• EdTech
  • Mobile

10 Minute English

Work on an existing English‑learning app close to Bright in its product model: 20+ locales, iOS/Android.

Product
10 Minute English
Role
Product Designer
Work period
2022‑2026

Product Context

10 Minute English is an English‑learning app that is close to Bright in its product logic: short daily sessions, vocabulary growth, word repetition, and a steady return to already learned material.

When I joined the product, the app already existed. Changes had to fit into a live system: 20+ locales, iOS and Android, existing navigation, subscription logic, analytics, content infrastructure, and current user habits.

Feature 1. Books

Books grew step by step: an MVP with catalog, reading, and book vocabulary; access and upsell; then comfortable reading UX with audio and settings.

Hypothesis: if adapted books are integrated into the familiar 10 Minute English mechanics, reading can become a new subscription motivator and give vocabulary real context.

Full Books board: flows, audio, settings, practice, onboarding, upsell, and access.
Full Books board: flows, audio, settings, practice, onboarding, upsell, and access.
Key screens: upsell, catalog, book details, and reading with translation.
Key screens: upsell, catalog, book details, and reading with translation.
Appearance settings: text size, reading modes, and state checks.
Appearance settings: text size, reading modes, and state checks.
Text‑size and layout‑density checks for different reading settings.
Text‑size and layout‑density checks for different reading settings.

Iteration 1. MVP: From Catalog to Reading

The first pass assembled the feature skeleton: bottom navigation entry, catalog, book details, reading, word translation, saving, and practice from book vocabulary.

PartHow It WorkedWhy
EntryA dedicated Books tab in the bottom navigationMake the new value visible without rebuilding the whole navigation
CatalogA book list with type, level, duration, genres, and reading progressHelp users quickly pick a book for their current level
Book detailsCover, description, word count, estimated time, level, and a continue actionGive enough context before a longer action
ReadingA basic text screen with translation, progress, and saved wordsValidate that reading connects to learning instead of living separately
PracticeA list of book words and a workout entry for that materialBring new words back into the core repetition loop
Learning flow: word translation in context, saving, saved word list, and training entry.
Learning flow: word translation in context, saving, saved word list, and training entry.

MVP outcome: the team got an end‑to‑end prototype where users choose a book, read, translate words, save them, and return to practice.

+20% word‑view to add‑for‑study conversion
+31% more words added for study per reading episode
+0.9% book‑based workout completion
25‑26% of users started practice from books

Iteration 2. Access, Upsell, and First Launch

In the second iteration, Books was packaged as subscription value: promo screen, upsell, first launch, and states for users with and without access. The upsell appeared here; later it was refined once for audiobooks.

  • For users without access, the section worked as a value preview rather than an empty locked tab.
  • The upsell explained the purchase: reading, books, and the connected learning scenario.
  • The first launch explained why books appeared and how they connected to learning.

Outcome: Books got a clear launch frame and became a separate argument for subscription value.

+10.6% ARPU per install in the upsell A/B test
9.59% of users purchased the Books upsell

Iteration 3. Books UX: Reading, Audio, and Settings

The third iteration turned the basic reading screen into a full learning environment. The focus moved from simply providing content to making the process comfortable: audiobooks, text settings, quick sentence translations, and a link between reading, listening, and saved words.

Users could switch between reading and audio without losing context, and the section became clearer for people who wanted to work with text, not just learn isolated words.

Reading and audio as two modes of one flow.
Reading and audio as two modes of one flow.

Outcome: Books became closer to a learning flow, not just a library inside the app.

+79% more users started reading at least one book
-7.3% fast trial cancellations in the first 3 hours
+31% more words added per reading episode
+51.2% Advanced trial‑to‑subscription conversion

Books Outcome

The main point was not a text screen, but new product value: access, catalog, comfortable reading, saved words, and a connection to practice.

The baseline launch confirmed interest in reading and book vocabulary. Segmentation added nuance: advanced users read more actively, but they were a small segment; scale depended on engaging beginners, simpler A2 books, and softer access. The upsell could be monetized separately, but only without the feeling of a surprise charge.

Feature 2. Perfect Streak

Perfect Streak appeared in three steps: a base counter with progress and calendar, then lives for recovery, then goal selection and anniversary days.

Hypothesis: if daily consistency becomes visible and users get a softer way to save progress, consistency will become clearer and subscription value more visible.

Feature board: launch, regular flow, broken streak, calendar, localization, and adaptive layouts.
Feature board: launch, regular flow, broken streak, calendar, localization, and adaptive layouts.
Onboarding and rules: missed days, goal selection, and lives as parts of the overall flow.
Onboarding and rules: missed days, goal selection, and lives as parts of the overall flow.
Progress and calendar: the base version with counters, later with lives and recovery.
Progress and calendar: the base version with counters, later with lives and recovery.
Explaining the streak mechanic inside the product.

Iteration 1. Base Streak, Progress, and Calendar

The first iteration launched consistency without lives or goals: explanation after a workout, first streak day, daily result, progress, and calendar.

PartHow It WorkedWhy
EntryAfter a workout, users saw an explanation of the new mechanic and received the first streak dayIntroduce the mechanic after the user had already completed a valuable action
Daily resultShowed the current streak day, weekly calendar, and a cue to return tomorrowMake consistency visible inside the daily loop
Progress and calendarLesson counters, completed days, streak status, and the calendar lived in the progress sectionShow that the streak did not disappear after one result and was connected to broader progress
Missed‑day ruleDedicated screens explained that a missed day would break the streakReduce the risk of an unpleasant surprise after the first missed day

Result: the streak got a clear start. Users saw the new counter, understood what made it grow, could check the calendar, and learned the missed‑day rule upfront.

+20.4% more workouts per user
+19.7% more study days per user
+0.9% ARPU among users who practiced
+17.4% more progress visits among active users

Iteration 2. Lives and Streak Recovery

In the second iteration, we added lives: if users missed a day, the streak did not always reset — it could be saved, and new lives were earned by completing all daily tasks.

  • A missed day became a managed scenario, not a sudden punishment.
  • Lives connected recovery to completing all daily tasks.
  • The calendar gained saved days and current streak status.

Outcome: motivation became less fragile while the value of daily consistency stayed clear.

+58% average streak length after lives were added
+2.9% ARPU per active trial after lives were added

Iteration 3. Streak Goal and Anniversary Days

The third iteration added goal selection and anniversary days. Users chose a target, while milestone days received a dedicated celebration.

A regular streak day and an anniversary day with a stronger celebration.
A regular streak day and an anniversary day with a stronger celebration.
Anniversary streak day animation.

The outcome was mixed: ARPU increased, but the goal‑selection screen added friction to the daily flow.

+3.4% ARPU per active trial after goal refinements
-4.4% completion of any day after the first

The hypothesis behind the lower day completion: goal selection created an extra step at the moment users wanted to finish the daily loop. For some users it strengthened commitment, but for advanced users and people under higher load it could feel like extra pressure.

Perfect Streak Outcome

In the end, the streak became a motivation system around daily learning: result, calendar, lives, goal, anniversary days, pushes, and a clear missed‑day flow.

The base version statistically strengthened consistency, lives extended streak length, and goals showed a financial lift at the cost of behavioral friction. This reinforced the segment takeaway: motivational layers work best when they support the mass user instead of adding pressure for stronger learners.

Feature 3. Training Streaks

Training Streaks increased motivation inside the exercise: consecutive correct answers grew a counter, changed progress, and created a short challenge loop.

Hypothesis: if a correct‑answer streak is shown during the workout, repetitive exercises will feel livelier and harder to abandon.

Feature board: flow, progress states, praise, illustrations, adaptive layouts, and RTL.
Feature board: flow, progress states, praise, illustrations, adaptive layouts, and RTL.

What Was Built

The feature was built into the existing workout without a separate mode. The counter grew after correct answers, the progress bar changed state, and a strong streak triggered visible praise.

ElementHow It WorkedWhy
CounterShowed the number of correct answers in a row during the exerciseMake progress visible before the final screen
ProgressColor and a short effect changed as the streak grewCreate a sense of increasing value without adding a new control mechanic
Final praiseThe result screen depended on streak strength and workout successCreate emotional payoff without complicating the workout
Start, exercise, and final screen after a run of correct answers.
Start, exercise, and final screen after a run of correct answers.
State and final‑praise variations.
State and final‑praise variations.
How the counter and progress react inside the workout.
Enhanced praise animation after a successful streak.

Results

The feature did not require new behavior: it strengthened the same workout flow. In the first test, we observed a positive financial trend, while v1.1 statistically reduced new‑word workout cancellations and increased total workout volume.

+15.8% more workouts per user in the second version
-10.5% new‑word workout cancellations
+0.2% repetition workout completion

The v1.1 takeaway: the feature barely affected repetition, but it clearly kept users inside new‑word workouts. Segmentation was again more useful than the average metric: beginners contributed better financial lift, while advanced users dropped 10.1% in subscription conversion without statistical confidence.

Feature 4. Word Repetition

Word Repetition was an attempt to update a core workout: instead of calmly going through words one by one, we tested matching pairs, a timer, progress, and several result levels.

Hypothesis: a short matching challenge with a timer would make repetition less monotonous and improve retention.

Experiment board: old flow, new version, adaptive layouts, RTL, and completion states.
Experiment board: old flow, new version, adaptive layouts, RTL, and completion states.

What We Tested

We tested it in one pass: old direct flow versus a new matching mechanic. The new scenario added pairs, a timer, mistakes, result grading, and retry.

VersionWhat ChangedRisk
Old flowOne word at a time, clear validation, simple completionCan feel boring with a large word set
New flowMatching pairs, timer, mistakes add time, graded resultMore cognitive load and less calm focus on the word
Old flow: counter, translation choice, success/failure, and completion.
Old flow: counter, translation choice, success/failure, and completion.
New flow: goal, matching pairs, mistakes, timer, and final grading.
New flow: goal, matching pairs, mistakes, timer, and final grading.
Early drag‑and‑drop concept
Final prototype with pair selection and timer.

Why It Did Not Work

The new flow looked more active, but became heavier: users watched the timer, searched for pairs, remembered mistake rules, and repeated words at the same time.

  • The timer added pressure where a calmer repetition mode was needed.
  • Matching tested search speed almost as much as word knowledge.
  • The flow required more states and rules, but did not create enough clear value.
  • 48.9% of cancellations happened on the preview: users saw pairs, the timer, and the mistake penalty before starting, then left the workout.

Takeaway: not every more interactive solution makes learning better. Sometimes the older, calmer scenario solves the user's task more precisely.

-7% ARPU per install in the test group
-17.6% repetition workout completion

Diagnostics pointed to cognitive overload: median duration grew by 247%, error conversion increased by 127%, and almost half of cancellations happened before the first in‑workout action.

Outcome as a Learning

This version was not worth developing further. It helped separate useful motivational layers from changes that break workout clarity.

Main takeaway: word repetition should be improved through small feedback layers, not a full scenario replacement.