Science in hospitality? Talk to these guys...
Welcome to the second edition of Meet Mews Partner Devs, this time with the guys behind Pace who aren’t just “anyone”. A couple of years ago, when we created an open API for anyone to integrate with our application, Pace helped to shape it, and we’ve seen them grow and push the limits of hospitality using data.
Who will you meet today? Lukas, the Product Lead, who’s an entrepreneur at heart and, in the early stages at Pace, had to wake up at night and pretend he was the computer running ETL scripts manually. Also, I talked to Matt, the CTO, who joined Pace last year when they were already settled and who’s now on a mission to bring the technology to another level. When the conversation went too technical, I had Vita, our backend developer who was ready to save me.
Enjoy the read!
Hospitality is our home ground
When it comes to the origins of Pace, I read two dates: September 2016 and March 2017. Which one is the correct one? Because the first email conversation I had with you was in March 2017…
Lukas: Originally, Pace wasn’t started as a hospitality company. We wanted to solve the problem of pricing no matter the industry. It’s such an interesting concept as it’s something that influences how people purchase and it influences the margins for how much you are selling. Essentially, it’s maximizing the economic outcome for everybody.
What was the trigger for the founders to start a company with such focus?
Matt: I know the answer for this one.🙂
The germ of the idea came from one of our founders, Jens. He previously worked at iZettle, a large fintech company, and was struck by how differently their customers handled transaction fees: while some passed it all on to their customers, others had to swallow them themselves. He found this difference in price elasticity really interesting and wondered if there was a way to work out the perfect price algorithmically.
So you mentioned that hospitality wasn’t the focus at the start, but it is now, right?
Lukas: You correctly pointed out that the beginning of the company was in 2016, and March 2017 was the beginning of the hospitality focus. Basically, the time between was the discovery phase and finding where the problem was big enough. So, next to hospitality we explored areas like gyms, car sharing or event tickets (there is a great podcast on the whole world of event ticketing). Each of these areas had their own challenges and complexities – think state of digitization for distribution, barriers to entry etc. Eventually we decided to focus on hospitality as it’s very similar to the airline business where pricing had a huge impact over the past years, and we felt it had the biggest opportunity.
Matt: We debated a lot, especially during the COVID times, what would be the virtue of diversifying into another vertical. For example, e-commerce retail was one of the spheres we looked into. However, considering our size and the long-term potential of the hospitality industry, we decided we want to continue our focus on hospitality as there is still so much more for us to impact.
How many PMSs are you integrated with and how many clients (hotels) are using your solution?
Lukas: We have five integrations, with Mews being our first and only one for quite a while. 😅 In terms of hotels, we have about 500 of them using our solution. It’s important to say that the majority of clients came within about six months prior to the pandemic. At that time, it felt like everything came together – but then we experienced a slowdown. However, we see more positive trends in the last couple of weeks. We are a platform that optimizes hotels’ revenues so we see that many, often larger clients, are now looking into smart solutions like ours.
Matt: We've managed to keep almost all of our customers throughout the COVID crisis, and we've been generous in terms of supporting them, because even though we were in a difficult situation ourselves, we could see hotels also having a tough time.
Small, compact and smart teams FTW
Matt, as a CTO, how often do you touch the code that impacts your partners?
Matt: Very rarely the stuff that is customer or partner facing. I am fairly hands-on and I like being in the code, but it’s usually the tools for the internal use of our product and developers. My role is about making the engineers' lives easier, and that's more about empowering them than chipping in to help with the coding.
How big is your R&D team and how is it structured?
The guys start counting their colleagues on their hands. 🖐
Matt: There are 17 of us in the whole company and about a third are engineers and a third are scientists. Science encompasses the algorithm design, research and implementation of the model. Engineering is responsible for the infrastructure, data pipelines, and the app that clients interact with Pace.
Even though we are London based, we had pretty well established remote-working practices. We have one phenomenal developer, Ondrej, who is based in Prague, and I was usually remote before the COVID crisis too.
With a small engineering team like yours, is everyone doing everything, or have you distributed the work?
Matt: This is a good question. We reorganized into cross-functional teams that are made up of both scientists and engineers not that long ago. We have a Customer Experience team which is responsible for the app that hoteliers use and its underlying APIs. Then we have the Integration & Data Engineering team. Their mission is to get hold of the data that sit in systems like Mews and bring it to our system to make it available with fast and idiomatic APIs, for consumption by the data scientists or the Customer Experience team. The third one is the Pricing Models team, who are mostly scientists. Their mission is to design and build the best possible model for price recommendations.
What’s your tech stack?
Matt: The Pace app is a single-page React app that sits on top of some REST APIs and increasingly a GraphQL layer. The backend is exclusively Python, including the APIs, microservices and the algorithms. Most of our team is focused on the backend. I would say that we have only one fully frontend engineer with one more being more of a full stack engineer.
When it comes to the data scientists I'd describe them all as full stack. They are responsible for implementing the algorithm and well as research and analysis.
There's a fairly common pattern where data scientists just prototype an algorithm in an iPython notebook and hand it over to an engineer to build it. We don’t think that's a good approach, the same way we don’t want to have separate people for engineering and operations. And let’s not forget about testing – that's done by our engineers as well.
Vita: What is your release period?
Matt: Great question. We are working towards Continuous Delivery but the truth is that our current situation is not as good as I’d like it to be. When I joined, everything was deployed as a big monolith and the process would take up to a day for one developer. Along the way we’ve added microservices so different bits of the stack can be deployed much more frequently and more easily. We have just introduced some new continuous deployment tooling and are working towards getting everything deployed through that pipeline. As a startup, we can't afford to stop the world in order to refactor, so these changes have to be added incrementally as we go. New stuff gets done "the right way".
You're positioning yourself as a company with a scientific approach towards business. Additionally, you were co-founded by an MIT doctor of Aeronautics and Astronautics. How does that affect your day-to-day operations?
Matt: There are a lot of fancy words flying around the office, that’s for sure. 🤓 I remember being fairly intimidated when I first came to the office. It takes a bit of courage to say "can you explain that to me in really simple terms", but fortunately everybody is very friendly, and happy to do so without making me feel foolish.
What’s the background of your R&D team? We love having the right balance and the dynamics between theoretical people with degrees and people who either started programming as a hobby or who started working in engineering very young.
Matt: I am a big advocate of diverse engineering teams. I think it works really well in software engineering and product management disciplines. Everyone right now has some kind of higher education but that won't be the case for long as we grow.
Pushing the limits of the API
Is the AI at Pace just a fancy word? Is it rather some form of data science or neural networks?
Matt: We don’t have any neural networks. We have some experience with reinforcement learning and we have some agent-based simulations. It's definitely more than a fancy word, although most of the algorithm would probably be better described as "Maths" than "AI".
You need to ask our Chief Science Officer, David, this question, because he will have an answer to impress you.
Lukas: What’s great and unique about him is that he has this incredible CV (with lots of fancy institutions) but also a great business understanding. He’s very capable of connecting our scientists with the needs of our product and our clients.
But to come back to your question, I believe we would characterize Pace more as a machine learning company. It is as much as a buzzword but it much better describes what we actually do. And what is unique about us is that our model adapts to every single property and even within the property when it comes to room categories, etc. And with every new piece of data we get every day, the system re-evaluates and adapts its approach for the respective property.
About Pace in their own words:
It is our belief that the future of revenue management lies within continuous pricing and automation. We enable RMs to manage prices by exception, drive revenues up and save time with our fully automated pricing and analytics solutions. Our industry-leading revenue forecast accuracy helps users plan for the uncertain future and make the most of every booking night.
Pace is one of the most connected partners we have in our Marketplace. How is it/how was it integrating with Mews?
Lukas: Without trying to please you, the integration with you guys is just so smooth and we’ve got big love for your API.
Matt: What we like about it is that it’s a modern and straightforward API where we don’t have to worry about pulling and pushing the data. And even though you guys are very easy to integrate with, we’d love to see a GraphQL based API one day.
Lukas: Right and you guys always have strong opinions about how products should be executed which translates to the way your API works and it is something we take a lot of inspiration from. And as Matt said, we can always trust the data. There are no weird corner cases that you’d set up for whatever reason. It’s always logical.
I believe it comes down to the fact that from the beginning our tech team challenged the founders when it came to certain features or common practices in hospitality. If it didn’t make sense, we didn’t do it, but it made us outcasts as we had to explain the “new normal” to clients as well as partners like you. And I’m sure you guys had many API changes requests which we always thought about and if it made sense, we added it (sooner or later) or we explained why we wouldn’t do it.
Lukas: There is one specific example where a couple of years back, we’d ask you for an extension which you denied and we accepted and worked our way without it. But then, two months later, you wrote to me: “Hey, it was added, hope you are happy now.” But we didn’t need this anymore. 😅 What I want to say by this that people will always ask for things which doesn’t mean they need them. So kudos to you for being strict as well as always taking into consideration the impact of feature requests. After all, an API is just another product that needs to be managed and maintained with customers (partners in this case) in mind.
What was the biggest fuckup or a worst moment in building your product along the way? Was it the time we had to switch you off because you were overusing our API?
Lukas: To be fair, that wasn’t only our fuckup – it was also your fuckup for not having any rate limiting (which you have now).😁 But yeah, that wasn’t the best moment, we were embarrassed as we really didn’t want to make our favorite partner angry. But you have to imagine we were at such an early stage back then. In fact, myself and a colleague of mine Philippe, we would run the dashboards for the 3-4 properties we were working with at that time, the whole ETL, manually at night mimicking automated process. So probably this was the moment when we ran a script that quoted your API too much.
It was a great learning point for us as well as we were pushed to introduce rate limiting and other API features. But the moment when I was walking around Amsterdam and our CTO called me to say that we gotta switch you off wasn’t too pleasant. 🙈 How does the onboarding process look like now?
Lukas: It’s all automated. I’d say the connection takes about one hour and the hotel basically just sets up the minimum and maximum price and they’re ready to go. There’s one thing where Mews could improve a bit and that’s the historical data. We want to know all the changes on bookings and rates. And that’s why we always try to pull so much data from you guys as we build a full history of your data on our side 😛. (Note: we are actually working on an improved action log.)
Vita: Since I was the one working on the Marketplace a lot, and it’s something that affected you a lot, I cannot not ask: what did you think about it when we released it?
Lukas: You know, this is something we have in common with you and that’s giving freedom and options to the clients. The business historically was always trying to tie you with long term contracts. Now, we have many hotels who try two different RMS on two different properties and then they decide which one is the most suitable for them. And I think this is the future, not just of hospitality but business in general, and you did a great job in paving the way. Look at how many marketplaces are out there now. 😉
From ZX Spectrum and the circus to the London tech startup scene
Matt, is it a coincidence that your previous two employers were also from the hospitality industry? It’s not so common for engineers to stick in one industry across their career...
Matt: Yeah, looks like hospitality is tied to me but it is a coincidence. Although it’s certainly helpful to have the domain knowledge and understand the business better from different perspectives.
What is your academic background? And what is your number one programming language when it comes to work or side projects?
Matt: I studied Computer Systems Engineering in Bristol, which is something between Computer Science and Electronic Engineering. For one year I was an exchange student at the University of Texas in Austin, where I returned after I graduated hoping to get a visa to settle and work there. That didn’t work out in the end and I moved back to London to join the startup community over here.
My go-to programming language is definitely Python and has been from the beginning of my career. I'm a father of two and there's no room in my life for side-projects, but I am the maintainer of a framework for microservices in Python called Nameko. It was developed at my previous job at onefinestay and is part of our stack at Pace too.
Is it true that Austin is like the coolest city in the US?
Matt: It is incredible. The city was the main reason for me wanting to settle over there. It's an amazing place with a great vibe and so much going on all the time. The biggest event is the annual South by Southwest which is a conglomerate of festivals with every possible venue hosting events – bars, restaurants, old churches, you name it – everywhere there's space for someone to perform, basically.
And even though I am trying to eat less meat recently, the Texas BBQ is certainly very delicious.
And last question before you’re off to play with your kids - why did you want to become an engineer?
Matt: When I was about 8 or 9, a friend of mine's older brother got a ZX Spectrum. I was totally amazed by it being a TV that we could control. I think from that time I knew I would be doing something with computers, I just wanted to figure out how they worked. There are lots of engineers in my family and computer engineering just sounded like a perfect career. And it is!
Lukas, now over to you! Your field of study was Technology Entrepreneurship which sounds very fancy. But what did you actually study?
Lukas: I did my undergrad in Germany and it was a mixture of computer science and business. What you’re referring to was my postgrad here in London at UCL and I always explain it as a year of trying to fail as much as you can by trying out different business ideas. You are surrounded by your teachers who are very well-connected people in the London startup ecosystem who constantly challenge your ideas and connect you with relevant contacts to test your business ideas.
During your studies, you co-founded a cool looking startup Teebly which seems to be doing pretty well. However, now you are an employed man. Is there an entrepreneurial mind in you?
Lukas: It’s definitely part of my DNA and honestly also the reason why I joined Pace. As one of the first people in the company you have an opportunity to create and shape the business. And then, with Pace, there were the people. It’s just a bunch of inspirational people with amazing experience from whom I could learn a lot from. I just really enjoy the process of starting up something new and challenging the status quo. And people who share this passion are usually people I get along with very well. I actually enjoy it so much that I also work with a company called Uprise Academy which is all about helping people break out of their corporate career and embrace their entrepreneurial mindset.
When I do interviews, I always try to prepare myself by really scanning the interviewee properly beforehand. With you it wasn’t that difficult to find out that you also had a career in circus. Do you juggle or are you an acrobat?
Lukas: Oh my God, is that on my LinkedIn? Unfortunately, I will disappoint you but I wasn’t a performer – I used to organize summer camps for kids in my hometown of Darmstadt. And one of them was focused on running circus workshops and organizing a show. We had a circus tent and all – it was pretty cool actually. 😊
And how high did you get when you visited the Himalayas?
Lukas: I got to the base camp of the Annapurna which is around 4,200 meters above sea level. But all the famous giants are everywhere around you and you feel like you can just walk there. It was a proper six days trek but my next bucket list thing is to do some proper mountain climbing with David. Yes, not only is he the smartest guy in the company but he’s also a pro mountain climber. We already do a bit of bouldering here in London.
This is the last topic I have and I couldn’t really avoid it: tell me the story of your previous name Prix and the rebrand you went through. Are you happy you are Pace now?
Lukas: Oh yeah, definitely. I think it’s a great name for our product. It relates to everything we analyze as it’s about the pace of price changes, bookings and so much more. So, it resonates much better than the previous one which we were called in a fun way. At the beginning, we were part of the Google Machine Learning Incubator in London and it came with a sign for us and we still proudly have it in our office as it’s part of our identity. We’re still Prix in heart. But we’re happy that “Pace” is a lot harder to mispronounce than “Prix” (if you don’t speak French). 😉
If you feel like reading more data, here are two suggestions by the founders of Pace. I swear, we didn’t get any sponsorship from Uber but there is a word out there that boys and girls in Pace are hiring. 📈
See you next time! ✌
For more engineering insights shared by Mews tech team: