We will also be presenting real world case studies of Musicmetric Pro demonstrating how it can be useful to you and all levels of the industry. We will also present where and how companies have implemented our API to great effect so don’t miss our presentation on Visionary Monday at 4.00pm in the Innovation Factory.

At our stand we will be having an informal drop in “beers and demo” session on Saturday 28th and Sunday 29th from 4pm till 6pm.

We will also have a presence at the Midem hack day so head to the presentations or register to meet the hackers if you want to discuss our API and data feeds.

Or catch us at the Music Tech Meetup @midem for networking and drinks on Sunday evening from 6:30pm to 9:30pm. More info here.

Attending from Musicmetric are:

• Gregory Mead – CEO
• Marie-Alicia Chang – Head of Business development
• Jeremy Silver – Chairman
• Jameel Syed – CTO
• Juliana Obeng – Communications executive
• Ben Fields – API / Data lead (Midem Hacker)

You can find us during the conference at stand MEDIUM 12 but if you want to organise a meeting then drop us an email on midem2012@musicmetric.com.

See you in Cannes!

Register here!

With more and more fans discovering and interacting with bands online, looking at online buzz as a major part of Brand New for 2012 is a fresh and forward looking take. The metrics we chart add a completely new and dimension, arguably giving more realistic insight as to how the artist will fare in the real world – beyond the list and label marketing.

Artists like Lana Del Rey [click for stats] making such a big impact through social media channels (such as YouTube) in the last year only serves to highlight how important these metrics are now for all artists. For MTV to allow an unsigned artist onto their Brand New list (alongside Del Rey) based on online buzz is a great step.

During the course of the unsigned competition The 500 long-list artists together generated the following across social media –

New Fans: 107,181
New Plays: 3,552,848

From the start of the competition and throughout, each artist was tracked by us providing daily statistics on which artists have risen the most in terms of their buzz; an aggregate of votes, plays, new fans and comments.

So congratulations to Context! you can check him out on twitter here

N.B. The scale refers to the number of open download sessions monitored rather than the total number of downloads.

I’m co-chairing this workshop (along with the other co-chairs — Amélie Anglade (Soundcloud), Òscar Celma (Gracenote), Paul Lamere (The Echo Nest), and Brian McFee (Computer Audition Lab, University of California San Diego)) which will cover a diverse array of approaches and angles for music recommendation and discovery.

The workshop is part of RecSys 2011, which runs the full week, though I sadly can’t stay for most of the conference aside from our workshop (see below). It should prove to be an interesting day of research. Are you planning on attending? Let us know.

Early Monday, I’ll be dashing off from Chicago to Miami to attend the International Society for
Music Information Retrieval Conference (ISMIR). This conference is “world’s leading forum for research on the modelling, creation, searching, processing and use of musical data.” There I’ll be demonstrating some new (not-yet-released) API features from Musicmetric (on Friday the 28th). Here’s abstract for our demo:

In this demonstration, we will show how both social and peer-to-peer network analytics are aligned by track across large datasets. We present this data as track specific timeseries from various domains and communities. These streams of timeseries data describe behaviour of fans for a track (or the artist that produced the track) in various ways include: friend transitions (follow/unfollow), demographics, and (peer-ro-peer) downloader location (by region) and activity level. This data is aligned via common identifiers to the tracks making up the Million Song Dataset (MSD) [1]. As this data is also easily available via the Musicmetric’s Web API this trend data provides an alternative source of data above and beyond the content-based features provided for each track making up the core of the MSD.

We will demonstrate prototypical usage of this contextual data live against the tracks of the MSD. These calls will be exploratory and modifiable during the demonstration, allowing the audience to interact with our data and understand how it might be used in their research.

I’ll be around the ISMIR conference all week, so if you have any questions about our service, especially about using our API, come find me in Chicago or Miami and let’s see what we can do!

- Ben

[1] Thierry Bertin-Mahieux, Daniel P.W. Ellis, Brian Whitman and Paul Lamere, “The Million Song Dataset”, ISMIR 2011.

Andrew Morgan (@minkymorgan) has built some helper functions in R for the Musicmetric API. These allow quick and easy access to our time series data for any artist given their ID (or any third party ID including Last.fm, Twitter and Facebook), and for top artists based on our chart api endpoints.

The code uses the Quantmod library, which has a whole bunch of time series analysis and charting features.

Rmusicmetric can be downloaded on Github right here and you can sign up for the Musicmetric API here.

Check out some example output plots:

and

So we hacked together a little chart using our real time play monitoring that ranks the artists by the change in their play counts online over the last hour. This ranks higher the artists that people have started to listen to over the last hour.

Here’s the result for 16:00 to 17:00 GMT today:

1 Maroon 5 – Moves Like Jagger [feat. Christina Aguilera]
2 Britney Spears – I Wanna Go
3 Chris Brown – Beautiful People
4 LMFAO – Party Rock Anthem
5 OneRepublic – Good Life
6 Cobra Starship – You Make Me Feel…
7 Jason Aldean – Dirt Road Anthem
8 Lady Antebellum – Just A Kiss
9 AFI – Beautiful Thieves
10 Gym Class Heroes – Stereo Hearts Feat. Adam Levine
11 Demi Lovato – Skyscraper
12 Jay-Z – Otis
13 Colette Carr – We Do It (Primo)
14 Dierks Bentley – Am I The Only One
15 Miguel – Sure Thing (Main Version)
16 Brad Paisley – Remind Me (Duet With Carrie Underwood)
17 Lady Gaga – You And I
18 Jake Owen – Barefoot Blue Jean Night
19 Coldplay – Every Teardrop Is A Waterfall
20 Avril Lavigne – Smile

And good news everyone – we’ve even made it into a Spotify playlist for your delectation right here! Sure it’s a bit mainstream, but what do you expect – this is the internet. When we’ve got some time to spare we’ll throw in some personalisation.

They have done this by offering an alternative to the larger commercial events, with a combination of home-grown, family-friendly and village fair vibes, and by being virtually or totally sponsorship-free. Added to this, some differentiate themselves by carrying a theme that attempts to involve attendees in dressing up. Standon Calling is one such festival. Its theme this year is ‘Monsters and Gods’, which as well as inspiring fancy dress, will influence the stage and set designs, art installations and interactive theatre at the event.

Forty minutes away from London, Standon Calling started in 2001 as a gathering of friends celebrating a birthday around a swimming pool with a barbeque and some record-decks. It has since developed into a so-called ‘boutique’ festival, set in the grounds of a 16th century manor house, and in the last few years has hosted Brits winners and Mercury music nominees including Florence and the Machine, Friendly Fires and Mumford and Sons.

This year’s ‘next big thing’ on their bill is US electro lo-fi musician Washed Out, who is currently topping the Standon Calling Band Tracker that the Guardian put together using the Musicmetric API.

The buzz about Washed Out began in 2009 when he posted a number of bedroom recordings on his MySpace page. Since then he has released two EPs, and his debut album Within and Without, which came out last month, is a summer record that I’m sure you will be able to enjoy at any time of the year.

Washed Out is the alias of Georgia, USA-based songwriter Ernest Greene, and one that very much sums up the sun-bleached, blissed-out aesthetic of his process-heavy sound that allies him with other bedroom acts associated with what is known as the Chillwave movement. His idiosyncratic brand of Chillwave – a reverb-drenched romantic psychedelia that mixes 80s synth-pop with early-90s dance beats – also sits well alongside the current crop of West Coast new-surf acts Best Coast and Girls, and even the ethereal Warpaint.

As well as monitoring Washed Out’s position on the Standon Calling Band Tracker, you can track his buzz and sentiment on social networks, bit torrents, and the web, and also compare it to other similar artists, using the Musicmetric Fantracker (above).

Oh, and if you are going to Standon Calling this weekend, be sure to bring your swimwear – they still have a swimming pool!

Toby Burton

It shows a very significant spike in activity around the album, but how common is this? Do all artists get a huge increase in Facebook fans for every album release? Does it make a difference how popular or established the artist is? What is classed as a big spike or a small one – how well did my artist do? And why should I care?

That last question is easy to answer, the more fans on social networks you have, the more fruitful engaging with them will be. Tell them about new gigs or releases, offer merchandise to them, etc. It’s obvious that the more fans you have online, the more response you will get from direct online promotion. The other questions are a bit more tricky, and we’ll try to answer them in this post, so read on…

Looking at the data

What I did was to automatically measure the size of the peaks after an album release (like the Lady Gaga one above) for around 1500 US album releases from the top 1000 artists over the last two years. One constraint I decided to use for this investigation was to restrict the data to what we have available via our API, so in principle anyone could run a similar analysis. Of course, these constraints aren’t exactly restrictive as we make available all the data for the top 1000 artists anyway.

The first thing is to ensure that we compare like with like, so I split the top 1000 artists into three tiers, with the combined popularity of each tier being equal (area under the curve) [see graph below]. The results of this are interesting in themselves:

• Tier 1 : 40 artists
• Tier 2 : 183 artists
• Tier 3 : 777 artists

This means that for the top 1000 artists on Facebook, 33% of all activity is down to just 40 artists, and popularity wise you need the top 777 long tail artists to match the top 40.

Now that we have our three sets of artists (you can see them here if you want) we can start looking at their trend data. I grabbed the time series Facebook Fan data for all 1000 artists from our API, some summary data and their album release events from Musicbrainz.

To show the general impact I compared the number of fans gained for each artist in each tier during the 14 days before the release to the number gained over the 14 days after the release (inclusive of the release date). It’s quite a crude measure but it serves our purposes well to give a rough overview of impact, especially with the fairly large sample size being used. We can do some more rigorous time series analysis / projections another day.

If I plot the impact as a percentage change in the number of friends gained before and after for the Tier 1 artists on a histogram, it looks like this:

The graph shows in general, releasing an album gives a positive impact on the number of Facebook fans! (seems obvious), with an average at around a 25% increase. That means that on average, a Tier 1 artist will get 25% more fans over the two weeks after a release compared to the two weeks before. It also shows that some bands actually get fewer fans the two weeks after than the two before. Working out the precise reasons for this is an exercise for the reader, but some reasons may include a larger pre-release ramp up compared to post-release, the album being a re-release or the artist not being particularly active on Facebook.

You might have noticed that red line on the graph called “T1 Random”. I put this in to compare the general trend of the social network to the impact caused by an album release. I generated 300 randomly placed ‘events’ and measured the number of fans before and after the random event. Think of it as the ‘average’ trend for randomly selected two week intervals. As you can see it’s further left than the album release impact line, confirming that album release events do increase fan activity above the average. Interestingly the average (T1 Random) line shows a higher value for positive increases than negative, which illustrates the growth of Facebook in general.

The graphs below show the same for Tier 2 and Tier 3 artists:

It shows that Tier 2 and Tier 3 artists have a similar impact to each other, and they tail off slower that Tier 1 artists, showing they have more room to grow than the more successful artists.

The histograms above are plotted for the range -200% to +200% to show the typical trends. There were however some outliers (very large increases) that we can take a look at now.

To show these I’ve taken each album release for the the artists in each tier and plotted them against the impact they created. The graph below shows this for Tier 1 artists:

This really illustrates the difference between impact caused by an album release event (blue) and the average trend for the network (red). Looking at the same for Tier 2 and Tier 3 show some interesting differences to Tier 1:

You can see that T2 and T3 are much more variable, and in some cases get an impact in excess of 2000% ! This is because those artists didn’t have many fans to start with so they have a lot of room to grow, and small absolute changes in their trend correspond to high percentage changes day on day. Imagine you have only one fan then the next month you get 2000 from your album release – that’s a 200,000% increase. A Tier 1 artist like Lady Gaga isn’t likely to get that percentage increase any time soon because she’s already got the majority of her fans (a 200,000% increase in Lady Gaga monthly fans on Facebook is around the same as the population of the world…).

Just out of interest, I’ll show the graph of the difference between impact and the network average for Tier 1 artist album releases on Facebook (the difference between the red and blue line for Tier 1 above):

You can see around 125 releases showed a small negative difference to the average, where as around 170 give a positive impact, but the magnitude is much higher for the positives.

Conclusion

That’s all great, but after all this you’re probably wondering how good is good and how bad is bad. Well I’ve created this handy look up table for anyone who wants to know if their album release had a low, average or high impact on Facebook fans!

You can use Musicmetric to measure the Impact values for your band.

The table above also shows that for Tier 2 artists, their average impact is lower than both Tier 1 and Tier 3, this could indicate that the majority of Tier 2 artists have reached the peak of their Facebook popularity, and don’t have much more space to grow, or they are not very active on Facebook.

It also shows the high variance that Tier 3 artists have in album release impact because of the factors mentioned previously.

Future Investigation

Clearly there is room to expand on these findings, we could do the same for plays over time, for singles rather than albums or even gigs. There is also room to improve the time series analysis using more advanced time series modelling algorithms. Check out our API documentation and T&C’s if you’re interested in doing something similar.

This is made easier by the fact that these days, in order to raise their profile, engage with their fans and promote their music releases, the majority of up and coming bands have a presence on all the major social networks. And it is often through these networks that I can find out more about a band, and establish how popular they are becoming.

But one band that has chosen to buck this social media marketing trend in the last year, and as a result has been a little trickier to track, is Manchester 4-piece WU LYF (pronounced ‘Woo Life’). Instead of focusing themselves across the major social networks, they have chosen to cultivate mystery about the band and to maintain a level of anonymity by refusing interviews by the press, and by only releasing a limited amount of information, photos of the band and music via their website. And their website, which is chock-full of semi-religious iconography, evocative imagery and enigmatic text, only serves to heighten curiosity about them.

It soon becomes clear from the website that WU LYF is more than just a band. For those in and around it, it is more like a lifestyle choice – as well as being: a club, a gang and an organisation. In fact, The Lucifer Youth Foundation, from which their acronymic name derives (the first two letters stand for ‘World Unite!’), is actually described on their website as a “not-for-profit organisation”. In order to be a part of it, you need to pay a £15 membership fee (+£3.50 p&p) and adopt a set of disciplines and interests including making art, music and film; and be prepared to indulge in petty crime, money-making and “busting heads”. In return, you will receive, among other things: music, discounted entry to gigs, a statement of intent, and “involvement and a democratic input with all LYF activity”. But don’t worry, if this all seems like a little too much commitment, and a bit dangerous, you can just buy their new album for £9.99.

All this esotericism, and the band’s clandestine activity, has as you can imagine driven the music industry crazy with curiosity, and made them even keener to get a piece of the action. But the band continue to reject much of the attention and to perpetuate the secrecy that surrounds them by operating in the shadows of the sub-culture.

To add further frustration to the music industry, WU LYF’s self-released debut album Go Tell Fire To The Mountain totally lives up to expectation – exceeds it, even. Their song lyrics – all available to read on their website – are abundant with biblical imagery of blood, fire, mountains and caves, and are delivered by the rasping impassioned vocal of front-man Ellery Roberts. And their atmospheric music, which often builds to crescendo, combines church-organ keyboard sounds, Xx-style guitar reverb and syncopated drum rhythms, to entrancing and dramatic effect.

WU LYF have apparently declared they have plan three albums before they disband, so they are definitely one to watch in the coming months as they continue to try to avoid the enticement of the traditional music industry and a conventional path to bigger success. Whatever the outcome, they have already proved that there is more to getting noticed than being active on Facebook and Twitter. In fact, they suddenly make bands operating on these, and other social platforms, seem very unhip indeed.

Despite their elusivity, we at Musicmetric will have find plenty of data to track on WU LYF. We will be monitoring websites that mention the band and their releases, and analysing the sentiment of comments made there. We will also be collecting data from the social networks frequented by their fans, as well as, P2P networks used to trade their music.

So why not follow the effect of this ‘mystery’ marketing strategy by adding WU LYF to the artists you follow on the Musicmetric App?

Toby Burton