Innovative Automation for User Acquisition: An AppChat with Hopper
Shani Rosenfelder: Everyone, welcome to another edition of AppChats. I am Shani from AppsFlyer. Today, we’re going to talk with Simon Lejeune, he is the Head of User Acquisition at Hopper. Hopper, in case you haven’t heard of it, is a Top 10 travel app in the US, at least according to Appstore rankings from today. And Hopper’s doing some pretty amazing things overall in its approach to travel and, specifically, as it pertains to user acquisition. So, Simon, let’s get started by telling us a little bit about yourself, professionally, how you ended up at Hopper.
Simon Lejeune: Yeah, thanks for having me. My name is Simon and I’m the Head of User Acquisition at Hopper, so I’m in charge of bringing those install numbers up, spending the budget as wisely as possible. We’re a small team at Hopper, but we’re a very technical team. We have a couple of User Acquisition experts, we work with some of the best creatives, and we work very closely with the engineering group and the data science group.
Before that, I was also in travel, I was working for Busbud – it’s one of the largest bus travel booking websites – where I was focusing more on SEO and SEM. So, it’s a nice change to go to Hopper, but we had similar challenges that required automation work, which is what we’re going to talk about today, so I’m really excited.
Shani: All right. So, yeah, tell us a bit more about the product and a bit more of what I’ll just briefly explain. I know you guys raised a massive fund a couple months ago. How much was that?
Simon: We raised $100 million.
Shani: $100 million, that’s amazing. Well done.
Simon: Yeah, you know, the travel industry is huge, the pie that we’re in is really big, so you see those high raise and high valuation of companies. But essentially, Hopper just sells travel, just like you said, and we are mobile-only company and so we are app-only. And we’re actually the largest – the most downloaded flights and hotels booking app – in North America, which is really crazy when you know that companies like Booking or Expedia spend together over $10 billion in marketing every year, yet we have more installs than them. We like to say that they spend our entire annual marketing budget every 6 hours.
The reason we are able to do this is because, obviously, we focus on mobile and we spend all our marketing budget on mobile app install campaigns on social media. So, we also sell travel, but we’re selling it differently. What we do is create trust and inject trust in the travel marketplace, where there isn’t a lot of trust in the first place. So, you see people shopping around for their flights, having those conspiracy theories about flight prices. You will have 25 tabs open checking hotel prices everywhere, so I think there is an opportunity there and Hopper was able to see it.
By using a lot of data and really smart data scientists, we were able to build this algorithm that predicts the future prices of flights. So you search on Hopper and we will tell you if you should book now or wait for a better price. If we tell you to wait, then you subscribe to push notifications. We keep watching your flight in the background and as soon as the price drops to what we think is the lowest point, we send you that notification. Basically, we take you by the hand and tell you what to do until you have to book your travel and people really like that.
Shani: That’s really cool. So, I think this, I guess, extremely data-driven company is not only the product, it has also trickled to marketing, right?
Shani: That takes us to the next point, you know, our main focus or topic of this conversation is how you guys use a pretty innovative user acquisition strategy – basically looking at automation in UA. Can you explain a little bit in general about what that means, what automation UA is, how this idea was born, and how this whole thing got started?
Simon: Yeah, so when I started about two years ago, we were both advertising the app itself –you know, the classic video ads about what the app was doing really well – and we were advertising specific deals. We had users and we’re already trying to find the best deals for them, so we were like, “Hey, how can we reuse those deals that we’re finding for our existing users and use this really good data to acquire new users, to attract people to the app?” And we realized that the deals, ads, and campaigns were performing much better than your generic video campaign. In a sense, it is because it’s very very targeted; if you are, for example, in Dallas, you’re going to see deals out of Dallas, so the ads feel more relevant and closer to what you’re looking for. That also meant that the interaction rates with the ads were really strong, the click-through rates, the swipe-up rates were really high, much higher than a generic video ad.
And so, we were trying to move more budget towards deal campaigns, but then you run into this issue where you can’t generate or create hundreds of deals out of every airport manually and the prices change all the time and you have different currencies, different languages, etc. – there was no way to do this manually. But it all starts manually, it’s a manual test and then you want to scale it, you want to extend it, and then you take the necessary steps to do that.
And I guess, like if you were a traditional e-commerce, you know about product feeds, Google shopping and all that automation part, but for travel and social, those ad products weren’t really built in the first place. Facebook had some kind of dynamic remarketing for travel, but they didn’t have dynamic user acquisition and so that’s what we set out to build.
Initially, we started to work with Smartly.io – it’s a Facebook marketing partner – to kind of have a shortcut through the Facebook API. So the input was a spreadsheet with a bunch of flight deals, the output was a campaign on Facebook, targeted to users around airports, showing them specific deals. That was really cool, that worked pretty well, but there was not a Smartly for Snapchat or for Pinterest or for Twitter, so we started to build that ourselves and to plug in what we call our Hopper Ads Engine, which fetches all the flight deals and plugs that into the Snapchat API, the Twitter API, and the Pinterest API.
All that we call user acquisition automation because the vision for it is to have one big red button that we would come in the morning and just press and then all the deals are created, all the campaigns are created, automatically on all social media.
Shani: So, basically, we’re talking about creatives, creative generation, and targeting.
Simon: Yeah, exactly. Because with the data that we sent to the networks, there was the geolocation of the airport. Most social media have a kind of geo targeting available, so you say “I want people in a 100-kilometer radius from the airports and show these people the deals from that airport.” So, now you have the automated targeting, the creative generation, the campaigns creation, the ad sets creation, and you can also automate the initial budget that you will set.
After that, the second part is really about optimization and rebalancing your budget from one deal to another. Something that we found out is that if you advertise 200 deals – maybe two or three of them are just crazy, like a 99-dollar flight to a really good destination – automatically, you want to shift your budget towards that best performing deal.
So that’s what we’re working on right now, to replicate this process within a campaign, across campaigns, and the next and final step would be across different channels. We would launch all the deals and find that, “Hey, this one on Twitter is performing really well, let’s shift some of the Snapchat budget towards Twitter” – something like that.
Shani: That happens automatically?
Simon: That’s the vision, right, that’s the vision, that’s what we want to get. There is nothing that would prevent you from doing that today because we get the data – performance data – in real time, so you just need to build a bunch of rules to then send the right message back to the API; for example, ‘reduce the bid’ or ‘increase the bids’ or you know, your daily budget, things like that.
Shani: Let’s say I’m a marketer and I’m just getting started and I’m excited by this idea. So, tomorrow morning, I want to go to the office, what is the first thing I should focus on? I mean, in terms of what I should do, in terms of maybe hiring someone who can write a script, hiring a data scientist, etc. What are the things that I can do?
Simon: Yeah. So, I guess like what you said is true, you have to have technical people, in-house people who understand what an API does and how it works and can run a lot of small tests. I’m lucky enough to have a colleague that’s very technical, to have a detached engineer from the engineering group that can tell me if my ideas are doable or not sometimes.
But basically, I think the first thing you want to do is build or think of the craziest plan and the craziest vision you can have in terms of automation. If you could just decide what your marketing or campaign structure should be, just think of that and then think of the smallest first step you can do to get there. At the end of the day, it’s not about the destination, it’s more about the journey and it’s about building an automation culture within your team.
That’s what I realized that we were doing, you know, I just repeated ‘automation,’ ‘automation,’ ‘automation’ all the time and at the end of the day, it’s just like everyone now when they start to test a new channel. They think about how they can scale it, they think about efficiencies, about technical integrations. So, it’s more of a culture and a style that your team will have than a specific, well-thought out plan.
For us the vision is very simplistic and funny – it’s ‘the big red button’ – but I talk so much about it. Actually, some of my colleagues bought me an actual physical red button that’s on my desk, but that was just great because I was like, “I repeat this so much that people make fun of it.” But then again, that’s when you know you’re on the right path to creating a culture in your team. So, now we joke, as though my button was the front end, I’m like “why are you guys waiting, like I’m pressing, nothing is happening, so just build it better.”
Shani: Your own nuclear suitcase. So, I mean, besides, again, better marketing performance which probably is obviously the result of automation, what other benefits does automation have?
Simon: Well, you get to do interesting podcasts like this one. But I think, well, it definitely helps being a more interesting client for your ad partners. So, because we focus so much on automation, Facebook and Google and Snapchat are really interested by what we do because they see it as the future and they want to bring that to their other clients, so that kind of helps your relationship.
We get access to a bunch of new ad fads and betas from the different partners. Then, it also helps with your recruiting, right, because some people, smart people who think about that stuff, they hear about Hopper and they see we have this automation culture, so I guess it makes us more attractive and we were able to hire some of the best user acquisition marketers in the market – that was really a strong point.
And then, I guess it also sends a message to your backers or investors, it’s like “my machine is ready, send more million dollars. I don’t need to increase my team to be able to keep the same efficiency in my marketing,” so that’s also interesting to send a message to people who fund your company if you’re in that kind of space or industry.
Shani: Yeah, so that sounds really good, but there are obviously some challenges and some pain points. Can you share some of those and maybe how to overcome these challenges? Or not, not yet, you know, we definitely do not have everything figured out yet.
Simon: Yeah. I mean, the danger can be to obsess too much about the details, because we get a lot of data from attribution partners. That’s why we use a lot of data from Facebook and from Snapchat. If you try to find like the truth about your real return on investment (ROI) and the last percentage and making sure that every number matches with the others, then you will lose a lot of time because you can’t get there. So, over-obsessing about the perfect picture of what’s happening can be dangerous.
And then, the pain point is just finding out the real attribution, the quality attribution. So, obviously we work with, I think, the best attribution platform, AppsFlyer, we get this quality data, but this data is just not telling you the full story. We know that there are different touch points and it’s not always perfect – you have limited ad tracking, things like that, that prevent you from having the perfect crystal-clear picture about attribution. That’s really a paint point.
So we’re really interested about that and trying to figure out where our organic in-sales come from. Are they the result of paid campaigns and so should we attribute all the sales that are made by organic users to our paid campaigns, things like that? And recently, we actually ran a survey, because we were like, “OK, how can we have a better picture of our attribution?” And then we realized we can just ask our users. We just asked them ‘Where do you come from?’ or ‘What did you click on?’ – in other words, why did you download the Hopper app?
So, we selected 30,000 users, send them a notification, and say that it was a link to a type form. Type form is my new favourite marketing tool – it’s the best way to uncover interesting things about your users. In this case, we asked our users like ‘Why did you download Hopper? Is it because a friend told you about it, did you see an ad, did you see it in the App store, or did you hear about it and then you used the main channels?” And we realized that over 20 percent of our users were organic, or said that they actually saw an ad.
We also realized that, for example, some said a friend told them about the app, but we saw in our attribution data that they actually clicked on a Pinterest ad. So, we were initially thinking like ‘Oh, our Pinterest campaign is doing really well,’ but later realized that if the friend didn’t tell them about Hopper, they wouldn’t have clicked on the Pinterest ad. And even within the people who said, ‘I saw an ad,’ because we’re able to tie that back in with their attribution data, once they filled out the questionnaire, the form, then we actually knew who they were in the app.
Some of them would say, ‘I saw an ad on Snapchat or on Twitter,’ although we had Facebook or Instagram as their media source. So, the issue is not with the attribution data itself, it’s about how you interpret it, because they actually clicked on those Facebook or Instagram ads, but they thought it was a bug, and said that it was ultimately Snapchat that made them download the app.
It’s very complex. If you try to obsess about these things and uncover the truth, you will never see the end, so yeah, it’s about making comparison between different channels and making decisions that are in the right direction, but not exactly certain. You never know exactly how much you should increase your bid or how much you should decrease a budget to the last dollar, for example.
Shani: I definitely see that also as a huge trend is the breaking of organic discovery. Maybe there’s no such thing as organic at all, like at some point on the journey, there was a marketing touch point?
Simon: I can tell you that 60 percent of our users told us that someone told them about Hopper, which was really interesting. Also, app store discovery was a very very small percentage and that was also an interesting insight that allowed us to stop obsessing about our app store rankings. And then, we then asked our users, “Have you talked about Hopper to someone else?” and 80 percent of them actually told someone to download Hopper.
So, we’re like wow, we have 80 percent of our customers, 80% of the people who downloaded Hopper, who have become ambassadors. Then it was like, “Now that we know this information, how can we accelerate this?” So, yeah, attribution and then, you know, you have your technical attribution, maybe you try to add some self-attribution surveys and I’d love to see more information and more marketers do that and share the information.
Shani: So, just a final question for our session today. How do you see automation in UA in five years from now; are we all going to be out of jobs?
Simon: Yeah, well I hope so. Haha. Well, no, interestingly, I think working with automation now gives us an edge on the other advertisers, because Facebook and Google haven’t done all the work. But I think five years from now, companies like Facebook and Google, but also AppsFlyer, will have integrated all the automation best practices. And you see that, slowly, Facebook is adding layers of automation, like they recently launched Campaign Budget Optimization. You launch your ad sets, you don’t have to worry about which deal is performing better, Facebook will automatically send your budget towards that ad set – that is something that we used to do that we don’t have to do anymore in our ads engine.
And you see Google going with universal app campaigns, where they don’t let me bid on a single keyword. So, if you see that trend going five years from now, well, every advertiser will have access to the same kind of automation, which is kind of exciting if you’re a new and small company. You can wade out and have access to the same level of marketing automation that bigger companies do.
And I think what’s going to happen is that the real space where you will have to be competitive will be creative, because creative will come back and be back at the core of your marketing strategy. Now, we have focused on a lot of shiny things like targeting and optimization and automation, but a few years from now, we will go back to the original – maybe like the Mad Men vision of marketing, where the message and the creative will be back at the center of your campaigns.
Shani: And that’s human, probably.
Simon: Yeah, yeah.
Shani: So, there is hope after all.
Shani: Alright. This was super interesting. I think this is almost like a look into the future, which you guys are already implementing.
Simon: One last thing I want to say is like, you know, I repeat to my team, ‘automation,’ ‘automation,’ ‘automation,’ but what we have at Hopper, we’re lucky to have a CEO that has a really good vision. He always tells us this very simple quote, “Skate where the puck goes.” That’s just a good quote if you’re a Canadian company, but in general, try to be a few years ahead of the market. And because marketing and automation is accelerating, being two or three years ahead means shooting really really ahead, so yeah, that’s like what we’re trying to do and what I would suggest other people keep in mind when they think about their marketing strategy.
Shani: Alright, Simon, this was really interesting. Thank you so much for being a part of another edition of AppChats. More to come soon. Thanks a lot for tuning in.
Simon: Thanks, bye.