Duolingo Beat Its Own User Target, Then Quietly Reached for the Monetization Dial
DAUs accelerated to 29%, video calls got 30x cheaper, Max is being retired — and the stock still trades 76% below its peak.
DUOL · Earnings Call · 2026-11-03
The beat nobody was positioned for
Coming into this print, the consensus framing of Duolingo was a company that had deliberately traded revenue for users and had guided the Street to expect roughly 20% DAU growth for 2026 as the price of that trade. It delivered something better. “Daily active users accelerated for a second quarter in a row to 29% year-over-year growth, up from 23% in Q2” — Luis von Ahn, Co-Founder · 2026-11-03 — and management said Q4 would look similar. Back in May, the promise had been narrower: “We expect that it is going to stay at around 20% throughout the year” — Luis von Ahn, Cofounder and CEO · 2026-05-04. That guide is now quietly stale by roughly nine points, and it was the single biggest number in the 2026 story. Breadth is the tell that this isn't a promo stunt. DAU growth accelerated in eight of the ten largest countries, and subscription bookings grew double digits in most regions. Revenue of $298 million was up 18% year over year with gross margin holding at 72.6%. Yet the equity sits 76% below its May 2025 peak of $540.68 and roughly 19% below its early-September high; the 90-day tape is up only 10%. The market has been paying for the monetization hole, not the user machine.Max dies so that Super can live
The most consequential line of the quarter was a cost figure. Video call once cost about $0.30 a call in inference; “We’ve managed to reduce that cost to well under $0.01 per call. And that’s mainly by moving to open-source models or open-weight models” — Luis von Ahn, Co-Founder · 2026-11-03. That single curve is what turned Duolingo Max — the double-priced tier built to wall off expensive AI — into an artifact. Max Individual gets migrated to Super Family at a similar price; Max is down to single digits of bookings and subscriber count and is being retired outright. This is a promise kept rather than a surprise. In August, Luis von Ahn had hedged: “Another possibility truthfully is that we may actually sunset Max” — Luis von Ahn Arellano · 2026-08-05. He has now done it, and the strategic logic is unusually clean: give the best teaching feature to a base roughly ten times larger, accept near-zero near-term bookings impact, and let retention and word of mouth compound. The inference cost collapse is the precondition, and Duolingo is not alone in noticing it — open weight model surfaced as a fresh global theme in the market's keyword set this quarter, with the same open-weight-versus-frontier trade-off showing up across semiconductor and software reporters. Duolingo is riding a market-wide wave here, not inventing one.The monetization engine is a testing machine, not a toll booth
What replaces Max is not a new price tier — it's throughput. Experiment volume is up about 90% year over year against roughly 10% headcount growth, because non-engineers can now ship small changes end to end. Management says next year will be "multiples" of this year, and that the constraint has migrated from engineering to bottlenecks like having enough good ideas (and enough engineers to fix the bugs product managers introduce). Next year's mandate adds monetization experiments on top of the user-growth and teaching buckets, without the friction lever. That last clause is the real tension of this call. The clearest strategic commitment management made is a negative one: “we have a strict guardrail inside the company to not degrade the free user experience anymore” — Luis von Ahn, Co-Founder · 2026-11-03. It is a direct echo of the 2025 confession that the company was “at the same time under-monetized and over-monetized” — Luis von Ahn, Cofounder and CEO · 2026-05-04 — and it constrains exactly the tool that historically moved conversion. Von Ahn is candid that he does not know whether conversion rates improve; he only knows the squeezing stops. Higher-quality ads, more IAPs (avatar suits), expanded free user trials and chess monetization carry the load instead. Chess is the sleeper. Ten million daily actives, and management calls it "completely undermonetized" — a rare admission that a growth engine is currently creating a monetization asset rather than consuming one. The longer arc is stated plainly: become as good as a human tutor without looking like one, because the failure mode in learning is dropout, not pedagogy. The Brawl Stars and Eggy Party partnerships are the cheap expression of that (no money changes hands; it is pure cross-promotion, a couple of times a year).What got walked back, and what it costs
Two things quietly deflated. Super Lite, merchandised aggressively last quarter, is now described as useful but not a home run, possibly confined to lower-GDP regions. And direct payments, a favorite bull talking point around consumer AI agents routing traffic straight to apps, was waved off: used tactically in performance marketing, but "not currently in the cards" as a major driver. The bill for leaning back in shows up in the fourth quarter. Guidance is bookings +13%, revenue +14%, and adjusted EBITDA roughly flat year over year at $84 million — with an $11 million bonus accrual in Q3 and a second tranche still to come, plus video call rolling out to all paid subscribers at some gross-margin cost. R&D is already running ahead of revenue at $92 million, up 25%, which is why operating margin slipped to 15.2% and free cash flow less stock comp fell 21% to $43 million.There is a coherent bull case in that arithmetic: a company paying itself a one-time bonus to fix its own funnel, having proven the fix works across geographies, now re-arming a monetization machine on a user base several quarters larger than the one its 2026 bookings box was built for. Subscriber growth, per the CFO, bottoms in a couple of quarters as prior-period bookings work through. At 4.7x revenue, the market is no longer paying for the old story. The question is whether it starts paying for the new one.Q3 operating expenses include an $11 million bonus accrual. This bonus rewards our team for reaccelerating user growth. We’re on track to hit the targets, so we’ve begun to record the expense.