In August and September 2026 we released a nine-track album called Your Kingdom First. It is Christian trap-metal, and it was produced using AI tools, which we have never hidden.
This page is every number it produced. What it cost, what it earned, what worked, what did not, and the three things we got wrong and had to correct in public.
We are publishing it because almost nobody does. Independent artists are surrounded by advice and almost no audited figures — and the few numbers that circulate come from people with something to sell. We have nothing to sell you. These are the actual readings from Spotify for Artists, TikTok Ads Manager, YouTube Studio, Google Analytics 4 and Google Search Console.
Every figure here is read off a dashboard, not estimated. Where a number is uncertain we say so. Where we were wrong earlier, we say that too.
What it cost
| Item | Cost | Type |
|---|---|---|
| DistroKid distribution | R826.99 | annual |
| TikTok Ads | R538.00 | one-off |
| YouTube promotion | R500.00 | one-off |
| Total | R1,864.99 |
Advertising is R1,038 of that — 56% of everything spent. Production cost nothing beyond time, because of how it was made.
DistroKid amortised across three releases this year comes to R275.66 per release, or R75.18 per recording across eleven tracks.
What the advertising bought
TikTok Ads: R538 for one click
| Metric | Value |
|---|---|
| Spend | R538.00 |
| Impressions | 6,136 |
| Clicks | 1 |
| CTR | 0.016% |
| CPM | R87.68 |
| Conversions | 0 |
A typical TikTok ad CTR is roughly 0.5–1.5%. 0.016% is about 31 times below the low end of normal.
We originally wrote this up as “R538 per click” and that framing was misleading, so here is the correction: the CPM of R87.68 is normal for South Africa. The campaign was almost certainly optimised for impressions rather than clicks, which means it was never buying clicks at all. A properly configured traffic campaign has still never been tested here. The money was not stolen — it bought exactly what it was told to buy, which was the wrong thing.
YouTube promotion: R500 for 1,005 views
R500 promoted one long-form visualiser and returned 1,005 views, at R0.50 per view.
Over the same period, fourteen organic Shorts produced 1,177 views for nothing.
| Views | Cost | Cost per view | |
|---|---|---|---|
| 1 promoted long-form video | 1,005 | R500 | R0.50 |
| 14 organic Shorts | 1,177 | R0 | R0.00 |
Free out-performed R1,038 of paid on total reach. That is the single clearest finding in this entire exercise.
What the organic reach produced
Roughly 18,900 video views across TikTok, Instagram Reels and YouTube Shorts during the launch period. Here is where those people went.
| Stage | Figure | Share of views |
|---|---|---|
| Video views | ~18,900 | — |
| Website visitors (28 days) | 7 | 0.037% |
| Sessions on the listen page (28 days) | 4 | 0.021% |
| Spotify monthly listeners | 10 | 0.053% |
| Spotify streams | 71 | 0.375% |
| New Spotify followers in launch week | 0 | 0% |
The website took 7 unique visitors and 24 visits in 28 days. Google Analytics 4 reported the traffic channel split as Direct 62.5%, AI Assistant 12.5%, Organic Search 12.5%, Unassigned 12.5% — and Organic Social 0%.
We need to be fair about that last number. TikTok and Instagram in-app browsers routinely misattribute to Direct or Unassigned, so some social traffic is hiding in that 62.5%. But the ceiling is the whole site. Even if every Direct and Unassigned visit came from social, that is 18 visits from 18,900 views — 0.095%.
Cost per outcome
| Promotion only (R1,038) | All spend (R1,864.99) | |
|---|---|---|
| Per stream (71) | R14.62 | R26.27 |
| Per monthly listener (10) | R103.80 | R186.50 |
| Per save (6) | R173.00 | R310.83 |
| Per follower (4) | R259.50 | R466.25 |
At a typical independent payout of R0.05–R0.09 per stream, recouping R1,864.99 requires 20,700–37,300 streams. The catalogue has 71.
We are not complaining about that. It is simply what the arithmetic says, and most artists never run it.
The thing that surprised us most
The music retains the people who hear it. 10 monthly listeners produced 71 streams — about 7 plays each. Nobody who arrived listened once and left.
And in the same 28 days that Spotify gained zero followers, YouTube gained seven subscribers.
People will commit. They will not travel to do it. They subscribe on the platform they are already standing on, and almost none of them will move to another app to do the same thing. Every piece of advice we had read assumed the opposite.
Three things we got wrong
Including these is the point of the page. Marketing write-ups almost never show the corrections.
Declaring a winning channel off one day of data
After day one we wrote that TikTok was clearly the channel, based on a single day where it produced every like and save while YouTube took 17 views. Two days later YouTube did 118 views against TikTok’s 43. The call was made on one day and reversed within three.
Comparing two different platforms and calling it a format win
We reported that switching to vertical video lifted reach — by comparing 253 TikTok views against a YouTube Shorts average of 84. Different platforms, meaningless comparison. When measured properly on one platform, the letterboxed post beat the vertical one on every measure: more views, double the like rate, four times the comments, and the only saves the account had.
Assuming one video length would work everywhere
Cutting a teaser from 27 seconds to 13 was meant to fix retention. On TikTok the shorter cut won decisively — 267 views at a 7.1% like rate against 143 views at 5.6%. On YouTube the same idea failed: the 13-second cut took 25 views in a week, while the 20–27 second teasers averaged 52. The platforms want different things and we had assumed they wanted the same thing.
One thing we predicted correctly, in writing, before the data existed
TikTok reach collapsed across three posts in September: 397 → 253 → 43 views, an 89% fall. The theory was repetition throttling — too many near-identical clips from one song, posted too close together.
On 21 September we wrote the test down in advance, with the threshold set beforehand so it could not be moved afterwards:
Reach recovers (200+ views) → repetition throttling was the cause. Reach stays low (under 100) → repetition was not the cause.
Written 21 September 2026
With spacing introduced, reach went 43 → 143 → 267, and the like rate returned to 7.1% against an all-time best of 7.3%.
Posting less often produced six times the reach. We would not have believed that without having written the threshold down first.
What we would tell another independent artist
- Do not buy ads before you can measure a conversion. R538 bought one click because the campaign optimised for the wrong thing, and no pixel was installed to tell us.
- Free out-performed paid on reach. Fourteen Shorts beat R1,038.
- Post less often than you think. Spacing recovered six times the reach on TikTok.
- Do not send people to your website on the way to a streaming platform. Ours took four sessions in 28 days and lost half of them.
- Expect almost nobody to change apps. 18,900 views produced 10 monthly listeners.
- Write your threshold down before you run the test, or you will move it afterwards. We did that once by accident and it is the only finding here we fully trust.
On the AI question, since it is the first thing people ask
The music is produced with AI tools. We say so on the record, in the metadata, and in a separate article about what that actually involves.
We mention it here only because it is relevant to reading the numbers: none of these results are explained by it. Nobody declined to follow because of how the music was made — the people who heard it played it about seven times each. The failure in this funnel is distribution and platform-switching, not production.
Methodology
- Sources: Spotify for Artists, TikTok Ads Manager, YouTube Studio, Google Analytics 4 (via Site Kit), Google Search Console and Rank Math Analytics.
- Period: album released 16 September 2026; most figures are 28-day rolling windows read between 17 September and 1 October 2026.
- Currency: South African Rand. We have not converted, because the conversion rate would date the figures.
- Spotify caveat: the platform only holds history from 21 August 2026, when the first single landed. TikTok’s advertising window is 365 days. These are not like-for-like periods and no comparison between them should be read as one.
- The 2 “super listeners” recorded on Spotify pre-date the launch and are almost certainly people close to the project. They are not converts and we have not counted them as such.
- Nothing here is modelled, projected or annualised. Where we do not know, we say we do not know.
If you want to cite any of this, please do — a link back is appreciated but not required. If you think a number is wrong, tell us and we will check it and correct the page.
Frequently asked questions
How much does it cost to release an album independently?
Ours cost R1,864.99 all in: R826.99 for DistroKid distribution across the year and R1,038 in advertising. Production cost nothing beyond time because the music was made with AI tools.
How many streams do you need to break even?
At a typical independent payout of R0.05 to R0.09 per stream, recouping R1,864.99 needs roughly 20,700 to 37,300 streams. We have 71.
Do TikTok ads work for musicians?
Ours did not. R538 produced 6,136 impressions, one click and no conversions — a 0.016% CTR, about 31 times below normal. The campaign was almost certainly optimised for impressions rather than clicks, so a properly configured traffic campaign remains untested.
How many video views does it take to get a Spotify listener?
For us, roughly 1,900. About 18,900 video views produced 10 monthly listeners, which is 0.053%. Those 10 played about seven times each, so the retention was fine — the loss happens at the platform switch.
Does posting more often help on TikTok?
It did the opposite. Posting near-daily coincided with reach falling 397 to 43 views. Introducing gaps of several days took it back up to 267 — roughly six times the low point.
Should I send fans to my website or straight to Spotify?
Straight to the platform, unless you are running paid traffic and need the page for measurement. Our listen page took four sessions in 28 days and lost half of them.
Does AI-produced music get rejected by listeners?
We saw no evidence of it. Our 10 monthly listeners averaged about seven streams each. The problem in our funnel was distribution and getting people to change apps, not the production method.