Growth Experiments
Growth UX designer
Role
Jun 2019 - Mar 2020
Period
UX and UI Design, leading ceremonies, creating and structuring hypotheses, planning and reviewing experiments, designing experiments.
Responsibilities
eduK is an edtech that provides skills-based learning for individuals seeking their own ways of making money.
Creating an GH operation
Operation consisting of 1 designer (myself) + 3 developers.
We conducted experiments using Growth Hacking methods to create and prioritize initiatives, following best practices with statistical significance.
The subscriber base was not growing, which could eventually make the business unviable. We launched an initiative to run experiments targeting different stages of our product funnel: conversion, retention, and revenue.
Problem Statement
New Checkout Funnel
53% of users dropped out of the subscription conversion funnel between the redirection to login and the moment they were logged in/registered.
Problem
Hipotesis
Allowing login without leaving the flow for another page can increase the number of conversions. We will be satisfied if we achieve up to a 20% increase.
#Experiment case
after
Experiment process
I added the knowledge I already had about conversion funnels with tests on various types of checkouts, heuristic analysis on the best cases, and then applied these learnings to define the new information architecture.
Although I designed the architecture from scratch, I used the existing component library at eduK to save time, knowing that some interactions would need to be improved after validation.
I modified the current version of the checkout, incorporating login as a step in the payment funnel for course and subscription purchases. I presented two potential proposals to the team.
1
2
3
Discovery
Design
Ui / Visual
4
I conducted a test with 5 people before moving on to development. They provided small insights to refine some details without major changes to the proposed design.
Usability interviews
5
We defined two clusters, and one of them started receiving the new checkout. I created the hypothesis by listing statistical relevance, sample size, and the desired improvement estimate, with a goal of 20%.
The experiment took 2 weeks to reach the necessary sample size.
A/B Testing
Usability test with Rosa
Web mobile New Version
Outcomes
The new version converted 40.5% better in the login/registration steps during checkout and has a 98.8% chance of outperforming the previous version.
+40,5%
Improvement over baseline
Results board
Insights
Testar sempre gera bons insights
Este caso reforça a ideia de que validar rápido, errar rápido e pivotar quando preciso é o melhor jeito de criar serviços ou produtos, muitas vezes precisamos de pouco tempo e recursos para chegar a conclusões importantes para o negócio.
#1
Processos precisam ser facilitadores
O processo de compra dificulta a rotina do atendente e facilita erros na compra, especialmente em momentos de pico. Aqui entendemos a necessidade de explorar um processo que cause menos atrito para garantir o sucesso da operação.
#2
