← Work/2021 – 2022 · Product Designer, research

iStoria

An English-learning app built on graded stories, now with more than 5,000,000 learners. I redesigned its onboarding one step at a time: explain first, ask later, measure each step.

In 60 seconds

Role
Product Designer, research
Timeline
2021 – 2022
Team
3 from stakeholders and product, 2 developers and 2 from marketing

The challenge

Only 45% of installs became users and 8% of users became lead subscribers, and registering took about 15 minutes. The first screen had too much text and no explanation of how the app works, and later steps asked too much too early.

The insight

Onboarding asked for too much before it explained anything. Mapping the flow gave each step its own time and drop rate, which turned one problem into five that could be changed and measured one at a time.

What I did

Explain first, ask later. Heat-maps and recordings showed what people did, interviews and moderated usability tests showed why, and each change was measured against its own step.

45 → 80%

Installs that became users

Analytics · Quantitative

8 → 40%

Users that became lead subscribers

Analytics · Quantitative

15 → 5 min

To register, with social sign-up

Analytics · Quantitative

36 → 12%

Skipping the placement test, redesigned after moderated usability tests

Analytics · Quantitative

The evidence chain

  1. 01

    What people did

    Heat-maps and session recordings, with time and drop rate for every step.

    Analytics· Quantitative
  2. 02

    Why they did it

    One-to-one interviews and moderated think-aloud usability tests.

    Interviews· QualitativeUsability test· Mixed
  3. 03

    Change one step

    Each change was released and measured against its own numbers.

  4. 04

    What moved

    Analytics before and after each change, and an A/B test of the subscription model.

    Analytics· QuantitativeA/B test· Quantitative

The mapEvery step, with its time and drop rate

Five problems, not one long one.

Mapping the flow gave each step a time and a drop rate: why learning English 10%, placement test 30%, set reminder 30%. Each could then be changed and measured on its own.

BeforeThe flow as it was
AfterThe flow after the redesign

01 of 05

Sign-up

The first screen was all text, and no one knew what Get Started meant. It now shows what the app does, then offers Email, Apple or Google.

Before1 screen
After2 screens

What improved, part by part

The first screen

The logo and a line of text, with Get Started at the foot.

A visual screen: Improve your English and increase your vocabulary, a phone showing real stories, and words such as Gain, beautiful and Secretly.

The button

Get Started, which people did not understand.

Create new account, with Already have an account? Sign in beneath.

Signing up

A long, text-heavy form.

Sign up with Email, Apple or Google, and Support and FAQs buttons.

Explaining the app

No walkthrough.

A short how-it-works video after the learner picks the interface language, in English or Arabic, which can be skipped.

  • 45 → 80%

    Installs becoming users

  • 15 → 5 min

    Time to register

02 of 05

Why are you learning English?

Every line had to be read and the touch targets were hard to hit. It gained icons, larger rows and a cue for how many reasons to pick.

Before1 screen
After2 screens

What improved, part by part

The list

Plain lines of text with a radio on each.

An icon beside each reason, in larger rows.

How many

Nothing said how many to choose.

Select 3 reasons, under the title.

Loading

No loading state among these screens.

A skeleton of the list while it loads.

  • 2 → 1 min

    Time on the step

  • 8 → 2%

    Drop rate

03 of 05

The placement test

It took about 35 minutes for eight levels, felt endless, and lost your score if you left. It now shows progress, takes about 5 minutes to start and can be saved, resumed and retaken.

Before3 screens
After2 screens

What improved, part by part

Starting

Would you like to take a placement test?, Yes or No.

Take a test to jump ahead: it takes about 5 minutes and adapts to your level, with Start Placement test and No, Skip Test.

Progress

A question counter, such as Question 1 from 5.

A level progress bar: Test for 8 Levels of 12, and Level 3 along the top of each question.

Leaving

Leaving lost the score.

Save and resume, and a retake from inside the app.

Words

No help with pronunciation.

Audio for every word.

  • 35 → 20 to 25 min

    Time for all levels

  • 8 → 3%

    Drop rate

  • 36 → 12%

    Skipping the test

04 of 05

The daily reminder

The reminder was asked for before the learner had read anything. It now comes after the first chapter, merged with the daily goal.

Before1 screen
After1 screen

What improved, part by part

When it is asked

During onboarding: What time of day would you like to learn?

After the first chapter.

What it is

A time picker with Allow Reminder.

Set your daily goal: pick the days for practising, then the reminder time.

Why

A time every day.

Days to choose, after support tickets said that some days people have no time to practise.

05 of 05

The assistant's voice

The voice was chosen in onboarding, before the assistant had read anything. It is now a sheet that appears the first time it reads a story.

Before1 screen
After2 screens

What improved, part by part

When it is asked

During onboarding, as a screen of its own.

The first time the assistant reads a story, in the chapter.

How

Assistant Voice, with Speech, Language Accent and Voice, then Next.

A sheet over the chapter with the same choices and Continue Learning.

Later

Reachable from onboarding only.

Reachable later from the chapter or from the app settings.

How we knewEvidence

Everything was tested in iStoria: observed, changed, then measured against the step's own numbers.

Analytics

Quantitative

45% → 80%

Installs becoming users, with users becoming lead subscribers rising from 8% to 40%.

Analytics

Quantitative

15 → 5 min

Registration time, from the long form to social sign-up.

Usability test

Mixed

Moderated tests where learners completed the placement test while thinking aloud, with notes on where they struggled and where they dropped off.

Interviews

Qualitative

One-to-one interviews with learners who had taken the placement test, and feedback from support tickets that some days they have no time to practise.

Analytics

Quantitative

Heat-maps and session recordings, used to balance the numbers with what people actually did on each screen.

A/B test

Quantitative

A paid plan with a free trial against a free plan with ads and paid-only features. The trial converted better first; the free plan with ads retained better.

DecisionsAnd why

What I saw, what I knew, what I chose, and what I gave up.

01

Explain the app before asking for an account

What we saw
The first sign-up screen was text-heavy and visually flat. Many people did not finish registering on their first visit.
Evidence
Analytics· QuantitativeA high drop-off between first open and the start of sign-up, with returning users engaging more.
Interviews· QualitativePeople did not know what 'Get started' meant, called the screen boring, and wanted to see how the app works before signing up.
So we
A skippable how-it-works video in English or Arabic, clearer copy, a visual first screen, social sign-up, and Support and FAQ buttons.
Not
A longer form with more text.
What happened
Installs becoming users rose from 45% to 80%, and registration fell from about 15 minutes to 5.

02

Let people scan the goals screen

What we saw
Learners had to read every line to choose, the touch targets were hard to hit, and nobody knew how many reasons to pick.
Evidence
Analytics· QuantitativeHeat-maps and recordings showed the struggle with the touch targets and the text.
Interviews· QualitativeUsers said the text-heavy screen was boring and wanted clearer instructions.
So we
Icons, larger touch targets, and a 'select 3 reasons' cue.
Not
Leaving the list as plain text.
What happened
Time on the step fell from 2 minutes to 1, and its drop rate from 8% to 2%.

03

Make the placement test something you can see through, and come back to

What we saw
It took about 35 minutes for eight levels, felt endless, and lost the score if you left.
Evidence
Usability test· MixedModerated think-aloud sessions found unclear progress, lost scores on exit, no retake, and no help with pronunciation.
Analytics· QuantitativeDrop-off at different stages of the test and time spent per question.
So we
A level progress bar, 'about 5 minutes', save and resume, a retake from inside the app, and audio for every word.
Not
A long test with no progress and no way back.
What happened
Time fell from 35 minutes to 20 to 25 for all levels, drop rate from 8% to 3%, and skipping the test from 36% to 12%.

04

Ask for reminders and voice only when they matter

What we saw
The reminder and the assistant's voice were asked before the learner had read anything, adding time to onboarding.
Evidence
Interviews· QualitativeSupport tickets said that some days people do not have time to practise.
So we
The reminder moved to after the first chapter and merged with the daily goal, with days to choose. The voice sheet appears the first time the assistant reads a story.
Not
Asking everything up front.
What happened
Both settings remain reachable later from the chapter or the app settings.

TestingWhat was tested

Everything on this page was tested in iStoria: moderated usability tests, interviews, heat-maps and recordings, analytics before and after each change, and an A/B test of the subscription model. iStoria now has more than 5,000,000 learners. Participant counts are not published on this page.

© 2026 Alhasan HosniHome