GetYourGuide’s VP of performance on creating their growth flywheel

During the first lock down, I took a three hour GetYourGuide Original tour in Amsterdam. My four kids and my parents were with me — plus Cato the Basset hound. And the guide absolutely captivated all of us for the full three hours about something as deep and serious as the life of Anne Frank and World War II.

Another unforgettable memory was in Tel Aviv. I was there with six pilot friends and to mix it up a bit from just partying, we did a full day private tour. The guide was not only extremely knowledgeable, he also knew we needed water after a heavy night out, so he brought a coolbox full. Then at noon, when he saw we were hydrated enough, he said “Would you guys like some beer?”. He had cold beers in a separate coolbox in the back. He took us through Jerusalem, Ein Gedi and the Dead Sea, it was an amazing experience.

My childhood dream was to become a pilot, and I was actually in KLM Flight Academy. Most of the flight hours were in Arizona, and after only twelve lessons with an instructor, they just hand you the keys and let you fly solo. It’s like a dream to feel the freedom of flight at such a young age. Later on, at your first long distance solo flight, they let you go across the state, which is a four or five hour flight including refueling. Beforehand you have to make your flight plan, calculate with the weather information how much fuel you need to get to the other airport and so on. During the flight, you continuously measure where you are and whether you are on track. How much do I need to steer to get back on track? And there are all these other external influences like weather, turbulence or traffic that can influence your plans. To make it a bit more interesting, I planned the whole thing over valleys so I had some company from cows and farmers.

In the end, I realized that the flying that I love is not the flying required nor desired by KLM. The school also realized this, especially after I parked my car in the dormitory building with a handbrake turn, they concluded, “Probably it’s best for all if you wouldn’t become an airline pilot.” It was time I figured out what my next career was going to be.

I started off in a call center, for those of you that know the movie ‘the boiler room’, I was selling subscriptions for anything imaginable from newspapers to energy, telco, lotteries even underwear. The only way out for me was to outsell everyone else, which opened up management roles and business to business account management. After two years, an old friend asked me to join his search engine marketing startup. As employee number one, I became head of new business and applied what I’ve learned in sales. But I also realized that search marketing and performance marketing in general is very much like performing a flight. First you determine your target, with what efficiency you’d like to get there and in what time frame. When the campaign takes flight, you have market dynamics, economical climate, viruses. You need to course correct continuously.

So what I was taught in the cockpit in Arizona is the whole build-measure-learn loop. And that’s what I’m still doing today, albeit with a much larger crew. The flight plan is to become the world’s number one platform for unforgettable experiences, and we’re well underway. That’s what made me who I am and why I love what I do.

The basis was already a good fit for growth marketing: I had an exact science profile in high school, combined with a Psychology Freshman year prior to the flight academy. I got lucky to figure out early in life what made me tick. And that was not flying from A to B, but something more explorative. It was innovating with the new technologies emerging and seeing the impact of what you do right away. With eight years of growth marketing experience at two start ups, I didn’t see the need for a diploma anymore. I had caught up with the market with this experimentation mindset.

Google was looking for a lead for Booking.com, their largest advertiser at the time. The five years leading up to that I was CMO at a direct competitor of Booking, EasyToBook.com. In fact, in all the cities where we had a good offering, Booking was always number one, and we were always number two. So Google saw my experience would be relevant for the job.

And you sense that Booking truly embraces the growth flywheel. They are demand led. The beauty of having such a large customer is they not only want to be in all the alphas and all the betas (who doesn’t?), but they will actually give great qualitative and quantitative feedback to the product teams. So as a Google sales person, I didn’t feel like selling. It felt like connecting all the dots at two very intriguing players in the online Travel industry. You represent the client internally and work across all of the marketing technology products, continuously improving their set up and seeing what’s around the corner.

Besides the co-development of the ad stack, Booking saw a need to build their brand in order to continue their growth in the US. That’s when the performance marketing heroes there started to ask how to build brand campaigns on the Google platform. That was a pivotal moment, how do you measure upper funnel campaigns that are mostly considered ‘spray and pray’? By applying a rigorous testing framework to the geographies and mediamix you unleash, measuring the effect on all metrics ranging from Top of Mind Awareness to queries, clicks and conversions.

The very first thing is to be as transparent as possible. It all comes down to trust. When you are a Google rep that has 40 or 100 clients, you don’t have enough time to build that trust because you have to get to your sales target and hit your OKR such as adoption of new product X, which might not be in line with your client goals. Ideally, you should first meet them to understand their business model, what success looks like, their setup and challenges across people-platform-process. Also important, make sure you have a fun activity as well, asides the mostly transactional meeting. Take them out to dinner or even a fun activity in order to get to know the client personally.

When you have good enough visibility on what the client’s company is like and their challenges, you can come with a joint business plan proposal. This will contain prioritized work streams over 4 quarters, so both companies are aligned and working on the topics that will deliver the results, be it on performance, brand, technology or data. Then you connect internally the right people within your company and theirs and track progress and prioritization on a regular basis. I think that in business the first point of building a relationship of trust is very many times just skipped.

With Booking we had the luxury of regular trips to Mountain View for meetings with all of the different products and engineering leads. This also means that you get to know your client well, for instance over a game of volleyball next to Stan the T-Rex on the Googleplex.

Indeed, I worked at the Telco Sprint in Kansas City, which might not be seen as the most innovative company around. First of all, connections go a long way: I was introduced to the CEO by a senior exec at Google. It enabled me to do my pitch as a contractor, face to face with him and a McKinsey partner. I was told he only had 45 minutes. In the end, the meeting lasted 2 hours. I had prepared a deck looking at Sprint outside in. I did user testing research, Similarweb competitive analysis, interviewed industry veterans about the market dynamics, and surfaced customer stories that were painful. And I knew how, if given the mandate, I would organize things around the customer journey. The CEO of Sprint was told many times that his baby was ugly, but not how beautiful she could grow out to be in the future. He asked me on the spot when I could join and that became the start of a great adventure. He gave me the mandate to lead the change in performance marketing, introduce multidisciplinary teams lead by product owners, perform omnichannel lift studies and inhouse digital media.

I think the core difference between being in Europe and America, or actually between tech companies and conservative organizations is the distribution of power through either a form of self steering teams or rather command and control structures. The latter was obviously the case at Sprint, which means you need to have excellent relationships all the way up the chain.

But it’s not the right way for innovation, right? Rather, create a strong sense of purpose and empower the whole organisation to achieve their goals. Ensure the right resources, technology and processes are in place. And then let the people who are closest to the customer decide what’s best. But it’s pretty far from the reality of work in the US, at least in corporate America. That’s the big difference with, for instance, Booking and GetYourGuide where the power to make decisions is actually as evenly distributed as possible in the organization. You create an environment where these teams get authority to just act the way they think is right. They have the right tool set and measurement systems. With the right talent density (be relentless on hiring!) teams are set up to reach mastery. Every individual should have a growth path and be able to learn from the best in their field. At a certain moment it’s no longer about the package, but about authority, autonomy and mastery.

The team we have in America are mostly colleagues from the supply organisation, who are the ones acquiring new inventory and managing the relationship with attractions to deliver on their sales targets. We collaborate closely with them, the demand data fuels their hitlists. We deliver bespoke marketing campaigns for our most important suppliers. We spot opportunities that are not included in our acquisition algorithms.

For example, there was this helicopter tour in Chicago, which we didn’t add to our website because it’s an hour outside the city center. Despite the distance, people searched for it, so the account manager acquired it and it immediately delivered numerous bookings. This collaboration led to so many improvements that right now we actually tweak the algorithm that creates the hit list based on our marketing insights.

Well, they picked up that check just before I joined, it meant we could definitely hire more senior talent, faster. It allowed us to be bolder in our marketing bets. But I think the real impact came when COVID hit — imagine having to raise cash in times of crisis! It gave a lot of security to all employees. In fact, SoftBank benchmarked our response to COVID and gave us the gold standard for crisis management. Another benefit of such backing is also that we could start spending wherever we saw demand.

Prior to the round, we were already a very performance driven organisation. Everything we do needs to have a clear impact — is it actually driving profitable cohorts or not? What changed is the timeframe in which you want the cohort to be profitable. You can become bolder in your bets on growing the upper funnel and app.

When you’re able to increase the headcount, you have the ability to become much more accurate in the way you increase speed of the growth flywheel. How to increase efficiency, whether it’s conversion rate, reach, the way you bid on your portfolios. And the granularity of the data that you insert in the machine. Take Smart bidding as an example — it wasn’t able to do a really good job in our old account structure. So we built a new one, with a better organization of origin, destination, language and intent. This was fuel for the algorithm and significantly lifted our performance. We also do faster predictions and updating of the ad copy, enabled by speeding up and stabilizing our ads pipeline.

Kudos to the collaboration of paid search marketing and tech, led by Paul. Thanks to the Series E, he can now execute his vision to have teams that own all aspects that impact our investment. The total amount you put in paid search hinges on the outcome of : (coverage * queries * click through rate * conversion * expected customer value)/(CPC * clicks). If you want that equation to be at your desired target and maximize volume, you better have teams diving in every aspect and winning in their respective focus area.

By default we have the English content in a new market, or one of our 14 other languages. If we see interest from the local language and we can expect a higher conversion rate from it, we make changes to translate the content to the local language. Then if it converts even better, we add more demand for that market on it. In turn, if we see it warrants a differentiated offering (eg. guided tours in that language), we can feed it into our hit list for experiences. And this is the flywheel that we’ve built. Now, the question is, when to enter a market with your brand campaign. That depends on the competitive pressure and the desire to build strongholds across markets. You can think of markets as operating in various maturity phases, which each have their desired investment profile and media mix to take it to the next level.

I think that the key cultural elements were already there. We already had a culture of learning, a culture of people with huge passion that are very diverse and customer obsessed. There’s a lot of commitment to own things as a team. And luckily, the teams were already set up with having business owners working closely with product analytics and creatives.

I’ll give you an example. In my Department, we have teams across paid search, CRM, display and SEO. Each of those teams have a counterpart, a product manager that they can talk with that has dedicated teams or resources for that channel. Depending on the size of the opportunity for that specific challenge, they get analytical power as well. We have a creative studio, that can deliver the assets needed, that works on improving our joint KPIs. And then on the back of that, we have an infrastructure that is built so that the creatives only focus on the creativity and formats, and the tech is there to scale things.

First thing that I worked on was team size. I needed to recruit, to find the right leads for all of those teams because none of them were in place. We’ve hired 90% of the performance team after I joined. Again, building a relationship of trust is key, only if the team believes that you truly care about their growth you can succeed. This involves a lot of 1-1s, mentorship, vulnerability and just having plain fun together. Lastly, I’m also ensuring the interfaces with other departments are as smooth as possible. The collaboration with Brand and the creative studio setup dedicated to performance is an example of this.

We have a team of marketing analysts, who can handle all kinds of different challenges. They’re set up per topic/specialty, ranging from elasticity, attribution, CLV, experimentation, paid search to segmentation. We’re hiring like mad, and soon I hope that every major channel will have at least a dedicated analytical resource. We have business analysts in the Financial Planning & Analytics team that support us on the investment analyses needed for marketing. On the product side, there are analytical resources to support product development and the building of our algorithms.

When I joined, there were three different attribution models. There was the data driven, one for paid search, that was a last click for our partners, and we had a U shape for the rest. It was a mess. So we needed to align across, and we were looking for a realistic model.

Before GetYourGuide, I had different experiences. At Sprint, with Visual IQ. It was a disaster. It took a year to set up, and it still didn’t deliver answers and costs like millions a year. And I worked with several advertisers on data driven attribution modelling. And then on top of that, doing all these lift studies to understand how to be less wrong.

Then, of course, you have Booking.com. They do everything in-session, last-click, with a lot of lift studies to see what target cost per acquisition is right for every tactic. This method kind of leaves attribution out of the picture. You focus on your tactics and you know what the incrementality is for them. It’s less suited for holistic media measurement as you do not have a like for like comparison of the success metrics.

So we went out to look for an attribution model and found an open source package for Markov Chain in R, which turned out was actually co-created by a good friend of mine. We got in touch with him to make sure he wouldn’t pull the plug on it and to get consultancy on how to use it. And after a few months of work we have it running as our base attribution. We of course do not use it for everything since for instance partnerships use a totally different model to understand value, and we couldn’t tell all our affiliates “sorry your value changed now since we have Markov”. However we do use it to tell the added value of affiliates in general.

To come back to incrementality, we also had this issue that Display wasn’t getting much value. So we decided to start with the click based attribution plus the outcome of the last lift studies we had, where we saw that we actually got a good multiple of incremental sales from display activities.

Additionally we’re doing holdouts, geo or audience tests in order to be less wrong. And we’re trying to innovate here. We are also now one of the first to have built the Markov Chain attribution model inside Google Ads Data Hub. Post GDPR, this is a place where you can upload your data and map impressions to the click path and hence model your data to see what would happen if you included impressions outcome. So far it confirms the same multiplier if you move from click to impressions impact.

Finally, we also have a whole work stream to understand CLV. So for instance, an app install or a newsletter subscriber won’t drive revenue right away, but we use micro conversions to predict CLV over time, and steer on a CLV return efficiency target accordingly.

We can innovate faster than they can imitate. Our competition does not have our team. They do not know how we actually implemented it fully and all of the learnings that we went through. I mean, even if they knew all those, they still have to make it work in their own environment with their own challenges and tracking.

I think defaulting to open is the best recruitment branding you can get in front of potential future candidates.

First, the end of the pixel I’ve not seen yet. It’s not the end of the first party data. It’s more third party. So that’s already a difference. We’re not using many third party cookies anyway. We’re using the audience capabilities of these platforms. So I see issues for companies that use a lot of third party data. Already we see companies moving from third party to second party to first party — our attribution model is built on first party. There’s also the part of Europe and the rest of the world.

That’s when we have the fallback scenario, which is our in-house built experimentation framework. So we’re able to create these test and control sets that are robust enough to run Google’s causal impact package based on our geos.

In general, our marketeers do as little manual work as possible. So what they usually do is they manage large portfolios on a certain efficiency metric. They work together with our creative team to find new hypotheses and test ad formats against each other. After a successful experiment, the format gets filled with our catalog data and other feeds. So that is all optimized.

Of course, we also have automation tools like Braze for CRM and in app messages like Adjust for the app attribution. But we also have many homegrown products that our team can be proud of. For instance we have a scraper that figures out which publishers to partner with or advertise on. We have an ad optimization pipeline that can quickly predict which ad will perform or not, in order to increase our speed of experimentation.

All of this doesn’t require any human effort anymore and is based on our catalog. If there’s a new product sold, it gets populated with the right content on the fly. It goes live on Google in all of the languages, and is tested on our display campaigns.

We’ve gotten so good in deep fakes: why don’t I have my deep fake video material yet? I want to be able to upload a couple of images from my catalog and create a video out of it of somebody who goes on holiday and shows the sites, you know, based upon all of the material that we have online, all of these touristic places. Why can’t we create a first person experience walk through for instance. You could think of some grand experience where you do a Harry Potter tour and decide which character you want to be during that tour.

Jokes aside, I wish it was easier and faster to build user experiences across all platforms. But if it were, we’d have a lot more competition.

First of all, that could not have happened if I didn’t have a very supporting partner. She’s a flight attendant for KLM, working part time, which solves a lot of the logistics. But I also need my support network for when she’s away. And the support from the company for letting me take my kids to school and back, bringing them to whatever practice there is with the trust that the work will be done. In previous companies I sometimes had to explain that family always comes first. At GetYourGuide this is no problem. So it’s about being able to be in an environment where you have enough people that you love and trust that can help you out. It’s also the core reason why I didn’t move to Berlin, because I wouldn’t have that. And I have a lot of respect for all of the people who moved with their families. That’s extremely difficult to do, especially with many kids and a working partner.