Ambilyrics is a lyrics based social network. Users are allowed to add artists, revise lyrics, and vote on lyric submissions for accuracy and presentation. It is fully featured, complete with profiles, messaging, notifications, account settings, privacy options, commenting, following, etc.
This is our first major in-house development. We chose this type of project due to the ability for it to be mostly automated. Through the use of cron-jobs and APIs, we don't have to spend much management time. The only thing we actively update are the lyrics (we only add lyrics to the top #50 - social network purposes). We are counting on our users to do the rest. With any luck, people will get interested in helping us out.
On a side-note, we also truly enjoy lyrics. We listen to music every day during our work hours, it seems to improve morale around here. Songs have the ability to connect with people on any number of emotional levels, depending on the individual's current situation. We believe this is why music is so widely listened to and appreciated.
We understand that there are a huge number of lyrics sites out there. We don't see competition as a problem, it means the business is worth trying. The problem is that most lyrics sites are slow and completely ad driven. They constantly barrage the user with nonsense. We will fix that by having a clean layout and minimal advertisement space, of which will only show if the user is not logged in. We will only cater lyrics, no unwanted news, cross-promotions, or other inconveniences.
We have a script that crawls iTunes' API to validate artist submissions and update their individual discographies. If the artist is not found, then it is deleted. Along with this script, we have another that will poll existing artists to update them should they come out with a new album or single.
We also have a script that will search through the top 300 songs on iTunes, and if any song or artist does not exist, it is added. We will also rank each song that is in the top 300 so that we don't have to use their RSS feed on every home page request.
With these scripts, our artists, albums, and songs are constantly being updated and retooled. iTunes frequently updates their song names for accuracy, so we follow suit.
We are allowed to do all of this by their terms because we act as affiliates. We have every artist, album, and song linked to their respective iTunes purchase page. This will help the artists get some money back from us borrowing their words. We make a (very) small commission on any qualifying sales.
We have scripts in place that will send a random portion of a lyric to our Facebook page. This way, even if the song is a duplicate post, it will always be different.
If there is a YouTube video submitted with the lyric then it will be posted as a video link. If no video is present, it will use the album artwork. There is a 25% chance for the lyric to be completely random, and a 75% chance for it to be a top #50 song. Accompanied with that, is a link that points back to the respective page on our website.
We do the same lyric posting on Facebook as we do on Twitter, the lyrics just have to fit inside of the 140 character limit. We then add a hashtag of either #nowPlaying (for random) or #top50 #(whatever rank it is). At the end of the tweet, we use our custom ambi.ws link shortener to direct users back to our website.
Then we work on auto-promotion. One script will follow anyone who writes lyrics via the #lyric or #lyrics hashtag, as well as unfollow those who haven't followed within a few days. It will also unfollow people that have followed us and are a few weeks old. Hopefully, this will weed out the users who only followed us because we followed them, and it will get our foot in the door to users who really enjoy these types of tweets. This only bothers a few people every couple hours, so it's not seen as too aggressive to violate their terms.
Any users that are currently following us get messaged, telling them to like our Facebook page. This has been somewhat successful, we've gotten about 10% of our followers to do this, which is a very respectable number. It also opens up a communication bridge with the user, and we respond to every message.
All of these scripts are completely randomized and set to certain time intervals, culminating into a pretty cool experiment. With this we'll see just how many Twitter followers and Facebook likes we can get while not being aggressive. We'll also be able to test out the effectiveness. We've come to find that by getting followers through this process, the quality isn't too great. We've been tweaking these scripts periodically to make them less intrusive and therefore get more quality customers. Eventually (when perfected), we plan to roll out these scripts to our own clientele, automatically promoting their website and business.
Ambilyrics also forces us to experiment with a large database, something we've always wanted to do. We have almost 200k songs, so optimizing the database has started to become a priority. This will give us practice should the user-base ever get out of control and make a huge spike. We will continue to make the site as simple as possible so we can get more users on board.