Booking.com packages with Can Ulker
This is a newsletter about big bets, explored through conversations with the product leaders who worked on them. For this one, I chatted to Can Ulker, a former product manager at Booking.com, about a bet they took on building a product to offer package holidays. Founded in 1996 in Amsterdam, Booking.com has grown from a small Dutch startup to one of the world’s biggest travel companies, which is famous for its data-driven approach to product development.
About 4 years ago, Booking.com decided it was time to expand beyond accommodation, and formed a business unit called New Products to start taking some big bets. They decided to build this as a completely separate organisation from the accommodation business to avoid causing distraction. The plan was that only once new products were mature enough would they be integrated with the existing accommodation product.
The overall company strategy was to provide the ‘connected trip’ - so if you were looking to buy any travel product, Booking.com wanted to be the one place you needed to go. Some of the bets that naturally came out of this strategy were a flight product, a rental car product, attractions, tour groups and more.
Can Ulker was a junior PM in the New Segments organisation. One idea that was floating around was a product for package holidays. I ask Can who came up with this idea and got it started, but he says it didn’t work like that: lots of people had the idea, because after deciding to go into new areas of travel, people naturally thought - why not combine multiple areas into a package?
In fact, the idea had been around for so long that the strategic partnerships team had already built a basic product around it for the European market. They’d created a partnership with Lastminute.com, which was essentially a revenue share in which Booking.com would sell a white label version of Lastminute.com packages for flights and accommodation. This was a typical approach, Can tells me: ‘Lots of things happen at Booking.com through small, simple experiments that give quick insights into what actually works.’
Having started with the partnership, now the company wanted to invest more deeply in the connected trip vision. They knew that if they wanted to sell more, and more diverse, packages, they couldn’t do it through a third party. So Can was assigned as the PM for packages, and tasked with working out what the value of this product would be to the company. Initially, the team was just him and a designer, and they spent their first few months in research to create a business case.
They started by trying to understand how the package holiday market works. They spoke to Lastminute.com to understand how a package product is built, and they visited physical locations for tour operators like Thomas Cook. On these outings Can and his designer even pretended to be a couple planning a honeymoon, to get a realistic experience of what booking a package was like. Armed with all this information, they created some mock ups and usability tests, and fleshed out how the users would experience the value proposition.
In parallel, they ran a few tests on the existing package offering with Lastminute.com. One thing they wanted to know was how much users care whether they’re using a white label product or not, so they tried removing the Booking.com branding. They also launched other experiments to optimise the ranking of the product on third party sites. It was a good way to learn, but was slow going because they were dependent on Lastminute.com to build each test.
Having conducted all this research, Can finalised his business case. As well as articulating the value proposition for users, he set out why they should do it as a business and why now, including a detailed financial model. He worked with commercial experts in the company to establish the incremental value of a package product. Flight booking itself is a low margin business, but it’s an enabler of travel, and combining flights with accommodation in a package opens up a segment of users who would never buy them separately.
A few months after the project started, Can pitched the business case to his own director and other senior leaders and got a lot of interest. They connected the dots with work going on in another team, called Resorts, who had also identified an opportunity to go after packages. The decision was taken that the teams should join forces and build a trial together.
The Resorts team had arrived at the problem from a different angle. They found that many of the users who were interested in staying in an all-inclusive resort hotel also liked to buy through packages. These were users with a preference for convenience over control and customisation. The Resorts team focused on their largest market, the US, and decided they wanted to build a Package trial there.
The two teams didn’t have exactly the same goals and hadn’t defined the user problem in the same way - Can was focused on the global opportunity, for example. But there was enough there for them to design a trial that worked for both. They picked a popular travel corridor, US to Cancun, and started building an MVP.
The team was now 2 PMs, 2 designers and 6 senior engineers. While they didn’t consider using Lastminute.com again, for the sake of speed they did partner with another company within the Booking holding: Priceline. This meant that it was possible for their small team to ship an MVP in 3-4 months working with APIs and allowing the heavy lifting to be done behind the scenes by Priceline.
The hardest part of the engineering was the post-booking experience, but Can knew from his research how critical this was for people who book package holidays. In fact, they often considered it the main value add of booking as a package. If your flight is cancelled and you booked it separately, you alone are responsible for figuring it out, but if you booked it as a package, the package provider is responsible for finding you a new flight. Since the team were building an MVP, they were constantly debating what to include and what to leave out, but they concluded the post-booking service was something that couldn’t be cut.
After just a few months they were able to launch their MVP as a test. They controlled access to it through adding an entry point to an existing landing page for resorts in Cancun. This page was not getting a huge amount of traffic - perhaps 200-300 users a day - so it felt safe to start a test, and they spent on PPC advertising when they wanted to increase the volume of usage.
The first question the team asked of their MVP was whether consumers wanted it, and the success criteria they’d defined for this was the number of people hitting the ‘book’ step. Their first booker came after 3000 visits. As a benchmark, they used their previous experiments with Lastminute.com, but to the team’s dismay, they fell way below that figure. In fact, conversion was about 10 times poorer. So they started diving into what was wrong.
Some research turned up the main answer: the pricing was off. Users could find the exact same US > Cancun packages on competitor websites for cheaper. They also identified problems in the user experience that were tripping customers up and started iterating to improve these. The conversion rate picked up and they got more sales, so they opened up a few other destinations including the Dominican Republic.
At this point Can left the team, as they’d decided it didn’t need 2 Product Managers. But he kept following the fortunes of the Packages trial from his new team inside Booking. Despite initial positive signs of growth, the product stalled. The team knew that what they had built was not scalable beyond the US, because their partner Priceline only supported US travel. But when they started looking into other markets, they found a huge amount of complexity waiting in different legal issues around travel, such as the European Package Travel Directive. These were insurmountable complexities of building an in-house solution.
Can points out that he had anticipated some of this complexity in his original business case. ‘One of my recommendations had been: do not do this,’ he laughs. ‘Or at least, don’t do full packages but rather upsell flights to people in the accommodation booking funnel, so you avoid the complexity of the post booking experience.’
I ask him what the experience of working on this big bet taught him. He tells me his main learning was not to spend so long in the research phase, but get going with building the real thing. Throughout his time at Booking.com, he continued to receive interest and compliments on the business case, but he feels he didn’t get enough momentum going by actually executing in those early months. Nothing beats the insights you get from real users trying out a product.
My top takeaways:
Do not underestimate the end to end experience of a product. You’re not just building an up-front journey, you have to follow through and build for the cases when things go wrong.
Keep your eye on the competition: if you’re entering an existing market, users will always judge your product (price, speed, convenience) against the nearest alternative
MVPs are great, but if markets differ significantly for your product, be aware that your learnings and tech may not scale at the rate you want them to. Manage expectations accordingly.
Forwarded this email? Subscribe here. Enjoyed this email? Forward it to a friend. And do hit reply if you have any feedback or bets you’d like me to cover.