Web Watch
Web Watch in One Page
The Duolingo thesis turns on whether the behavioural moat survives the generative-AI substitution cycle while the bookings reset proves transitional. Five active watches cover the load-bearing variables outside the quarterly calendar. The first follows paid-conversion durability — the single leading indicator that distinguishes a freemium compounder from a Chegg-style monetisation erosion. The second tracks a live sovereign-government tender (UK Home Office migrant English testing) that would re-rate the Duolingo English Test as a standalone franchise. The third watches whether free large-language-model providers expand directly into language tutoring, the structural risk the 10-K self-incrimination already named. The fourth follows Apple and Google in-app-purchase fee policy and DMA-style enforcement that directly drives ~28% of cost of revenue. The fifth tracks whether the six open pre-litigation probes into 2025 user-growth disclosure consolidate into a federal class action that would force reserve disclosure and re-anchor the management-trust discount. Together they cover the long-duration thesis test (paid conversion), the asymmetric optionality (DET), the structural failure mode (LLM substitution), the margin lever (platform fees), and the overhang (legal/governance).
Active Monitors
| Rank | Watch item | Cadence | Why it matters | What would be detected |
|---|---|---|---|---|
| 1 | Paid % of MAU and subscriber-vs-DAU growth gap | Daily | Single leading indicator of whether the freemium funnel survives free-LLM substitution; the durable thesis breaker is two consecutive quarters of paid % flat-or-down YoY | Quarterly KPI disclosures, sell-side panel notes (JMP, Cleveland, AppFigures), Sensor Tower / data.ai updates, management commentary on paid attach and renewal cohorts |
| 2 | UK Home Office migrant English-test tender decision | Daily | First sovereign-government adoption of the Duolingo English Test would re-rate DET as a separately-disclosed franchise; £816M live tender with zero embedded value in consensus | UK Home Office award announcements, shortlist updates, contract scope, competitor bids, security-review outcomes |
| 3 | Free large-language-model substitution in language learning | Daily | The structural failure mode the FY25 10-K already named — ChatGPT, Gemini, or Claude expanding into dedicated language tutoring is the canonical Chegg-substitution vector | New language-learning features from frontier LLM providers, AI-native tutoring apps gaining material traction, ChatGPT category crossover in mobile education rankings, comparative engagement studies |
| 4 | Apple and Google in-app-purchase fee policy and DMA enforcement | Daily | ~28% of FY25 cost of revenue is platform fees; every 5pp shift = ~3-4pp gross margin; the FY26 margin guide is one regulatory decision away from re-rating either way | New Apple/Google commission policies, EU Digital Markets Act enforcement orders, US Open App Markets bill progress, web-billing carve-outs that would let Duolingo bypass IAP |
| 5 | Securities class-action consolidation and governance escalations | Daily | Six pre-litigation probes are anchored on undisclosed Q2 2025 DAU deceleration; consolidation into a federal complaint would force reserve disclosure and re-anchor the management-trust discount | Federal-court complaints with class periods covering 2025 statements, lead-plaintiff motions, settlement disclosures, 8-K Item 8.01 filings, proxy-advisor commentary on the CFO-from-Audit-Chair transition |
Why These Five
The report identifies one load-bearing thesis variable (paid % of MAU), one live binary catalyst (UK DET tender), one structural failure mode (LLM substitution), one margin-and-cash lever (platform fees), and one durable overhang (litigation consolidation plus governance signals). Each monitor maps to a specific 5-to-10-year underwriting driver or failure mode rather than to the next earnings date. The set is deliberately built around evidence that arrives between the quarterly prints, where the multiple actually moves before consensus updates — paid-conversion panel data lands weeks ahead of earnings, the UK tender will be decided outside any earnings window, platform-fee rulings come from regulators on their own schedule, and class-action complaints land when the plaintiffs' bar is ready, not when Duolingo reports. The two report-rated highest-impact items — paid conversion durability and the UK sovereign tender — anchor the top of the ranking; the structural risks (LLM substitution and platform-fee shifts) fill the middle; the legal/governance overhang sits last because it widens the discount rather than re-rating the underlying business.