the travel industry for a long time. My co-founders and I managed the largest distributor in Central Europe, a company called TravelTainment. We ran hundreds of websites and supplied thousands of travel shops and what we noticed is that all of our clients were interested in conversion and what went into the call centre but none of them understood the customer decision-making process. We felt that e-commerce could be much better if people had an idea of what customers were after, what they wanted and why they were making certain decisions. So, we decided to try and understand digitally what customers are trying to achieve and match them with the right products and experiences. Instead of throwing many offers at them, we could try to meet their needs in real time.

Essentially, bd4travel was founded on the conviction that at that moment a lot of e-commerce was about shouting at people rather than listening to them. If we could listen to them, we could provide a better service.

The core of our offer is a profiling service. We work with direct channels, such as apps and websites as a first party data collection source by trying to build a profile of their users. After every click they make, we interpret their decision. We have now profiled over a billion customer journeys in as much detail as possible.

Our initial idea was just to provide profiling solutions, but on the way we found that most of our clients did not have very fast websites and could not react efficiently to the information we offered. We then started building products that would react to the profiles we created. For example, if you are showing someone 12,000 hotels, instead of displaying them according to price or to your commercial deal, we would sort them based on what we think is relevant to that individual. If we’re making recommendations, we will try to work out what to show that individual and, when they are ready to book, we’ll start honing on the thing they are after and give them more details on that.

While the core product is profiling, we built our business model around those auxiliary services that are driven by machine learning and hypotheses to put products in front of people in the right way.

I think the main innovation we bring is working at the level of the individual in real time. We do personalisation rather than segmentation. In this way, we can treat each person as a unique user rather than build the best website for the average user.

Our contention is that currently e-commerce is about building a site that is suitable for the average person. We are trying to make every element of the website respond to a particular individual in their current process, and not target users as a whole.

Absolutely. We have to comply with data protection legislation and always get consent. We never share data among our clients, it’s always on a first party basis, so we always work within consent frameworks. We also have to be very careful with what personally identifiable information we allow to be shown and transferred. We’re largely based in Germany, so we’re conscious from the start of the European Union laws that we need to provide anonymous data which can only be shared with the customer’s permission.

While a lot of our data is being hosted in Germany, we now also work with global partner Microsoft Azure, which allows us to host data in different locations.

When we set up in 2013, AI was not very popular and machine learning was not too interesting to people. The first couple of years we had to keep explaining these ideas and why they are so beneficial in processing data a lot quicker than humans could. This is completely different now and people are very interested in AI, but the first years were very challenging.

Another challenge was that e-commerce was mostly based on trying to find the best solution for the greater number of people. Most of the people who use e-commerce tools are trained to think in averages and segmentation and cannot conceive how to work at the level of individuals. It took a giant leap of belief for people to let AI choose what to put in front of individuals. Once they were convinced, they saw a whole new way of doing things.

A lot of the core systems that are in place require setting up hypotheses about how rules should work. If a person has chosen a particular experience three times then the system will show them the same thing next time. But the more complex products you have and the more changeable they are, the harder it is to run them with rules.

We can improve this process because we can help people move beyond segments. The AI learns what the best target customers are, then shows them a series of options and sees which ones perform best. So, we bring a new dynamic way of working for our clients. We also make automation more flexible and adaptable to the needs of customers.

It is a leap of faith to get involved because AI is still quite scary, but once you do, it becomes a mindset.