Hello, on behalf of myself (Jason) and Dylan, I’d like to welcome you to our 2020 update on the Influence Score.
Before we get into the nitty gritty, perhaps it’s best to clarify the intention of the “Influence Score”.
The Influence Score exists to help artists answer the following question:
“If I get shared by Outlet A, how much might it move the needle for my song vs. Outlet B or Outlet C?”
The Influence Score is a relative measure that attempts to compare “apples to oranges” (YouTube vs. Spotify vs. blogs, etc.). It isn’t perfect, but it’s a heck-of-a-lot better than looking at fan counts.
Each outlet’s score is likely to change over time, and a high Influence Score doesn’t always mean we recommend sending to them. The genre match, quality score, and overall approval rate are also important metrics for artists to keep an eye on.
When SubmitHub launched in late-2015, each curator profile was accompanied by a snapshot of the total number of fans they had across all their various networks.
As the selection of curators on SubmitHub grew, it became apparent that “total fans” was not a good reflection of how engaged an audience actually was. This discrepancy became even more pronounced when we opened up to Spotify playlisters, many of whom had tens of thousands of followers but failed to bring in a corresponding number of listeners.
SubmitHub circa 2017
Some time in the middle of 2018, I sat down to tackle the idea of a “universal” Influence Score that factored in “engagement” and could be applied across various outlets. Fan count still provided a foundation, but points were added or subtracted based on whether that curator managed to successfully move the needle for the songs that they shared.
Metrics analyzed included number of Spotify listeners, YouTube views, Hypem-listing, Instagram story views, and SEO rankings.
Over the years, new data became available and the methodology for generating the score was regularly tweaked.
Eventually that constant tinkering resulted in a “bloated” formula that meant pulling a lever in one location would result in wild swings elsewhere. And while these “swings” weren’t obvious initially, it wouldn’t take long before emails and chats flooded in from curators who were wondering why their Influence Score had dropped a point. Funnily enough, no one ever complained when their score went up! ;-)
Last week, I decided it was time to revisit the formula with the following key objectives:
The starting point for this “rewrite” was to dive into the metrics available and determine how to weight them.
The metric here is pretty straight-forward: # of views per video (shared via SubmitHub).
YouTube channels tend to be able to pull the biggest numbers out of anyone on SubmitHub. The average video upload gets 5,000 views, and the top 10% of YouTubers can easily generate 120,000+ views per video. Getting that many eyeballs (or earballs) on a song is definitely going to have a spillover effect into various other streaming platforms.
It’s worth noting that YouTube channels tend to have a significantly lower approval rate than many other outlet types, largely due to the fact that they tend to not upload more than a handful of videos each week. This doesn’t factor into the Influence Score as the approval %age is already displayed separately next to each outlet.
There are a number of metrics available here, but the most-important one is without-a-doubt the number of listeners for each song that gets shared. Crunching the numbers, the average playlister pulls in ~350 listeners, while the top 10% can bring 5,000+ to each song.
Spotify is a priority for many artists today, and after a few discussions, it felt safe to conclude that for most, 5,000 listeners on Spotify is generally considered “more valuable” than a similar number of views on YouTube.
We have a few methods for determining the # of listeners. I’ll keep those private for now because they’re a bit of a “competitive advantage” that we have. Ultimately the numbers are based on stats provided by artists who have been shared via SubmitHub, and are weighted according to which playlists are shared to most-often.
Another cool metric we have access to is the % of premium listeners behind each song. This is something that does get factored into the overall score. Someone who tries to boost their listeners via dubious sources will end up hurting themselves more than it’s worth.
Our Spotify numbers are not 100% accurate, but they’re a good benchmark and I believe they’re the best thing out there. The only other platform I can think of that publishes listener counts is Chartmetric, but my understanding is that they stopped updating their number in ~March when Spotify began listener counts on artist profiles.
Once the darling of the music industry, most blogs do not have the audience they used to (including my blog Indie Shuffle, which has dropped from a peak of~6 million streams per month in 2013 to ~1 million per month in 2020).
That said, there are some really important upsides of getting blog coverage:
Blogs use various methods of analytic tracking, so it’s hard to narrow in on one. So… I’ve actually chosen a few.
Pretty much everyone has an account on these social networks, but the engagement levels are horrendous.
For example, Indie Shuffle has 113,000 real organic fans following our Facebook page, but when we share something we’re lucky if we get 30 likes. I crunched the numbers for everyone on SubmitHub and it’s embarrassing (Facebook gives us a nice “talking about” metric via their API).
It’s clear that Facebook doesn’t think new music deserves much time in the average user’s feed, and as such Facebook should no longer have a material impact on the Influence Score.
Same goes for Twitter. It’s short-lived and the average account is lucky if they get 2 or 3 engagements on a Tweet.
On the subject of VK, it seems like they’re way-more geared toward music sharing. And so for those Russian curators on SubmitHub who have a good presence on there (VK shows views per post — yay!), there’s a boost available toward the Influence Score.
SoundCloud seemed to take a pretty steep turn downhill when they rolled out the “repost” feature. Sure, there were plenty of other factors going against them, but that one messed things up for me.
What about for artists, though? What can they gain via a share to SoundCloud? Well, the platform can be leveraged two ways:
We have some curators on the site who do both of these, but most simply repost. The former (uploading) tends to have a stronger impact, so we kept that in mind when factoring available SoundCloud metrics into the influence score.
That said, the numbers we’re seeing curators generate on SoundCloud are significantly less than those being generated on YouTube and Spotify. SoundCloud still matters, but not as much as it used to. And our influence score attempts to capture that shift.
We’ve got a few people who have active Mixcloud “shows”, and I actually think that for those select accounts it’s a pretty important metric. Mixcloud’s “niche” = people passionate about music. Chances are that if they’re listening to your show, they’re there to find some good new music.
If you’re an influencer and you’re reading this, I think it’s important for me to emphasize that SubmitHub’s Influence Score is not a reflection of how influential you are overall. It’s specific to the music industry.
Earlier this year we started signing up Instagram Influencers (and TikTokers) to use SubmitHub. Many of them have been great additions to our community. That said, in many cases we’re not sure if it’s actually moving the “music” needle in any considerable direction for the artists who are being shared.
Getting 1,000 views on an Instagram story vs. 1,000 listeners from a Spotify playlist is *very* different. Heck, even 10,000 views on a story doesn’t hold a candle to 1,000 Spotify listeners. Not even close.
Ultimately, Instagram isn’t where people go to listen to music. And while Facebook is putting a lot of effort into growing that part of the platform (they’re supposedly paying royalties, and almost all distributors now allow for easy upload of music to Instagram), it’s clear that a top-tier music-focused channel will have significantly more impact than a top-tier fashion influencer.
TikTok’s an interesting platform with loads of potential. However, the reality is that “indie” songs aren’t the focus of TikTok. While new acts have certainly leveraged it to “break out”, it tends to be a regurgitation of the same “viral” songs and isn’t well-suited to indie music discovery. As such, engagement for TikTok is accounted for but does not have a heavy weighting in the overall score.
This is one of the toughest to measure.
It’s near impossible to reliably calculate reach, and the whole medium of radio is consistently losing its following in developed countries. On top of that, verifying when or how many times a song was played is difficult unless you’re actively tuning into the show.
The only decision we’ve decided to make here is that if a radio station is terrestrial (that is, they have an FM or AM frequency), they’ve probably got a larger audience than the average online show.
After much tinkering, I decided that the best approach was to determine a “percent rank” for each outlet within the respective metric being analyzed, and then multiply that by a fixed amount of points.
For example, if a YouTube channel picks up an average of 7,500 views per video they fall at the 63%ile relative to all other YouTube channels on SubmitHub. And if the number of available points for the YouTube metric is 50, they’d gain 50 * 63% = 31.5 points toward their overall Influence Score.
The available number of points for each metric varies, but in general it’s assumed that the highest-impact methods of sharing songs in 2020 are Spotify, YouTube and blogging.
Channels are primarily compared among themselves (YouTube vs. YouTube, Spotify vs. Spotify), and consideration is factored for the relative impact that each sharing medium might have on a song’s growth (per the above metrics breakdown under “What are we measuring?”).
Not everything works out nicely when using a percent rank methodology. Alexa Rank comes to mind as an example, as does being listed on Hypem (it’s either true or false). So in some cases, a slightly different methodology must be used.
Each outlet can gain points in multiple measures, though a weighting is factored in based on where they typically share (eg, a big YouTube channel that has zero video uploads via SubmitHub won’t get as many YouTube points).
I’d get into the nitty gritty, but I worry that sets the formula up for being gamed :(
In the end, one of the most-important objectives of the Influence Score is to properly set artist expectations.
If we say someone has a Influence Score of 1/10 and a song they share gets virtually no new listeners, we don’t want that to come as a surprise.
On the flip side, if someone has an Influence Score of 10/10, we expect their share to have a noticeable impact on that artist’s song.
A low Influence Score should not be taken as a judgement of that curator’s content or passion for music. We have many excellent curators who share to their chosen medium passionately, but don’t even bother with other social media services / platforms. They might have a low Influence Score, but they’re still totally worth sending your music to (and that will be reflected in their sharing history + Quality and Genre Match scores).
We spent a lot of time working to fine-tune the new Influence Score. We’re fairly confident about where we’ve landed, but there are more than 1,400 active outlets, and corner cases exist. If you spot something that looks horribly off, feel free to ask us to take a deeper dive.
Jason & Dylan