Suno Case Study: How a Fitness Brand Created 500 Custom Workout Tracks and Grew App Retention 28%

The Problem: Licensing Music for a Fitness App Costs More Than Development

A fitness startup with 200,000 monthly active users had a music problem. Their workout app offered guided exercises — HIIT, strength training, yoga, running, stretching — and every workout needed background music matched to the exercise intensity. Users consistently rated “music quality” as the second most important feature (after exercise content itself).

The licensing landscape was brutal:

• Spotify integration: could not be used for synced workout experiences (licensing restrictions on programmatic playback control)
• Licensed music library (Epidemic Sound, Artlist): $500-2,000/month for commercial app use, limited to their catalog, could not customize BPM or energy to match specific exercises
• Custom music production: $200-500 per track from a music producer, needed 500+ tracks for full catalog coverage, total: $100,000-250,000
• Royalty-free libraries: low quality, every fitness app used the same 50 tracks, users complained about repetition

The founder needed hundreds of unique, high-energy tracks that were:

• BPM-matched to specific exercise types (120 BPM for running, 140+ for HIIT, 70 for yoga)
• Varied enough that users did not hear the same track twice in a week
• Licensed for commercial use in the app
• Producible at a cost that did not bankrupt a startup

Suno’s Pro plan ($22/month with commercial rights) could generate unlimited tracks at the right BPM, in any genre, with no per-track licensing cost.

The Music Strategy: BPM-Driven Genre Map

Exercise-to-Music Mapping

The team created a systematic mapping of exercise types to musical parameters:

Total target: 500 tracks across 9 workout categories.

Genre Variety Within Categories

To prevent listener fatigue, each category needed multiple genre sub-options:

HIIT (80 tracks):
  - 20 tracks: EDM (festival energy, synth drops)
  - 20 tracks: Drum and bass (fast, rhythmic, driving)
  - 20 tracks: Trap/hip-hop (heavy bass, hi-hat rolls)
  - 20 tracks: Industrial/electronic rock (aggressive, dark)

Users could select their preferred genre or let the app
randomize within the workout type.

Production: 500 Tracks in 3 Weeks

Week 1: Prompt Template Development and Testing (100 tracks)

The music director (a team member with production background) spent the first week developing and testing prompt templates:

HIIT EDM template:
"High-energy EDM workout track. BPM: [specific BPM].
Powerful four-on-the-floor kick drum, driving bassline,
synth stabs, build-ups and drops. The energy stays
consistently high — no slow sections. Designed for
intense interval training. Festival-ready production
quality. No vocals. 3 minutes."

HIIT Drum and Bass template:
"Drum and bass workout track. BPM: [specific BPM, 170-175].
Fast breakbeat drums, heavy sub-bass, sharp synth leads.
Relentless energy, no breakdowns longer than 4 bars.
Electronic, aggressive but not abrasive. Designed for
high-intensity exercise. No vocals. 3 minutes."

Yoga Ambient template:
"Gentle ambient yoga track. BPM: [specific BPM, 65-75].
Soft pad textures, occasional singing bowl or wind chime
accents, subtle nature sounds (distant water, birds).
Spacious, meditative, unhurried. No percussion. No
sudden changes in dynamics. Designed for holding poses
and deep breathing. 5 minutes."

Each template was tested with 5 generations. The best 2-3 were selected, and the template was refined based on what worked.

Week 1 output: 100 tracks (testing + initial production)

Week 2-3: Batch Production (400 tracks)

With templates proven, the team entered batch mode:

Daily production schedule:
  Morning (2 hours): Generate 40 tracks (4 categories x 10 tracks)
  Afternoon (1 hour): Review, rate, and flag regenerations
  Late afternoon (1 hour): Regenerate flagged tracks, post-process

Daily output: 30-35 approved tracks
Week 2: 175 tracks
Week 3: 225 tracks

Quality Control Process

Each generated track was evaluated on:

Rating criteria:
1. BPM accuracy (does it match the target BPM?)
   Verified with BPM detection tool (Ableton, mixed in key)
   Tolerance: +/- 2 BPM

2. Energy consistency (does it maintain workout-appropriate energy?)
   HIIT tracks: no energy drops longer than 8 bars
   Yoga tracks: no sudden energy spikes
   Running tracks: steady energy, no major tempo shifts

3. Production quality (does it sound professional?)
   No audible artifacts, glitches, or unnatural sounds
   Clean mix: bass not muddy, highs not harsh
   Loudness appropriate for headphone listening during exercise

4. Distinctiveness (does it sound different from other tracks?)
   Compare against the last 10 tracks in the same category
   If too similar to an existing track: regenerate

Rating scale:
A = Ship immediately
B = Minor adjustment in post-processing (EQ, loudness)
C = Regenerate with modified prompt
D = Discard

Acceptance rates by category:

• HIIT/High-energy: 55% A/B (Suno excels at energetic electronic music)
• Strength/Hip-hop: 45% A/B (good but sometimes too repetitive)
• Yoga/Ambient: 65% A/B (Suno produces excellent atmospheric music)
• Running/Steady: 50% A/B (maintaining consistent energy is harder)

Post-Processing Pipeline

Every approved track received standardized post-processing:

Post-processing chain:
1. BPM verification and minor tempo adjustment if needed
2. Intro trim: ensure the beat starts within 2 seconds
   (users are exercising, no long intros)
3. Outro fade: 5-second fade-out ending
4. Loudness normalization: -14 LUFS (louder than music
   streaming standard, needed to be audible over gym noise)
5. Frequency balance: gentle bass boost for headphone
   listening during exercise
6. Loop point marking: identify seamless loop points for
   tracks that need to extend beyond their generated length
7. Metadata tagging: BPM, genre, energy level, workout type,
   duration, generation date

Integration with the Fitness App

Smart Playlist Generation

The app built playlists dynamically based on the workout:

Workout structure: 30-minute HIIT session
  Warm-up (5 min): 2 tracks at 110-120 BPM, building energy
  Intervals (20 min): 6-7 tracks at 140-170 BPM, high energy
  Cool-down (5 min): 2 tracks at 80-100 BPM, decreasing energy

The app selects tracks from the library matching:
1. BPM range for the current phase
2. User's preferred genre (if set)
3. Tracks not played in the last 5 sessions (no repetition)
4. Energy level matching the exercise intensity

User Personalization

User preferences:
- Preferred genres: [user selects from available genres]
- Energy preference: [higher / standard / lower]
- Vocals: [no vocals / some vocals / any]

The recommendation engine learns from:
- Skip rate (if user skips a track, lower its weight)
- Completion rate (if user finishes workouts with certain
  tracks more often, increase their weight)
- Explicit likes/dislikes (heart/skip buttons during workout)

Results After 6 Months

App Retention

The 28% improvement in Day 30 retention was the headline metric. In the fitness app market, a 6 percentage point retention improvement translates directly to lifetime value: more retained users = more subscription renewals = more revenue.

User Feedback Themes

Positive:

• “The music actually matches the workout intensity — finally!”
• “I never hear the same song twice in a week”
• “The yoga playlists are actually relaxing, not just quiet pop music”

Negative (addressed in iteration):

• “Some tracks are too repetitive” (regenerated with more variation instructions)
• “I want to save favorite tracks” (feature added in app update)
• “Can I use my own music?” (Spotify integration added for users who prefer it)

Cost Comparison

The cost difference was staggering: $264/year versus $100,000+ for equivalent custom music.

What Went Wrong

Problem 1: BPM Drift in Generated Tracks

Approximately 15% of generated tracks had BPM that drifted 3-5 BPM from the target over the course of the track. A track prompted at 140 BPM might start at 140 and end at 145. For running (where pace sync matters), this was noticeable.

Fix: Added a post-processing step using Ableton’s warping feature to lock BPM throughout the track. This added 2 minutes per track to the post-processing pipeline but ensured perfect BPM consistency.

Problem 2: Vocal Fragments Appeared Despite “No Vocals” Instruction

About 10% of tracks included unintelligible vocal-like sounds — not lyrics, but vocalizations that distracted users during exercise.

Fix: Added explicit negative instructions: “Absolutely no human voice, vocal samples, choir sounds, or vocal-like synthesizer patches. Instrumental only.” This reduced vocal artifacts to under 3%.

Problem 3: Users Compared to Spotify and Found Suno Tracks Lacking

Some users who also used Spotify during workouts felt the AI-generated tracks lacked the “polish” of commercially produced music. The tracks were good — but not Beyonce-good.

Fix: The team acknowledged that AI music occupies a different quality tier than top commercial productions. They positioned the music as “custom workout music designed for your exercise” rather than competing with mainstream artists. They also added Spotify integration for users who preferred commercial music, making the custom tracks a complement rather than a replacement.

Lessons for Fitness and Wellness Apps

BPM Is the Most Important Musical Parameter for Exercise

Users do not consciously evaluate music quality during a workout. They subconsciously respond to BPM matching their movement pace. A mediocre track at the right BPM outperforms a great track at the wrong BPM for exercise motivation.

Volume Matters More Than Variety Initially

The team could have launched with 200 tracks and grown to 500. But the initial 500-track library ensured that no user heard the same track twice in their first week — which is when retention is won or lost. Front-loading the volume investment paid off in retention metrics.

Let Users Control Their Experience

Some users love AI-generated workout music. Others want their Spotify playlists. The best approach is not either/or — it is offering both and letting users choose. The AI music is the default (works for everyone), and Spotify integration is the upgrade (for music enthusiasts).

Frequently Asked Questions

Can Suno generate music at specific BPMs reliably?

Suno follows BPM instructions with approximately 85% accuracy. The remaining 15% are close (within 5 BPM) but may need post-processing to lock precisely. For applications where BPM precision matters (fitness, meditation timers), always verify with a BPM detection tool.

Is Suno Pro commercially licensed for app use?

Yes. Suno Pro and Premier plans grant commercial usage rights for generated music. This includes use in apps, videos, podcasts, and other commercial products. Check current terms for your specific use case.

How do users react to knowing the music is AI-generated?

Most users do not care — they care about whether the music motivates their workout. In the app’s user survey, 68% said they “did not notice or did not care” that the music was AI-generated. Only 12% said it negatively affected their perception.

Can this approach work for other types of apps?

Yes. Meditation apps, focus/productivity apps, sleep apps, and gaming all benefit from custom BPM-controlled music. The same production pipeline works — adjust the genre map and BPM ranges for the use case.

How often should the music library be refreshed?

Add 20-30 new tracks per month to prevent listener fatigue for daily users. Retire the lowest-rated tracks quarterly. After 12 months, you should have 700-800 tracks with the bottom 100 retired — a constantly improving library.