If you’ve ever done a Duolingo lesson at 11:58 P just to save a 147-day streak, you’ll understand why people feel personally betrayed right now.

Earlier this year, the language-learning app rolled out its new AI tier - a beefed-up, more personalised experience - and inconspicuously moved some fan-favourite features behind a paywall. The internet did what the internet does best: 

completely lost the plot.

This update had users threatening to delete the app, boycott the owl and never conjugate a verb again.

And yet… none of this matters. 

Because Duolingo has been in a similar situation before.

They spent six years making almost no real revenue while investors happily handed them $183M. 

Why? 

Because engagement trumped monetisation - and they knew exactly what they were doing.

This week, I’m breaking down the strategy behind that decision and why it should matter to SMEs who are constantly told to monetise too early, too fast and often for the wrong reasons.

Here’s this whole newsletter in 30 seconds:

1/ Duolingo raised $183M pre-revenue because engagement beat early monetisation.

2/ Their strategy worked because software economics reward scale before billing.

3/ Short-term backlash ≠ real feedback - consistent, long-term behaviour always tells the truth.

4/ SMEs win by validating early and monetising when their category allows, not when social media thinks they should.

When accuracy matters, typing can introduce errors and slow you down. Wispr Flow captures your spoken thinking and turns it into formatted, number-ready text for reports, investor notes, and executive briefings. It cleans filler words, enforces clear lists, and keeps your voice professional. Use voice snippets for standard financial lines, recurring commentary, or compliance-ready summaries. Works on Mac, Windows, and iPhone. Try Wispr Flow for finance.

Start flowing free

Before Duolingo became arguably the world’s most famous pocket language teacher, it was just a free app burning money at a pace that would give most founders heart palpitations.

Between 2011 and 2017, they raised $183M across multiple rounds:

2011: Union Square Ventures kicks things off with $3.3M
2012-2014: Another $35M flows in
2015: Google’s CapitalG leads a $45M round
2017: They cap it off with a final raise that brings the total to $183M

The revenue, however, was almost non-existent.

Duolingo knew that charging £4.99 a month in year one would kill their customer acquisition before it even started. EdTech is brutally competitive, and they were up against free substitutes (YouTube), habitual behaviours (textbooks) and giants like Google Translate.

So they played the long game:

Ignore revenue, maximise engagement and let the product speak first.

And it worked. 

By the time they finally introduced paid features in 2017, they had 300M users logging in, practising daily, building streaks and forming habits that were almost impossible to break.

Investors didn’t fund revenue. They funded behaviour.

Engagement was the proof.

When Duolingo eventually flipped the monetisation switch, it was explosive.

Here’s what “six years of no revenue” turned into:

2021 IPO valuation: $3.7B
2023 revenue: $577M
User base: crossed 500M
Brand moat: borderline cultural phenomenon

That early patience became the foundation for everything they’re doing now - including the controversial AI rollout. Wall Street didn’t love it. Users definitely didn’t love it. Stock dipped, comments exploded, streaks were threatened.

But amidst the noise? The fundamentals stayed exactly where Duolingo needed them: engagement strong, churn stable, growth intact.

This is the part SMEs should pay attention to.

The backlash isn’t a sign they’ve lost their trust indefinitely. It’s the same play they used a decade ago: Sacrifice short-term comfort to build something competitors can’t copy.

AI is just the latest example. They’re using it to create features that others won’t be able to catch up with for years; the same way their early free strategy let them dominate the EdTech category before anyone realised what was happening.

Investors saw this long-term instinct early, which is why they wrote the cheques before revenue had materialised.

And it’s exactly why Duolingo’s “pre-revenue era” was one of the most strategically disciplined moves in modern consumer software.

Forget “mastering” a language, here’s what SMEs raising capital can learn from Duolingo.

Validation is non-negotiable. Duolingo didn’t wait six years to monetise because they were spiritually aligned with “the journey” - they had hard proof:

Real signals: retention, usage, streak obsession, churn that barely budged. 

Conviction paired with data is strategy. On its own, conviction can drift into something less useful.

And I get why. Conviction without validation is an easy trap to fall into, especially when you care deeply about what you’re building. I’ve met thoughtful, capable founders who stay pre-revenue for years - not out of stubbornness, but because the signals they’re hoping for never quite materialise.

But…

Conviction + real behaviour = progress.

Conviction + no behaviour = vision without traction.

One builds a billion-dollar engine. The other never leaves the drawing board.

Software can delay revenue because the marginal cost is basically zero. Spotify can give you a free song. Citymapper can plan your entire commute. Duolingo can let you butcher Spanish for years without paying a penny. That’s the luxury of code: one more user costs nothing.

Physical products? Completely different physics. You can’t give them away endlessly “to build traction” unless you want to speedrun insolvency.

If your category allows day-one monetisation, do it. If it punishes early monetisation, delay it. The smartest founders follow the maths rather than trends.

Short-term outrage doesn’t mean your strategy is wrong - if it did, Duolingo would’ve shut down the entire app after all of the AI drama.

Users often don’t know what they want until they’re shown it; remember, if you’d pitched TikTok in 2002 (“endless videos of strangers dancing in their bedrooms”), you’d have been escorted out of the building.

Feedback is fallible. Behaviour is truth.

The real signals are retention, churn, usage patterns and revenue health - the boring metrics that build an accurate picture of the future.

Ignore the noise. Track the behaviour. Develop conviction from data, not from comments written in all caps.

Strategy is not universal; it’s contextual. 

Every year there’s a new “rule” founders are supposed to follow - monetise early, monetise late, build community, build audience, build AI, avoid AI. 

It’s all useless if it ignores the physics of your category. Software plays one game. Marketplaces play another. CPG, hospitality, hardware - completely different economics.

Your growth model should reflect your unit costs, competition and distribution realities, not whatever is trending on LinkedIn. Duolingo can delay revenue; your bakery cannot. The best founders respond intelligently to the constraints of their industry.

Great things happen when strategy, behaviour and timing line up. Duolingo simply proved the formula.

At FundOnion, we’re really looking to scale over the next couple of years, and we want to become a global pioneer in SME financing. But for that to happen, right now we’re in this phase where we’re building infrastructure long before the scale shows up. We’re taking bets on how the market’s going to respond to the products we’re working on, and in the short term the results are basically nil. Nothing.

It’s all investment at this stage - capital, time, energy - and of course there’s a part of me that knows if it doesn’t pay off, it could all look like a waste. We could’ve stuck with business as usual. But I trust, very deeply, that the payoff will come.

And yes, it’s uncomfortable because you’re not getting applause or feedback while you’re laying foundations. But that’s often the only way great things actually happen. You have to believe in the work long before anyone else sees it.

The takeaway is (relatively) simple: build the foundations, then the revenue; not the other way around. Duolingo proved that when you pair conviction with actual behaviour - not vanity metrics - the long-term upside is massive. 

If you want help making sense of where your own business sits on that curve, I’m here. Just reply.

Till next time,

James