100 days, 30 features: a building log
A retrospective on what we built after StudioMode's v1 ship — what hit, what flopped, what we'd skip if we did it again, and the meta-lesson about building in public.
30
Features shipped
21
Commits queued
11
Share-card types
6
Blog posts
The setup
StudioMode v1 shipped to TestFlight on Apr 18, 2026. Core experience: AI-curated daily drops, save streaks, BeatStars/Airbit marketplace handoff. From there we ran a 30-feature push to reach product-market fit — adding the stuff we'd been promising ourselves we'd build "after launch." This is the autopsy on those 30 features.
What hit (4 features that compounded)
★ HIT
Public roadmap voting
Move from "what should we build?" being internal to a public votable feed at /roadmap. Users submitted ideas we hadn't considered. Vote counts taught us what mattered. The act of seeing votes accumulate is itself engagement.
★ HIT
Shareable cards (11 types)
Building
/api/og as a single function with query-dispatched card types compounded harder than any other engineering bet. Every screen now has a beautiful preview when shared. Tarot reading, Wrapped recap, compatibility score, streak milestone — all unfurl visually on Twitter / iMessage.★ HIT
Long-form blog posts
Six posts in 100 days. The vision essay (most-linked), the algo explainer (most-emailed), the 5,000-saves data post (most-quoted), the 12-functions engineering post (most-tweeted). Worth more than 100 LinkedIn posts; SEO compounds for years.
★ HIT
BeatPlayer songwriter trio
Loop, sleep timer, playback rate. Three small player features, each ~50 lines of code. Shipped together and immediately referenced as "I can't go back" by every TestFlight user we asked. The whole-is-greater-than-sum effect; loop alone wouldn't have moved the needle.
What flopped (3 we'd skip)
✗ FLOP
/wallpapers
Built it as a fun aesthetic surface — six procedurally-generated phone wallpapers. Almost zero downloads. Nice idea, wrong audience: the people who want StudioMode wallpapers are already on the app every day; they don't need a separate retention surface.
✗ FLOP
Theme picker (5 accent colors)
Pro feature; 4% of Pro users actually changed it. We thought "personalization" was the moat. Turns out the moat is "fits the way I write." Aesthetic preference matters less than we projected. Won't deprecate, but won't build more themes.
✗ FLOP
Email digest preview page
Built /digest-preview to let users see exactly what the daily email looks like before opting in. Beautiful page. Zero measurable lift in opt-in rate. The honest interpretation: people decide on opt-in based on copy ("daily" vs "weekly" matters more than seeing the rendered email).
What surprised us
⚡ SURPRISE
Beat Tarot
Built as a novelty ("for engagement"). Turned into one of the most-shared features. The combination of "personal data + ritual + tarot card aesthetic" creates a story people want to post about. The lesson: novelty isn't always wasted scope — if it's about the user, it can compete with utility for share-of-mind.
⚡ SURPRISE
Producer badge generator
Live SVG badge ("Featured on StudioMode — 47 saves") for producer bios. Built as a producer-incentive lever; turned into our top organic acquisition surface. Every producer who drops one in their BeatStars bio creates a free funnel back. Should have built this in week 1.
⚡ SURPRISE
12-function constraint
Vercel Hobby caps you at 12 functions. We treated it as a hard limit and packed everything possible into static HTML + Supabase REST. Result: cheaper, faster, more debuggable code than if we'd had unlimited headroom. Full breakdown.
Novelty isn't always wasted scope — if it's about the user,
it can compete with utility for share-of-mind.
The meta-lesson
The 30-feature post-launch push felt like trial-and-error chaos while we were doing it. Looking at it now, three patterns emerge:
- Personal data + ritual + share = compounding. Tarot, Wrapped, streak milestones, the daily challenge banner. Anything that turns user data into a shareable artifact lands harder than utility features.
- Producer-side leverage > consumer-side optimization. The badge generator and embed builder did more for growth than any artist-side feature. Building tools your supply side wants to share is a free distribution channel.
- Constraints force creativity. The 12-function cap, the 10,000 YouTube quota cap, even the choice to stay on iOS-only at first — every constraint forced us to either drop a feature or find a 10× cheaper path. Removing constraints sounds good. It usually isn't.
What's next
Three things on deck for the next 100 days:
- Android. The React Native codebase ships there with one config flip. The reason we haven't is bandwidth, not difficulty. Probably 2 weeks of polish.
- Spotify-as-source for taste import. #1 vote on the public roadmap. Pull a user's Spotify top tracks on signup, hot-start the recommendation algorithm with that signal.
- Stem packs for Pro producers. #2 vote. Producers opt in to attach stems; Pro users get them free with a paid lease. Closes the loop between discovery (us) and licensing (BeatStars/Airbit) without competing.
Reach out if you have ideas for the next 30. The roadmap is votable.