Duolingo's 'Creator Army' Is a Strategic Blunder: Why Killing the Unhinged Owl Will Backfire

Duolingo, famously a brand that took flight on TikTok, is now citing an algorithmic change as a reason to emphasize an army of creators over its viral owl shenanigans. Mark Ritson issues a warning to its brand managers in a language they’ll surely understand.

There is a special kind of stupid in marketing. It is not the ordinary stupid of bad creative or wasted budget. It is the strategic stupid – the kind that requires meetings, a reorg and a press release. I call it, Duolingo stupid.

Last month, the company announced it was dialing back its famous “unhinged” marketing approach. Fewer butt jokes. More balance. A recalibration from “80% unhinged, 20% wholesome” toward something more corporate. More generic. More safe. The reason, according to Duolingo, is that organic reach on TikTok has declined, and the platform now wants brands to pay for it.

So Duolingo is pivoting. Like most brands (especially at Unilever), it is now committed to influencers… sorry, creators… to use the current terminology. In May, Duolingo announced it was building what it called a “creator army.” According to Business Insider, it was reaching out to creators “to encourage them to become paid ambassadors and make new TikTok burner accounts for fun content about the Duolingo brand.”

The brand that built the most effective social media presence in the recent history of corporate marketing – 17m TikTok followers, 472m likes across five years, a following that made rival brands weep – has decided to abandon the thing that created all of that and pay strangers to pretend to like it on fake accounts instead. This is not going to work.

To understand why, you have to go back to the beginning and understand what did work, and what, I argue, could still work. In 2021, a 23-year-old hire named Zaria Parvez decided that working in a big Manhattan agency sucked. She got a low-tier job at Duolingo because – brace yourself – she genuinely liked the company and its culture. Then she asked if she could do something with Duolingo’s dormant TikTok account. She had no budget, no team and no brief. What she did have was a shit ton of savvy, a tolerance for chaos and a stupid green owl mascot. She’s left the business now, which goes a long way to explaining the new direction.

I hate the term “social first.” I hate any marketing term that ends in “first” because it invariably means putting the tactical horse before the strategic one; it always means someone is doing something wrong. But I’ll give it to Parvez. Her “social first” approach at Duolingo almost, almost made a cynical old man change his mind.

She started posting in a manic, on-brand, off-the-radar kind of way. Duo stalked celebrities. Duo had public meltdowns. Duo took a shit and then sold its shit. Duo turned up at events uninvited and behaved badly. Within months, the account had millions of followers. By 2024, Duolingo’s own shareholder letters were crediting the “unhinged and viral marketing campaigns” with driving user growth. The company cited the Dead Duo campaign – killing off the mascot in February 2025 – as a key driver behind 49% daily active user growth in Q1 2025. Forty-nine percent. For murdering an owl.

It would be dismissive to call this just a social media tactic. It was pure brand building of the highest order. Consistent, distinctive, disruptive brand building, conducted through a medium that most companies use for corporate announcements and employee birthdays. Duo had personality. That personality was the brand. It was a textbook case of distinctiveness, delivered cheaply and with enormous effect.

The results followed. Daily active users grew 40% or above for every single quarter from Q2 2022 through Q2 2025. Revenue was up 40% year on year through 2025. Paid subscribers passed 10m. These are not vanity metrics. That is a company making serious money, in part because its marketing made the product famous and beloved.

So why change it? To understand Duolingo’s current anxiety, you have to understand what happened to the company in 2025. Growth that spectacular creates its own gravity. Investors get used to 40%-plus DAU growth. Analysts build it into their models. The business, rationally, starts looking for new levers to pull. We don’t say this enough: growth might be inherent to the system, but it is also a motherfucker.

In that context, the marketing team suddenly faced two converging problems. The first was self-inflicted. In April 2025, Duolingo declared itself, ahem, “AI-first” and announced it was replacing contract workers with artificial intelligence. The internet, which had been Duolingo’s most enthusiastic collaborator, turned on it. The company reportedly lost more than 400,000 TikTok followers in weeks. Engagement collapsed. The brand that had built its entire appeal on feeling human, chaotic and authentic had just told its audience it was happy to automate the shit out of everything, including its humans. The audience noticed. They cared. They left.

A word here. If you missed Eric Schmidt getting booed during his speech to graduates at the University of Arizona, you missed a big cultural insight. Not just a general distaste for billionaires, nothing new there, but the specific moment when he got booed by twentysomethings in Tucson: the bit where he got all passionate and sweaty about the impact of AI. Old men like Schmidt love the techno-efficiency ahead. But if you are 23 and looking at the worst job market since the Great Depression, you aren’t feeling it. There is a generational divide in AI enthusiasm that no one, yet, is talking about. And it’s the reverse of the usual revolutionary demographics. The old fuckers are in the vanguard. The kids are being left behind.

The second Duolingo problem was structural. DAU growth decelerated throughout 2025, and Duolingo’s own shareholder letter in early 2026 acknowledged the company expected only 20% growth in 2026, down from consistent 40%-plus growth for three years. Some of that deceleration was natural at scale. But some of it was the direct consequence of the AI announcement breaking the relationship the brand had spent four years building.

When you are a public company and your growth rate halves, you look for explanations that are exogenous. That’s a $50 way of saying that if you work in finance, the answer can never be “we don’t know” or “we are a bit shit at the moment.” You have to find a bullshit excuse. The weather is the first and most relied upon option. FX is good too. So is consumer sentiment. When those hoary old options don’t play, you have to get a bit more creative. Duolingo went with the TikTok algo.

Blame the medium rather than the message or, God forbid, the messengers. Facebook reach declined, said everyone in 2013. Actually, brands had stopped making content worth spreading. Email open rates are falling, said everyone in 2018. Actually, the emails had become terrible. TikTok wants brands to pay for reach, says Duolingo in 2026. Actually, Duolingo betrayed its audience and then killed the creative that had made them care in the first place.

Nothing fundamental has changed about TikTok’s organic reach mechanics. The platform still makes unknown creators globally famous overnight on zero budget. The algorithm still rewards content that people want to watch. What changed is Duolingo went from creating content people wanted to watch to creating a corporate crisis, and then decided the problem was the platform.

The burner accounts proposal makes this worse. The brand that built one of the most recognizable and genuine social presences on earth is now planning to obscure its involvement by having paid people pretend to be organic fans on anonymous profiles. That is digital astroturfing. And, in the current regulatory environment around undisclosed influencer payments, a slightly ropey approach that leaves a sticky residue.

One of the most persistent pathologies in marketing is the inability to distinguish between what created a result and what merely accompanied it. Duolingo’s growth was driven by a distinctive brand personality, consistently expressed, on a platform its audience loved. The channel was TikTok. The driver was the brand. When performance dips, companies instinctively reach for the channel as the explanation. Re-platform. Buy more media. Build a creator army. Do what everyone else is doing. Do anything except confront the harder question, which is whether they have broken the thing that made them famous.

The correct diagnosis in Duolingo’s case is not particularly complicated. The AI-first announcement was a trust catastrophe with exactly the audience that had been its most fervent advocate. It does not require fewer butt jokes and more brand ambassadors with fake accounts. That will make things worse. It needs senior managers to actually understand their brand as well as the consumers who have rejected it.

Duo does not need to grow up. Duo needs to remember what it was. Before it’s too late.

Mark Ritson is a former marketing professor, brand consultant and seven-time PPA Columnist of the Year. He is the founder of the MiniMBA in Marketing.