Thousands installed SwiftChat every month, then left within days. I designed a guided first bot conversation that lifted D15 retention by 66%.
Android app and PWA. Design, Product Management and Growth teams.
Despite thousands of installs each month, many users dropped off after onboarding without engaging further. This directly impacted retention.
A contextual guided conversation in place of the old app tour, shown as a popup the first time someone lands on Discover.
D7 retention up 33%, D15 up 66% and M1 up 25% against a control group that never saw the message.
SwiftChat is an AI-powered edtech platform. We envisioned an interactive space where students could learn through AI-driven conversations, quizzes, and challenges.
But something was off. Users were leaving too soon. So we took a deep dive into our user data and found some alarming patterns.
A majority of users abandoned the app within their first 72 hours, leading to a sharp decline in retention and engagement.
To validate our assumptions, we spoke to students in classrooms and at home to understand their frustrations.
“I installed the app but didn’t know what to do next.”
“I looked around a bit, but nothing felt interesting.”
“I didn’t see a reason to start chatting with the bot right away.”
Users were unsure what to do after onboarding, as soon as they landed on the Discover page.
Many users didn’t explore quizzes and challenges because they weren’t aware of the variety of features.
Users did not feel a strong motivation to start conversations with the bot.
Before UX, I was a dentist. One habit stayed with me: don’t treat the symptom. So we looked for the single behaviour that separated users who stayed from users who left.
That set the brief. Not a better tour of every feature. A guided nudge towards one first conversation, in a way that felt natural and exciting.
24.16% of new users quit within three days.
They landed on Discover, saw too many options, and nothing told them where to start.
Users who had a bot conversation within the first 3 days were 2.5 times more likely to stay until the end of the month.
Users will be more engaged if they are introduced to fun, interactive elements as soon as they onboard.
How might we get new users talking to a bot right after onboarding?
From a passive tour to one well-timed conversation. Tested against a control group that didn’t see the message.
The previous onboarding walked users through the Bot Store in four steps: bots you might like, finding bots, filters and sorting. Useful, but it asked nothing of them.
four screens, zero reasons to chat
We replaced the old app tour with a contextual guided conversation, sent as an interstitial popup when users visited the Discover page for the first time.
One clear action: start a quiz with a bot. We did not send it to all users, keeping a control group for comparison.
guide them, don’t make them exploreWe tested 2 variations of the design: one with fact-based messaging and one with FOMO-based messaging.
FOMO won
Measured against the control group that didn’t receive the in-app message.
For users who received the guided tour.
For users who received the in-app message.
Uplift for users who received the in-app message.
FOMO-based messaging was highly effective in driving engagement. Timely nudges helped users take action immediately instead of passively exploring.
What went well, and the one thing I’d carry into any onboarding.
Replacing a passive onboarding with a guided conversation increased retention and encouraged meaningful interactions.
A data-driven approach with a control group let us measure success and iterate quickly.
This experiment proved that small, well-placed nudges can have a big impact on retention.
It’s not just about getting users to sign up. It’s about making them stay, and that starts with engaging them from day one.
Want to see the full experiment, or talk through the variants? I’d be glad to walk you through it.