My Product Work

Product case study · Duolingo · October 2026

Festival Cards

During festivals, learners send a friend a greeting in the language they're learning, with a free month of Super. The friend replies in that language on the web before installing anything, so the invite itself is their first lesson.

The prompt: pick a KPI and spec a new feature in the Duolingo app that would improve it, in five slides or fewer, with mockups.

Slide 1 of the Festival Cards deck: the problem, the solution, and the KPI

Problem

People learn a language to connect with its culture, but Duolingo skips that culture's festivals, and learners can't share them with friends outside the app.

KPI

Day-7 referred learners: invited friends still active a week after joining. It feeds daily active users.

Validation

Two cheap checks before building, then two A/B tests at Lunar New Year, Eid and Holi.

1

The problem and the KPI

Why festivals matter to learners, and an estimate of about 34K new learners from New Year alone, built from industry benchmarks.

Slide 1: Festivals are when the world says hello. Duolingo isn't part of them yet. Problem, solution, supporting data and the KPI.
2

How it works

The full loop from sender to new learner, a calendar of 27 festivals across 38 languages, and how it differs from Gift Super, Friends Quests and today's follow link.

Slide 2: Every greeting is someone's first lesson. The full flow from Maya to Lucía, the festival calendar, and a comparison with existing features.
3

The sender's side

Mockups of the sender's flow: a festival lesson, making the card, and sending it on WhatsApp, with each screen marked as new or existing.

Slide 3: Maya turns her lesson into a gift. Five phone mockups from the Diwali banner to sending the card on WhatsApp.
4

The friend's side

The friend replies in the sender's language in her browser, accepts the gift, and joins with her streak already started.

Slide 4: Lucía replies in Hindi before she installs anything. Five phone mockups from the web reply to the app.
5

How I'd test it

Three hypotheses with pass bars, two A/B tests with guardrails, and the tracking events from card sent to day-7 active.

Slide 5: Test cheap first, then run two A/B tests. Three hypotheses, two A/B tests, and the tracking events.