Answers/Problems & answers/Original research
Problems & answers
Original research: Every Question, Answered
First-party studies built from my own data — numbers nobody else can publish because nobody else has them. Every question below is answered here in full — the link takes you to the page it came from if you want the working behind it. 21 questions.
I Tested 40+ YouTube Titles Against vidIQ's Scoring Model — One Clause Moved the Score 24 Points
Across structured title-testing sessions against vidIQ's title-scoring model, the largest single lever was not the headline formula but the specificity of the middle clause. Holding the opening and closing beats identical and changing only the middle consequence moved the score from 73 to 97 — a 24-point swing from one clause. Two further findings: the best architecture differs by subject matter rather than being universal, and 97 was a hard ceiling that no amount of word-level revision broke.
Is this measuring actual video performance or a scoring model?
A scoring model. These are vidIQ title scores, not view counts, and a high score is a prediction rather than an outcome. The findings are about what the model rewards, which is useful for drafting but is not the same as proven view lift.
How many titles were tested?
Over forty across several structured sessions, including deliberate controls where only one clause was varied while everything else was held constant.
What was the single biggest finding?
That middle-clause specificity outweighed the choice of formula. A named, concrete, personal consequence scored 24 points higher than a systemic or abstract version of the same claim in the same title structure.
Why did 97 turn out to be a ceiling?
Six word-level variants of the two highest-scoring titles all tied or scored lower. The model's own diagnostic indicated the format category itself was capped, and that reaching higher would require a structurally different hook rather than further editing.
Does a higher vidIQ title score actually mean more views?
Not by itself, and I want to be careful here. vidIQ's score is a prediction generated from patterns in titles that have performed, not a measurement of how a specific video did. A title that scores 97 has the shape of a title that tends to work; whether the video behind it works depends on the thumbnail, the first thirty seconds, the topic and the audience. What the score is genuinely useful for is comparing two drafts of the same idea, which is exactly how I used it.
Can other creators use these architectures?
The specificity principle should generalise, since it is about concreteness rather than any particular phrasing. The exact templates were tuned against one channel's subject matter and one scoring model, so treat them as a starting hypothesis to test rather than a formula to copy.
Full explanation: I Tested 40+ YouTube Titles Against vidIQ's Scoring Model — One Clause Moved the Score 24 Points
What Music Platforms Actually Pay Per Play — Seven Years of My Own Royalty Data
Across seven years and 677,639 verified plays of my own catalog, the spread between platforms is far wider than the per-stream debate suggests. Apple Music paid $0.0066 per play and YouTube Premium $0.0092, while YouTube Shorts paid $0.0000032 — about 2,900 times less. The headline number: 18.7% of my verified plays came from social and short-form video, and they produced 0.5% of my money. A further 41.3 million Facebook plays sit in the export and are excluded here — they are fingerprint matches to other people's videos, not plays of my songs, and the section below explains how I found that out.
How much does Spotify pay per stream?
Across 186,519 of my streams from 2019 to 2026, Spotify paid an average of $0.0033 per play, or roughly 301 plays per dollar. In the US specifically the rate was $0.0039, about 259 plays per dollar. My yearly Spotify rate has been remarkably stable at around $0.0038 since 2020, apart from a dip in 2023.
Which platform pays the most per stream?
In my data, YouTube Premium (listed as YouTube Red) at $0.0092 per play and Apple Music at $0.0066 per play were the highest-paying at meaningful volume. Apple Music paid me $1,110 from 175,162 plays; Spotify paid $613 from 186,519 plays — more money from fewer plays.
How much does Facebook pay musicians per play?
On my own Facebook posts — the Social Media Pack line — 70,270 plays paid $1.79, about $0.000025 per play. The separate "Facebook" store in my export shows 41.3 million plays and $36.14, but those are Rights Manager fingerprint matches to other people's videos, mostly in Mexico, Argentina and Spain where I have no audience, and I exclude them. The mechanism is explained on this page.
Do YouTube Shorts pay artists anything?
Almost nothing per view. 51,461 Shorts plays of my catalog produced $0.17 total, roughly 308,000 plays per dollar. Shorts are a discovery surface, not an income surface.
Is this the same as total music income?
No, and this is the most important caveat. These are recording-side royalties reported by my distributor only. Performance royalties from a PRO, mechanical royalties from the MLC, and SoundExchange money are all separate pots that do not appear anywhere in this data.
Are these gross platform payouts?
No. These are net amounts reported to me as the artist. Rates are normalised for team percentage, so split tracks do not understate the platform rate, but they still reflect what reached my account rather than what the platform paid at the top.
Full explanation: What Music Platforms Actually Pay Per Play — Seven Years of My Own Royalty Data
I Paid Songtrust $100 to Collect My Publishing — Here Is Every Dollar That Came Back
Over payment periods from late 2023 through early 2026, covering usage of my 73 registered songs back to June 2019, Songtrust collected $261.39 in publisher-share royalties on my catalog and paid me $222.76 after commission. 96% of it was mechanical royalties — $189 of the mechanical-streaming money came through the MLC alone — and only $10.03 was performance. Against the $100 registration fee, my publishing admin has returned about $123 net so far. The same catalog earned $3,117 on the recording side in the same era: publishing was one-fourteenth of distribution money at my scale.
Is Songtrust worth it for a small artist?
On my numbers it returned the $100 fee plus about $123 over roughly two and a half years of payment periods, so it cleared its cost — barely. The real answer is that almost all of what it collected was mechanical royalties, and US streaming mechanicals are collectable free by registering directly with the MLC. What Songtrust added beyond a free MLC registration was mostly foreign collection and publisher-share performance, which together were a minority of my total.
How much of the money was performance royalties?
$10.03 gross out of $261.39 — under 4%. At my streaming volume the publisher share of performance royalties is almost nothing. The fear of "losing half your performance royalties" without a publisher was, in my data, a fear about single-digit dollars.
How long do publishing royalties take to arrive?
The median gap between the month a song was streamed and the quarter Songtrust paid it was about 290 days. A quarter of my royalty lines took under ~3.5 months; another quarter took over 2.7 years, much of it marked as reprocessing or historical adjustments. Publishing money is slow money.
What did each platform pay per thousand streams on the publishing side?
For US mechanical streaming in my data, per thousand units collected — Amazon $1.38, SoundCloud $0.96, Apple Music $0.85, YouTube $0.62, Spotify $0.44. These are publisher-share mechanicals only, a small fraction of the recording-side rates on the same platforms.
Is this all the money the songs earned?
No. This export is composition-side, publisher-share money only. Recording-side royalties from my distributor (about $3,117 in the same era), the writer share of performance royalties paid directly by a PRO, and SoundExchange digital-radio money are all separate pots that never appear on a Songtrust statement.
Did the Songtrust commission change?
Yes, and my statements show it. Rows through 2024 pay out at 85%; from 2025 Q1 the mechanical rows drop to 80%, matching Songtrust's published change to 15% performance / 20% non-performance commission effective January 1, 2025.
Full explanation: I Paid Songtrust $100 to Collect My Publishing — Here Is Every Dollar That Came Back
Get the paperwork done in one afternoon
The Zero to Beat Society walks through registration, splits and release paperwork step by step — with the templates and checklists already filled in.
See the tiers