vNEXTCreative intelligence for independent artists
Your audience,
written down.
SonicPush is a system that learns who you're for — then quietly grows that audience while you make music. It tests creative on your behalf, watches what resonates, and writes you a brief every Monday morning.
The BriefTasteLibraryVariant LabPatternsMapLive
ExampleJune Marlow · ACCT 00428JM
Week 47 · Nov 18–24
The brief.
Written by the system at 06:00, in your timezone. Read in three minutes.
Sources synthesized43 variants · 12 cells · 8 patterns
Confidence0.81 — High
PreviousWK46 · WK45 · Archive →
This week the system became more confident about who you're for. The bedroom-indie cluster in the UK began moving the way the NA cluster moved six weeks ago — same listening habits, same save patterns. We started colonizing it.
Variants spawned
23
+8 vs last week
Winners
6
+2 vs last week
New territory
T-04 UK
expanding state
Cost / follow
$0.42
↓ from $0.51
Audience added
+11.4K
compounding
02
Where you're becoming known
Cluster T-04 (sad-girl pop · UK · 18–26) is expanding 11.4%/week. The system spawned three probe variants on Tuesday — by Thursday one had reached scaling threshold at $0.31/follow. This audience wants what your NA audience wants: nighttime visuals, slow intros, captions that feel rather than instruct.
T-04Sad-girl pop · UK · 18–26+11.4%/wk
Illustrative weekly brief · numbers shown are example projections, not guaranteed outcomes
01 / Mechanic
The five inputs the algorithm actually wants.
Meta's optimizer isn't a black box — it's a learning system with known appetites. Volume of creative. Fast signal. Conversion-event clarity. Compounding audiences. Daily decisions. SonicPush feeds all five, every day. That's the entire mechanic.
01
Creative library
Variants generated from your assets.
We pull artwork, reels, lyric snippets, performance clips. The system composes 50+ creative variants per week — different hooks, copy, aspect ratios, soundtracks.
02
Test cells
Small, parallel, cheap.
Each variant gets a $5–10 test cell against a controlled audience seed. Cells run 48–72 hours. We measure cost-per-Spotify-follow, not vanity reach.
03
Winners scale
Budget concentrates automatically.
Variants that beat threshold cost-per-follow get scaled with bigger budget and broader audiences. Reallocation runs on Meta's optimization learning window, informed by rolling 7-day performance.
04
Losers killed
No sentimentality.
Variants that miss threshold get paused. Their learnings feed back into the next creative batch — what hook patterns worked, which audiences responded, which formats didn't.
05
Audience compounds
Each test makes the next one cheaper.
Custom audiences (engagers, page-viewers, Spotify-clickers) compound in your ad account. Lookalikes get sharper. CPM drops. Cost-per-follow trends down month over month.
Avg variants / wk
53
across active SonicPush accounts · [Phase 1 cohort]
Test cell cost
$5–10
before a winner gets scaled
Review cadence
24h
action paced to Meta’s learning window
Cost / follow
$0.40
target after 60 days running · [Phase 1 validating]
02 / Premise
What the algorithm rewards. And what it punishes.
Most artists who run their own Meta ads optimize the wrong thing. They tune audiences, sweat targeting, agonize over interest categories. The algorithm doesn't care. Here's what it actually responds to.
i.
Rewards: creative diversity.
The optimizer tests creative against many audience slices in parallel. With one ad, it has nothing to compare. With fifty, it builds a signal map of what works for whom. The audience finds the artist through the creative — not the other way around. This is the single most under-fed input in indie music ads.
VARIANT 001VARIANT 050
ii.
Rewards: compounding custom audiences.
Campaign-based ad management resets the optimizer every launch — it has to re-learn your audience from cold. Always-on lets engagers, video-viewers, and link-clickers grow continuously. Each becomes a lookalike seed; each new test starts warm. Six months in, your reach is cheaper because the algorithm already knows who responds.
MONTH 1MONTH 12
iii.
Rewards: a clean conversion event.
Reach, impressions, CTR — all proxies. Meta's optimizer gets sharper the closer your conversion event sits to actual value. We fire the pixel on the Spotify destination and optimize on cost-per-Spotify-follow directly. The algorithm stops chasing scrollers and starts hunting listeners. Everything else is noise we ignore on your behalf.
iv.
Rewards: decisions in 24 hours.
Manual ad management runs on weekly review cadence. The algorithm learns much faster than that — and starves between reviews. Variants that miss threshold cost-per-follow get paused inside a day. Budget moves on Meta's optimization rhythm — fast enough that the algorithm never goes cold between checks. By the time a human would write up a “let's optimize” memo, we've already moved through several cycles.
03 / Trigger
Why this triggers Spotify's algorithm — when playlist pitching doesn't.
Streams are an output. Spotify's recommendation engine is a behavioral system that watches what listeners doafter they click — save, follow, replay, finish, share. Paid placements buy the click. They don't buy the behavior. That's why the streams flatline the day the placement ends.
Pattern A · Playlist placement
Streams spike, then fall off a cliff.
WEEK 1DECAY → ALGORITHM IGNORESWEEK 12
Pattern B · SonicPush always-on
Listener behavior compounds. Algorithm picks up.
WEEK 1SAVES · FOLLOWS · REPLAYS → LIFTWEEK 12
i.
Spotify is a behavioral system, not an inventory.
Discover Weekly, Release Radar, and the personalized algorithmic mixes don't see “streams.” They see saves, follows, completion rates, repeat listens, share-clicks. Those are the signals that move you into recommendations — and they only fire when the listener arrived for you, not for a playlist.
ii.
Playlist placements deliver passive listens.
A listener on a curator playlist isn't choosing your track — they're listening to the playlist. Spotify reads that as low-intent. They don't save, don't follow, don't replay. The day the placement ends, the streams end. The algorithm files you under “played in passing.”
iii.
Meta ads deliver intentional arrivals.
Someone who saw your ad, scrolled past two others, came back, tapped through to Spotify and pressed play — that's the highest-intent signal Spotify can read. It's the listener telling the algorithm: “this is the kind of thing I want.” Stack thousands of those, daily, and the recommendation engine starts doing your distribution for you.
04 / SetupSetup is fifteen minutes.
Setup is fifteen minutes.
Most of it is reading the screen.
You stay the owner of your Meta ad account. You stay the owner of your audience data. We build the system inside it.
01
Connect Meta
Grant SonicPush partner access to your Meta ad account. Don't have one? We walk you through the five-minute setup. Either way, the account stays in your name.
02
Connect Spotify
Link your Spotify for Artists account. The pixel fires on streams, follows, and saves — the events the algorithm actually uses to decide what to surface.
03
Upload your assets
Tracks, artwork, reels, performance clips, lyric snippets. Whatever you have. The Engine composes 50+ variants from your creative library — different hooks, formats, aspect ratios, soundtracks.
04
We start testing
First test cells go live within 24 hours. First decisions inside 48. First scaled winners by end of week one. You don't push a button.
15 MIN · NO CALLS · NO CONTRACT · STAY THE OWNER
05 / Focus
What we don't do. On purpose.
Most music marketing tools widen — more channels, more dashboards, more line items. We narrowed. Every channel below has a real argument for inclusion. We left them all out for the same reason.
Channel · 01
TikTok's optimizer rewards sound, not destination. The path from a paid TikTok view to a Spotify save is too long, with too many drop-offs, for our cost-per-follow model to work cleanly. Organic TikTok matters — paid TikTok doesn't, for what we're solving.
Channel · 02
Audio ads inside Spotify reach existing Spotify listeners — the audience you already have access to organically. They don't pull new listeners into the platform with intent. The math doesn't compete with Meta-to-Spotify on cost-per-follow.
Channel · 03
A dashboard running Meta + TikTok + YouTube + Snap simultaneously sounds powerful. In practice it splits creative attention four ways and starves each algorithm of the volume it needs to learn. We picked one channel and overfed it.
Channel · 04
Pay-to-play playlist services and submission marketplaces sell access to curators. The streams that result don't trigger Spotify's recommendation engine — see Section 03. We don't run pitching, integrate with pitching tools, or recommend it as a complement.
One channel. One metric.
One reason: it's the only thing that compounds.Most artists disappear between releases. SonicPush doesn't.
One reason: it's the only thing that compounds.Most artists disappear between releases. SonicPush doesn't.
SonicPush v2 · product principle
06 / Pricing
Two tiers. Same engine. The bigger one runs more variants.
Subscription buys the system. Ad spend goes from your card to Meta directly — typical artists run $10–30/day to start.
14-day trial · No card · Cancel any time
Tier 01
Indie
$63/ month
+25+ creative variants tested per week
+One Spotify artist profile
+Live dashboard · cost-per-follow optimization
+Custom audiences in your ad account, forever
+Email support · <24h response
Most picked
Tier 02
Pro
$103/ month
+50+ creative variants tested per week
+Multi-track / multi-release support
+Lookalike audience expansion
+Quarterly strategy review with the team
+Priority support · <4h response
+Everything in Indie
What this replaces
Music marketing agency · typical
$1,500–3,000 / month retainer + 10–15% of ad spend. 2–3 weeks of onboarding calls. 3–8 creative variants tested per week. Reports every two weeks. Lock-in: 3–6 months minimum.
SonicPush · pro
$129 / month flat. 15-minute setup. 50+ variants tested per week. Live dashboard. Cancel any time.
Annual cost · agency vs. SonicPush: $27,000 avg vs $1,548.
07 / FAQ
Questions an engineer would ask.
Q · 01
How many ads will you actually run?
25–50+ creative variants per week, depending on tier. Each goes through a small test cell ($5–10) for 48–72 hours. Variants that beat threshold
cost-per-Spotify-follow get scaled; the rest get paused. The system runs continuously — you don't see a “campaign” structure because there isn't one.Q · 02
Do I need a budget for ad spend on top of the subscription?
Yes — and you set it. The subscription buys the engine. Ad spend goes from your Meta-connected card to Meta. You set a daily cap; we never exceed it. Most artists start at $10–30/day. Pause anytime without canceling SonicPush.
Q · 03
What happens if my ads don't convert?
The system kills them automatically. Variants that miss threshold cost-per-follow get paused at the next decision cycle (every 24h). The dashboard shows you which ones died and why — so you can see the system isn't burning your budget on creative that doesn't work. If your entireaccount isn't converting after 30 days, that's usually a creative-input problem (we'll tell you, and the trial covers the period to find out).
Q · 04
Can I see what's actually running?
Yes — every variant, every dollar, live. The Live tab in the dashboard shows every active ad with its current performance, the test cells in flight, the variants that died this week, and where today's budget is going. You can also open the ad in Meta Business Manager directly from any row. Nothing is hidden.
Q · 05
Can I override the system if I want a specific variant to run?
Yes, but try not to. Pinning a variant defeats the optimizer. We let you pin one or two for a specific reason (release date, tour announcement) — beyond that, the algorithm beats human intuition by a wide margin. We've measured.
Q · 06
What does "always-on" really mean?
The system never pauses on its own. No campaign cycles, no “between-release” gaps. New tracks slot into the existing creative library; old tracks keep running until they decay. Custom audiences keep compounding 365 days a year. You can pause manually any time.
Q · 07
Why no TikTok ads?
The conversion path is too noisy. Paid TikTok views to Spotify saves drops off at every step — sound discovery, profile click, link in bio, Spotify open, save. Our cost-per-follow model needs a tight optimization loop. Meta's pixel-on-Spotify path is the cleanest one we've found. Organic TikTok matters; we just don't run paid spend there.
Q · 08
Why no Spotify Ad Studio?
It reaches the audience you already have. Audio ads inside Spotify play to existing Spotify users — a closed loop. To grow, you need to pull new listeners intoSpotify with intent. That's a Meta job. We considered Ad Studio as a secondary channel and it didn't beat the focus tradeoff.
Q · 09
What about playlist pitching — should I still do it?
Editorial pitching, yes. Pay-to-play, no. A real editorial Spotify or Apple Music placement is a brand signal that compounds. Paid curator services deliver passive listens that don't trigger the algorithm (see Section 03). We don't run either; we just don't pretend the second category is doing what artists hope it's doing.
Q · 10
Will streams keep growing if I pause?
Some — then they decay. The custom audiences, the lookalike pool, the listeners who already saved you — those keep generating algorithmic streams for weeks. But the always-on flywheel needs ongoing creative volume. Pause for a month and the curve flattens. Pause for three and you're starting cold again.
Q · 11
How is this different from un:hurd / Linkfire / similar?
Different category. un:hurd is closer to a pitching marketplace; Linkfire is a smart-link / analytics layer. Both can sit alongside SonicPush — neither competes on cost-per-Spotify-follow optimization with always-on creative testing. If you're already on Linkfire, keep it; we'll route Meta traffic through whatever destination URL you give us.
Feed the algorithm.
Every day.
We're opening access in waves while the system learns. Join the waitlist — we'll email you the moment your spot is ready.