Part V — Release

Promotion after the record is ready

Test a record cheaply, decide from the result whether it has earned any spend at all, and know what to do with one that already underperformed.

Start at the wrong end and you will pay for it

The record exists, the metadata is registered, the accounts are open, the release is serviced. Now the question is money — and promotion is where you look last.

The diagnosis runs in a fixed order: the record, the visual, the call to action, the promotion (BC7TiuVATvM). A defect at an earlier base makes every hour spent on a later one worthless. Run it on your own release before reading further.

1. The record. A legible copy of one artist, badly recorded, or in a style that has moved on?

The fix is blunt: remix it over a new beat, or write it again.

1. The visual. Not "is it expensive" — does it carry the same feeling as the song. A visual

promising what the record does not deliver disconnects the viewer before any link.

1. The call to action. One action per post — in the post, the caption and the first comment.

The source says this is where most people fail, and it costs nothing to fix.

1. Promotion. Only now.

⚠️ That video attaches percentages to those bases — how often it is the music, how often the visuals, how much revenue is lost by stopping short. All three are stated as if measured and none is sourced. Keep the order, drop the arithmetic.

Test the record before you fund anything

This prevents spending a budget on a record nobody wanted, and the relationships along with it.

Decide in advance what the test is for — criticism, or anticipation. Without that choice the same comment section reads either way and every result confirms whatever you hoped (jrh2oibPRTw). Set the timeline and the budget up front too; that is what stops a test becoming a campaign.

Three cheap instruments, ascending in cost:

• The fragment. Post a couple of bars of something unfinished, watch which lines provoke a

reaction, write the ones that land into the finished record (vX-Kqp9XnUM). The source gates this himself and the gate is real: it works only if the writing is genuinely arresting, because the fragment has nothing else to carry it. Without that you get silence and misread it.

• A small deliberate panel. At most twenty people, inside artist communities — a cap, not a

broadcast, so the response is legible.

• Three finished-quality records as content, on one platform, across tempo. The uptempo wins

regardless, so the real question is whether anything else survives.

Two rules make the results readable. The sample must already be at sale quality, or you measure the mix and not the song. And read the content type separately from the song — a format lifts likes on its own, so a low-scoring song may be a badly framed one. Trend it; one post proves nothing.

The best question here, for the people whose opinion you most want: "Would you play this if I didn't ask you to?" Anyone will say a record is good. Almost nobody claims they would add it to their own rotation unprompted.

Write the failure branch before you run the test. Loved: ship. Mildly disliked: album filler. Truly disliked: back to the studio, and the unspent money is the win.

⚠️ The test-budget figures in jrh2oibPRTw are illustrations, summed aloud on camera and corrected mid-sentence, with no basis given. Do not budget from them.

The gates a record passes before any money moves

The sharper version uses gates rather than stages (jUVw5zddnwg). A stage is something you pass through; a gate may turn you back. A failed gate is a result, not a setback — a "no" at gate one just saved the ad spend.

The spend trigger is two independent signals, not one: DJ pickup plus climbing watch time. Gate four is not mandatory — "three out of four ain't bad" — which keeps the sequence usable by someone with no DJ relationships yet.

Two honest limits. Gate three requires actually releasing the song, so a record that fails there is already published. And "a small spend into reputable playlist tools" sits close to the paid-playlist practice he rejects a sentence earlier; he never defines reputable, and neither will we.

If you already have an audience, run the same sequence with the budget removed: your existing fans are the filter, and you service a record only once it is sticking.

⚠️ The cost claims in that video contradict each other by an order of magnitude — the opening prices validation one way, the same phase is priced ten times higher later, and larger budgets are used as hypotheticals throughout. Take the gate structure and the two-signal trigger, not a number.

Most of the money goes into the thing, not the ad

The allocation argument, the most quotable idea in this Part: of every ten marketing dollars, roughly one advertises and nine build the thing being advertised (2ExLSmSiz5Q).

The reason is the second impression. Assume the real viewer path — see the ad once, arrive at your profile, look for more of what they just saw, decide in seconds. What they find is your index: the work already on the shelf. If nothing is in the club there is no reason to come back, and running ads against an empty index makes you a false advertiser. Everyone knows a flyer flatters the night; the gap between the flyer and the room is what destroys trust.

Stated the other way, and this is the line to remember: an ad landing a curious listener on an empty profile has paid to disappoint them.

Ads buy awareness and only awareness (fGJkezf7Gt0) — the product working as designed, not failing. The waste is a category error: paying repeatedly for the shallowest layer of the funnel while the layers that convert stay empty, because paid reach cannot manufacture them. Attention, retention and conversion each need different content — talent content makes the attention, story content holds it and converts it. Hence the rule: amplify what already worked, never what might, and pair it with a call to action.

⚠️ Both videos name a budget floor at which ads start to make sense. One is contradicted by its own worked example minutes later and the transcript garbles those figures beyond reading; the other states a threshold once, in passing, with no working. There is no usable number here. Decide your floor from your own index and your own gates.

The audience paid reach assembles for a new artist is also scattered, because there is no concentrated seed for the system to model — followers everywhere, nobody at the show (67Q3LSnuBsw). Follower count cannot see that. This test can: if you played tonight, in your own city, would anyone come?

Salvage: what to do with a record that already underperformed

Some of you are not planning a release; you are looking at one that did not work. Run the discrimination first, because two opposite treatments hang off it (aDPnwO-HtwU):

• People heard it and were unmoved. A record problem, with a cheap next release inside it.
• Almost nobody heard it. A marketing and servicing problem. Re-cutting the record fixes

nothing; go back to the previous chapter.

⚠️ That distinction is buried a minute into a video whose title promises the fix without the test — the most important sentence in the source and the one nobody reaches.

In the first case the record is inventory, not a verdict. The money is spent; the recordings still exist. Keep the song that did work at the top, put one genuinely new song immediately beneath it, remix the middle of the record beneath that, ship it as an EP under a new name. The song that worked is the anchor drawing attention to everything shelved beside it. The new song is not optional — without it the release is entirely recycled and the anchor's pull is spent on nothing the audience has not heard. EP, not album: you are using only the pieces that worked, and the point is holding attention while the real next project is written.

Three cautions. The source gives no threshold at all for which middle tracks "did all right" — no stream count, no ratio — so the step deciding the whole tracklist is left to taste. His timings for reading the data and turning it around are underived, and he says himself it can be faster; keep the shape, not the clock. And his no-producer shortcut — buy a matching-tempo beat, drop the session over it, done — creates a new release with a new clearance chain, which Part I covers and he never mentions.

The same repackage extends the life of a project that did well while the next one is written — the more broadly useful version, and it gets ten seconds of the video.

What to do this week

1. Run the four bases on your current release, in order, and stop at the first that fails. Fix

that one.

1. Pick one record and define its test before running it — criticism or anticipation, which

platform, how long, and what result sends it back to the studio.

1. Look at your profile the way an ad click would. Arriving from one post, is there enough on

the shelf to justify a second visit? If not, that is where the next nine dollars go.

1. If you are sitting on a flop, decide which flop it was — heard and unmoved, or never heard.

Write down the evidence for your answer.