Duolingo Onboarding Teardown: Lesson Before Sign-Up

Duolingo's onboarding asks you to learn before it asks you to sign up: a handful of profiling questions, a goal, an optional placement test and a complete first lesson all happen before any account screen, and the subscription offer comes later still. Each step below pairs the screen with the design principle and, where available, the published test result.

Screens are described from the iOS app as observed in October 2025. Duolingo runs hundreds of A/B tests, so wording, order and options vary by cohort and country unconfirmed; treat this as the stable skeleton.

The sequence at a glance

Step 1: Welcome screen

The first screen is the owl, a one-line promise about learning for free, and two buttons: a large green "Get started" and a smaller "I already have an account". There is no carousel of feature slides and no sign-in-with-Apple button on this screen.

Principle: one decision per screen. Growth.Design's Duolingo case study credits the app's heavy use of a single highlighted action with reducing hesitation, a direct application of Hick's law (more options, slower choices) ⁴.

Step 2: Pick a language

A scrolling list of courses, each with a flag and the number of learners. Tapping one moves on immediately; there is no confirm button. The UI language is already inferred from the phone, so the app never asks "what language do you speak?" unless it has to.

Principle: start with the choice the user came to make. Screens Design's recorded-flow teardown describes this as establishing route and purpose before anything else, so every later question feels like it belongs to the user's goal rather than the company's ⁵.

Step 3: Profiling questions

Three quick single-tap screens follow, each with Duo in a speech bubble framing the question as helping to build the course: how you heard about Duolingo (TikTok, friends, App Store search, and so on), how much of the language you already know (four tiers from "I'm new" to "I can talk about most topics"), and why you are learning (travel, work, school, brain training, fun, family). A thin progress bar across the top advances with every answer.

Principle: questions are cheap when they visibly change what comes next. Screens Design calls this "calibrate rather than interrogate": your proficiency answer decides whether you see a placement test, and your motivation is reflected back in later copy ⁵. The "how did you hear about us" screen is zero-party attribution in a post-ATT world, bought for one tap while the user is cooperative.

Step 4: Daily goal

The user picks a daily time commitment: four tiers from roughly 5 to 20 minutes, labelled Casual through Intense (exact labels and minutes vary by test; unverified for any specific build). A second screen asks for a longer-term goal and offers to set a reminder.

Principle: a goal set by the user is a commitment device. Gina Gotthilf, Duolingo's former VP of growth, said in 2017 that having users define their own goal during setup was deliberately paired with the streak, and recommended letting users set it from the start rather than later in a settings menu ¹. The goal also defines what "done for today" means, which is the unit the streak, the widget and the reminders all depend on.

Step 5: Placement test or start from scratch

If the user said they already know some of the language, Duolingo offers two options: "Find my level" (a short adaptive test) or "Start from scratch". Beginners skip this screen entirely. The test gets harder as you get answers right and ends when it is confident, then unlocks the matching section of the course.

Principle: respect existing knowledge, but never force the test. Gotthilf's team found that lowering early difficulty generally helped, and Growth.Design highlights a deliberately easier "happy path" for activation ^1,4. Making the test opt-in avoids a drop-off point while still serving the intermediate learner who would quit if forced through "hello" and "apple".

Step 6: The first lesson, before any account

The user is now in a real lesson: tap-the-word translations, listen-and-pick, matching pairs. A green progress bar fills, correct answers chime, and the lesson ends with an XP total, a time-taken stat and a "day 1 of your streak" celebration. Nothing so far has required an email.

Principle: deliver value before you ask for anything. Gotthilf reported that pushing the sign-up screen a few steps later, after the first lesson, produced about a 20% increase in DAU, and that a later refinement of the soft and hard walls added another 8.2% ¹. Appcues puts it plainly: by the time an account is requested, the user has already answered questions, got some right and watched a streak counter start ³.

Step 7: The soft sign-up wall

After the lesson, a screen offers to "create a profile to save your progress", with Apple, Google, Facebook and email options. The escape hatch is a muted "Later" link. Progress continues anonymously; the wall comes back after a few sessions, and only eventually becomes blocking.

Principle: soft walls first, hard walls later. Gotthilf described the escalation explicitly: a soft wall just after the first moment of value, another after several sessions, then a hard wall; hard walls performed significantly worse without soft ones before them. Replacing a conspicuous "Discard my progress" button with a subtle "Later" was itself a measurable win, though no figure was published ¹. "Save your progress" frames sign-up as protecting something the user now owns, the sunk-cost effect Growth.Design points to in Duolingo's streak and wager mechanics ⁴.

Step 7b: The notification ask

Soon after sign-up (or after the first lesson, if sign-up was deferred) Duolingo asks to send reminders, tying the request to the goal the user just set. Only after an in-app "yes" does the iOS system dialog appear.

Principle: pre-permission, with a reason. The reminders are worth the ask: Duolingo's 2020 bandit paper reported that optimising reminder wording lifted total DAU by 0.5% and new-user retention by about 2% over a strong baseline ⁶, and earlier tests found a reminder roughly a day after the last session worked best ⁸.

Step 8: The soft paywall

Only now does the subscription appear. The screen presents Super Duolingo (no ads, unlimited hearts, personalised practice) with a free trial and an annual plan, and a plain "No thanks" at the bottom. Dismissing it costs the user nothing; the course remains fully free, as the App Store listing promises ⁹. Later triggers, such as running out of hearts, re-surface the offer when the limitation is actually felt. Appcues describes the trial as a two-week reverse trial that steps the user back down to free when it ends unconfirmed ³.

Principle: connect the paid benefit to a constraint the user has already felt. Screens Design's teardown phrases it as paid benefits connecting to "experienced constraints" ⁵. Duolingo's bet is that conversion follows habit: former CPO Jorge Mazal wrote that users with longer streaks are much more likely to subscribe ². The company reported 10.9 million paid subscribers at the end of Q2 2025, 9.0% of trailing-twelve-month MAU ⁷, a high rate for a mostly free consumer app with the gentlest paywall timing in its category.

How this shows up in the charts

Duolingo holds #1 in Education on the US iOS free chart with 5.5 million ratings and a 4.7 average ⁹. Because the first lesson and the account wall convert a high share of installs into day-2 users, each download is worth more to Duolingo than to a competitor with a sign-up-first flow, which lets it outbid them for the same users and keep the rank. See where Duolingo and the rest of the category sit today on the iOS charts and Android charts, and compare with the language apps moving in this week's US iOS movers.

What to copy, and what not to

Copy the order: a few taps of profiling, a user-set goal, a real win, then the account, then money, with every ask dismissible on first contact ^1,3. Copy the framing: "save your progress" beats "sign up" because it names something the user already has.

Do not copy surface features without the system underneath. The questions work because each changes the next screen; the late paywall works because the streak gives people a reason to pay later. A placement test bolted onto an app that cannot act on the result is just a harder sign-up form.