Brand managers the world over would probably do desperate things for the levels of organic engagement that Duolingo enjoys.
The language learning app has become a case study for how to do social media right, following users’ lead even when it means adopting a deranged brand personality in the process.
Rebeca Ricoy manages both Latin America and EMEA as Duolingo’s international marketing director.
Ahead of her session at MAD//FEST this week, she spoke to MediaCat UK about mascot fatigue, the brand’s North-Star metrics, and experimenting with content ‘that’s a little bit less unhinged’.
Duolingo is famous for its owl mascot, Duo, and its social-first marketing strategy. What are your top-performing channels at the moment, and how do you measure their success?
Usually, the top-performing channel is TikTok in most of our markets. From there, Instagram is typically in second place. YouTube also plays a very interesting role for us. We’re quite nascent ourselves in creating content on YouTube, but it is a great channel of growth for us because we see a lot of UGC content there, and that brings a lot of new users. So the way that we usually leverage YouTube for most of the markets is via influencers and just letting people create their UGC, and in some markets, we’re starting to test YouTube shorts, specifically for the global accounts and then also for Hispanic LATAM.
We have two North Star metrics overall and they go hand in hand in most cases. The first one is organic impressions. That’s really important. For our paid media, we only do performance marketing. Other than that, we might boost a post here and there, but that’s tiny. We don’t even have media agencies anywhere in the world. And then within organic impressions, we look at how many new viewers we’re getting versus the viewers that already know us. Within the markets, it’s also very important to see how many of those viewers are actually coming from the market because it does sometimes happen that France will go viral internationally, which is not necessarily benefiting the market.
The other key metric is organic new users coming from our social channels. Within our app, we have a survey we call ‘How did you hear about us?’. It’s not an exact science but it is pretty indicative when people select TikTok, YouTube, or Instagram. That’s a really good way for us to understand the impact of the work that we’re doing on social and how it translates to actual new users, which from the business side is our key metric and how we are measured.
Is there anything that you’ve found that you’re not able to do or achieve with organic social media?
I think the main thing that is a lot harder to achieve through organic social is trying to reach different audiences that are not in the Gen Z or millennial segment. If we want to reach older demographics, we usually have to go to other channels. For example, in some markets, we explore TV, and that’s a really good way of tapping into older segments. We also work with content creators who sometimes talk more to that audience, or we do more PR so that we can be in outlets that older people are actually reading.
That tends to be the hardest thing: tapping into older demographics or kids. If you want to reach kids, then you usually have to first go through the parents. But that’s why our core audience usually is Gen Z and millennials. We know that they share it with their parents and their grandparents. That organic word of mouth still happens.
What role does paid media play in your marketing strategy?
It’s usually for performance marketing, which we run worldwide in a centralised way from our New York office. We divide the world into four different groups depending on how mature they are, their monetisation potential, retention metrics, and other things like that. We divide all the markets into these different groups and from there we run performance marketing. And that’s an entire machine in itself.
Those are very specific ads that are optimised and running for performance marketing. Sometimes we’ll include or mix some assets that are created by influencers or by the brand side of things, but usually it’s just a well-oiled machine that is running on the back end on its own.
Beyond that, we have very specific paid media moments where we can, for example, buy a billboard. For Eurovision, we bought a billboard in the Malta airport. It was a stunt basically. We were playing with Miriana Conte and the whole KANT song. The copy was ‘learn how to serve Kant in over 40 languages’ and we had Duo dressed like Miriana with the leopard-type outfit. It was a fun, quirky, and unhinged thing that got some press and got the attention of the participants, which was way more than we expected it to do.
How do you go about collecting insights about your users and potential users? What are some of the most useful insights that you’ve picked up?
We always say that the brief is in the comment section, and we truly do believe that. So we can do a little bit of market research but it’s not really that relevant for us. For us, it’s more about reading the comment section and understanding what people are saying.
A lot of the lore created around the brand and Duo actually comes from that. The fact that Duo is so intense and will go out of his way in a very unhinged way to get people to do their lesson comes from the comment section. The fact that Duo is obsessed with Dua Lipa comes from the comment section, too. Honestly, it’s all from the comments. We just build on that. A lot of the campaign ideas and even potential partnerships that we work with all come from the comment section, so that’s that’s the main source of information.
Duolingo has both global and local social media accounts. What is the strategy for the brand’s international accounts? You’ve previously mentioned that the UK account did not have the expected impact. Why is that?
We launched the UK account just as an experiment because we had a hunch that it actually was not going to work. And our hunch was confirmed. The issue is that the UK audience is very well covered by our global account because it’s in English.
I’ve been in London for two weeks now and my algorithm has kind of shifted to more London content, but it’s still pretty US-centric. There’s a huge amount of UK people looking at our global account. And so that was our hunch.
We had a few viral hits but there was no new user impact. So now we’re taking a pause. But we’re starting to think of it as more campaign-focused, where if a good opportunity arises within the market, we’re able to leverage it and use our global account to talk about whatever is happening.
For example, if Dua Lipa is giving a concert at Wembley — that happened last week, and we actually wanted to do something, but we were too late. That would have been the perfect opportunity to leverage a team in the UK and do something stunt-y in the UK that can actually benefit our global account.
Is there a specific market that has seen really good results?
Yes, pretty much all of our Hispanic accounts. They’re much more mature markets and accounts. They’ve been going on for a while, and the impact there is 100% tangible.
We launched France last year, and it had immediate traction and impact, both in virality and with new users. The team in France is doing a spectacular job of creating amazing content that is very unhinged, extremely funny. And so their challenge is that they frequently go viral outside of France because their content is so good, but they have to make it very local so that they’re able to show impact in their market.
Even more recently, we launched Turkey two months ago. That’s also in an experimental setting, to see if it’s worthwhile keeping long-term. We haven’t defined it 100% but we’re definitely seeing impact in terms of new users and impressions.
Another thing you mentioned last year was that you were starting to look into mascot fatigue and what product-led content could look like. What did your research reveal and did that influence the stunt faking Duo’s death?
It wasn’t really entirely related. I guess subconsciously it was, but it wasn’t done on purpose. The Dead Duo campaign actually originated from the product. From time to time, we have app icon changes and that is created by the product team and the designers within the product team. One of the designs that they had was the crossed eye Duo, and someone in marketing said, ‘It looks like a dead Duo. Why don’t we just kill the bird?’ And that’s the campaign that we did around the app icon change. Of course, the ‘Why don’t we kill the bird?’ comes from this subconscious knowledge of mascot fatigue, so it does stem from that, but it wasn’t really a result or a strategy.
We are still quite focused and have a very clear strategy on the need to evolve the brand and the need to talk more about our product. During the last two quarters, we’ve been experimenting with creating content that is a little bit less unhinged, viral, brain rot-type content, and much more informative and educational, talking about our product and learning languages. This past quarter, we’ve gone to the extreme in some of the posts talking about the product.
And then we have a mix of content that is a little bit in between. We call it the pull-push content. It’s in that grey zone, which is talking a little bit more product but is still fun and unhinged in a way. Funnily enough, that’s the type of content that is actually working the best for new users and for resurrection. We have some good learnings there.
Right now, about 20% of the content that goes out is within this strategy. We need to evolve it more and we need to scale it, but it’s a hard thing to actually do. The hardest part is getting the teams — both internally and within the agencies — to stop thinking about brain rot and to think about this other type of content and make it work.
This week, you’re speaking at MAD//Fest about strategic partnerships. What do you look for in a strategic partner besides the language element? Do you have any requirements for how you will or won’t adapt the brand for the integration?
We try to look for partners where we can extend our audience. So that’s usually partners that are relevant to specific fandoms that we want to be a part of. K-pop and anime, for example, are two of the basics for us.
The other really important thing that we try to look at is brands that have an affinity to our style or our type of humour. And it’s not for everyone. We understand that not all brands are risk-takers and they can’t do that. They don’t have permission to do that.
We’ve been in situations where we think, ‘Yes, this is going to be amazing,’ and we have been talking with the marketing manager from the other brand, but the moment that it goes to the VP or the CMO, it’s dead. We try to get approvals and be sure that everyone is on board on the other side of the table to truly do something that is relevant and that surprises people.
Those are usually the main components that we look at. And then, of course, the other one is ensuring that it’s locally relevant for our audience. It really does need to tap into local fandoms, local pop culture, and local meme culture in order for people to like it and share it.
You’ve had quite a few interesting partnerships lately. Can you share an example of one that you’d say fits your ideal? What do you think made it so successful?
One of the most recent ones, which was earlier in the year, was a global one. It was for the launch of Squid Game 2. This was ideated when Squid Game season one went out because we saw a huge spike in Korean learners around the world. We didn’t have any official partnership there and we were too late to the game at the time. We had no idea that this was going to be such a big thing, but we saw that and we realised there’s something huge there.
We thought that we needed to work with Netflix so that for the second season, we can actually do something together. So we worked on a full-on campaign with them. It’s a very funny video of Duo with the pink guards and it worked quite well because the execution was on point. It’s very relevant to the audience and it actually also triggered the launch of Korean for many other languages.
Finally, what consumer trends are you seeing at the moment, both globally and across different markets, and how are they shaping current projects?
Very related to Squid Game, we see huge fandoms that continue to grow. For Korean culture in general, we see Korean TV content and K-pop. K-pop is huge worldwide, and it is growing quite a bit cross-generationally. The other one is Japanese culture, both from the tourism perspective and also anime. So both Korean and Japanese have grown a ton in terms of appetite.
For years, we’ve said that we need to launch Korean and Japanese. From when I first joined Duolingo, my only wish was to have Korean and Japanese from Spanish because everyone was asking for it. So this year, we launched Korean and Japanese from most of our main languages.
And then I would say that the other one that is not necessarily a global trend, but that we do see growing and that we wanted to tap into is chess. When I was growing up, chess was only for the geeks; it wasn’t cool at all. But apparently, chess is now a cool game to know how to play. So we just launched chess, and in September, we’re going to be launching it on Android and in other languages. So that’s something that we’re starting to see, but I would say that the biggest one that it is 100% global is Korean and Japanese. That’s huge.