Recent updates
New Feature: SubmitHub Packages
This week we're rolling out our latest feature, SubmitHub Packages. The idea here is to take something really complex and bottle it up for users who want us to take care of everything.
The normal submission flow is still available and easy to access, and I anticipate most people will still opt for that level of control.
The article below is quite meaty, but I want to lay it all out for you.
The problem I'm trying to solve
Setting up a campaign on SubmitHub involves a lot of work:
- 3 pages of questions to upload a song
- 3 more pages to start your submission, followed by a huuuuge list of curators to choose from
- 3 pages of questions to kick off an ad
For many users, especially first-timers, this is a lot to navigate, and we often see the same avoidable mistakes.
I've been stewing on this for about a year now, and kicked off work on the project about a month ago. I'm a "launch and iterate" kind of product developer, so this is by no means finished. Feel free to ask questions and/or share your thoughts in the comments section below.
The solution
I've found two promotion strategies that work consistently:
- Strategy #1: Trigger the Spotify algorithm
- A combination of playlists and Meta Ads is the way to go
- Playlists are good for quick streams, but the audience isn't as connected
- The solution is to run Meta Ads at the same time - it's more expensive per stream, but you're much more likely to get positive engagement signals (eg, a "save" or a new follower)
- Strategy #2: Branding via blog posts
- Teach ChatGPT, Google, etc., about your music (these big companies "scrape" blog content to teach their algorithms)
- Blog posts don't disappear either
- They're good for sharing on your social media
SubmitHub Packages makes it easy to use both of these strategies.
Launching a Package
I've tried to make this as simple as possible.
- When you land on the promotion page, it prompts you to choose between "easy mode" and "advanced mode"
- You're then presented with a list of packages - these cover the strategies I just laid out
- The final step is to choose your song and submit - that's it!
- What's really cool is that if your song is already on Spotify, all we need is your song link. My code takes care of filling everything else out for you (I'm planning to extend this "simply paste a link" approach to normal song uploads soon)
Once you've submitted your package request, our team will get everything set up for you.
Note: I realize that a lot of music promotion today centers around Spotify, but I think blog posts are still valuable (I'm a blogger, myself). To that end, there's a separate "Branding Package" (RGB blogs) as well as an "Extras" option within each Spotify package to select blogs and/or radio for an extra boost.
How the campaign side works
- Confirmation: Your package is confirmed, and you receive an email with a breakdown of your budget and what to expect.
- Learning Phase (Day 1): Within an hour, we send your track to a selection of curators using about half your budget. This is the learning phase - we're seeing how curators respond to your music. You'll get an email explaining the process and setting expectations. Responses typically take up to 72 hours.
- Round 1 Results (Day 3-4): Once all responses are in, we analyze the results:
- Strong response (50%+ approval): We move to round two with higher-impact curators who have larger audiences.
- Solid response (15-50%): We use what we learned about which curators responded well and send a second round.
- Low response (under 15%): We pause and refine. Your track gets queued for a quick genre check from our community of artists, who vote on which genres best describe your sound. Once that's done (usually a couple of days), we retarget and launch round two with adjusted genres.
- Round 2 Results (Day 6-8): You receive cumulative results across both rounds. If you have credits remaining, we continue sending in the background.
- Silent Cycling: After round two, we keep working through your remaining credits in smaller batches. No more email updates at this stage - you can check your package page anytime for the latest.
- Low Credits Warning: When you're down to your last few credits, we send a heads up with a per-type breakdown so you can decide whether to top up.
- Completion: Once all curator credits are used, you receive a final summary with your overall approval rate and results. If you purchased Meta ads, those run on their own separate cycle.
How curators are chosen
I know there's some concern from curators that automated selection will bias some over others. I don't take this concern lightly, and will be monitoring it carefully.
This section of the blog post is primarily aimed at curators who are looking to maximize their exposure.
When I match tracks to curators for packages, every curator gets a score based on several factors. Here's what matters:
- Genre match: The algorithm starts by checking how well each curator's genre preferences overlap with the track.
- Curators with fewer, more focused genre selections actually get a boost - if you list 8 genres and 2 overlap, that's a stronger signal than someone who lists 50 genres with 2 overlapping.
- If you've boosted specific genres, those carry even more weight.
- We look at how you've responded to tracks in the same genre historically. A consistent pattern of engaging with a genre - whether approving or giving thoughtful feedback - tells us you're a good fit.
- Playlist quality (for playlisters):
- We look at how niche your playlists are and whether the genres of songs in your playlists overlap with the track's genre.
- Curators with focused, genre-relevant playlists score higher than those with broad, catch-all playlists.
- Really Good Bloggers (for blogs):
- When someone chooses the "branding" package, we're looking specifically for written articles from our RGB curators
- Response time: Curators who take a very long time to respond (10+ days for playlisters, 21+ days for blogs) get filtered out entirely.
- Quality and reach: Your overall quality score and reach factor in. Curators with notably low quality or reach scores get a small penalty.
- Language and mood preferences:
- If you accept specific lyric languages or have mood preferences set, tracks that match those preferences will be routed your way.
- Mismatches reduce your score.
- A quality score below 6, and you're unlikely to receive any automated submissions
- History: If you've previously approved tracks from the same artist, you get a boost - the algorithm recognizes you're a fan.
- New recruits: If you're new to the platform (joined in the last 30 days), you get a baseline score that puts you in the mix even though you don't have history yet.
As a curator, here's what you can control:
- Keep your genre list tight and accurate - boost the ones you really love
- Set your mood and language preferences
- Respond promptly
- Build focused, niche playlists (for playlisters) and join the RGB program (for bloggers)
- Stay active and consistent - those in weekend mode or on a break won't receive automatic submissions
Submissions sent using Packages show up with the following note:
Meta Ads
We've been running Meta Ads for over a year now, and our amazing Tilly has set up more than 10,000 ads! Integrating this portion of the "package" was fairly straightforward:
- I automatically create a Link (landing page) for your song
- I then set up an ad for you
Sounds simple enough, right? But behind the scenes, I'm doing some pretty cool stuff.
The most interesting one is probably how I'm using lyrics. I've built a new AI model that:
- Extracts the vocal stems
- Analyzes them with whisper (voice to text) and extracts the lyrics (~85% accuracy)
- Takes those lyrics and puts them through Qwen3.5 (an open source LLM) to summarize what the song is about
I've also built an AI model that extracts moods and "thematic tags" from a song.
I can then use these extra bits of information to generate the ad copy / captions / select the appropriate videos for creatives.
This means we can kick off a Meta campaign with a few hundred lines of code, and all that's needed is a quick review to make sure everything's okay.
What's left for me to code?
As mentioned at the beginning, this is by no means a finished product. There are still a few things I'm buttoning up:
- Easy high-level reporting of Package results
- Topping up budgets
- Handling pre-release songs (ads must wait until release day)
I'm also going to be monitoring results very closely to ensure that a) we're effectively helping people promote their songs, and b) curators are well represented.
My hope is that this feature makes it easier for artists to promote their music AND brings an increase in submissions for curators. Fingers crossed!