Growing a romance reading app from 100K to 1M+ downloads in under a year

Background

FlingFM is a romance audio and reading app built for women: a personal retreat to unwind with stories from tender romance to steamy fiction, in both text and audio formats.

When I joined, the app had 100K downloads. For the first 8 months I was the sole product designer, owning end-⁠to-⁠end UX across two parallel apps, FlingFM (in-⁠app currency) and FreeFM (ads), each testing a different monetisation model. Later, a second designer joined and I onboarded them, sharing context, product knowledge, and established workflows.

I joined with a clear brief: bring in more users and keep the ones we already had.

The brief came from the product owners, and every decision on this project traces back to those two goals.

Problem

Users weren’t sticking around. The brief was growth and retention, and the data showed three problems standing in the way.

1. 1

   People left in their first session.

   About 30% dropped off during or right after onboarding. They’d browse briefly and leave without engaging.

2. 2

   Sessions were too short to become a habit.

   The users who stayed weren’t spending enough time in the app to form one.

3. 3

   Stories didn’t feel safe or personal.

   People started stories and didn’t come back. Discovery was generic, intensity could arrive as an unwelcome surprise, and the app wasn’t built for something as private as late-⁠night romance reading.

How do you design a product that feels like a trusted, personal ritual?

Process

I started by gathering stakeholder interviews, user feedback, store reviews, and product metrics. We used Amplitude for behavioural analytics and internal dashboards for business KPIs: revenue, subscriptions, content performance by cohort.

What users told us

Desk research, competitor analysis, and validation conversations filled in the why, and one realisation changed everything.

For most users, FlingFM was a wind-⁠down ritual, 20–30 minutes before sleep.

• Pre-⁠sleep is the key moment: the experience had to support that emotional arc
• Friction kills comfort instantly: interruptions feel disproportionately frustrating at night
• Privacy is non-⁠negotiable: discreet notifications and neutral UI increase trust
• Unexpected intensity damages trust: mismatched spiciness means users don’t come back
• It’s ritual, not completion: users seek controlled emotional closeness, not finishing books

From insights to hypotheses

Research gave us more than 10 hypotheses, each tied to an insight. A few of them:

• If users pick themes and a spice level during onboarding, their first stories will match what they came for, and fewer will leave in the first session.
• If stories work offline, buffering won’t break the bedtime moment, and sessions will run longer.
• If users can chat with book characters, they’ll come back more often.
• If the app picks stories by mood, browsing will take less effort and users will start more stories.

We tested them in sessions and A/B tests. The first two held up. The last two didn’t.

A/B testing as a continuous lens

We tested throughout (tabbed home screen vs. unified feed, book cover variations) and compared behavioural data across both apps (retention, session duration, ARPU, churn) to see how monetisation context shaped engagement.

What didn’t work

Not every hypothesis survived contact with real users:

• “Chats” with book characters: sounded appealing, but in sessions users found it distracting from their reading ritual
• Mood-⁠based auto-⁠curation (“Play Something Soft”): users rejected it. They wanted low-⁠effort browsing, but they still wanted to choose
• Trope filters / “Not for me”: no measurable retention impact. Liked in concept, unused in practice

These failures taught us that in this domain, agency and control are the product itself.

Solutions

Every solution answers one of the three problems, and through them the brief: keep people longer and bring them back.

Reimagined onboarding. Three steps: pick themes, choose a spice level, done, all changeable later. People’s very first stories match what they came for. First-⁠session engagement improved and early drop-⁠off decreased, so more new installs turned into readers.

Offline Mode. Stories download for reading and listening. Buffering used to break the bedtime moment right when people were settling in. Nothing interrupts the wind-⁠down now, so people stay in the story longer.

Text-⁠to-⁠Speech dual mode. Narration with synced text highlighting. People too tired to read can keep following the story instead of closing the app, which keeps the nightly session going.

Trigger warnings. Content warnings in book descriptions. People know what’s inside before they start, so an unwelcome scene doesn’t end their relationship with the app.

Narrator switching with preview. Audio previews before choosing a voice, like choosing who reads you a bedtime story. People pick a narrator they like upfront, so a voice doesn’t become a reason to drop a story halfway.

Parent Mode. App lock and neutral push notifications. Mothers could hand their phone to their kids without worry, so keeping the app on a family phone stopped being a risk.

The tension was UX integrity vs. ad revenue. Our foundational principle (no interruptions) directly conflicted with FreeFM’s ad model. We iterated: ads only between chapters, skippable formats, buffer screens so transitions felt intentional. A constant negotiation between UX and business. One that never fully resolved, but forced us to be creative about where monetisation could live without breaking trust.