Jun 11

Case Study — MusicNewsHq.com

MusicNewsHq was a network of automated entertainment sites. It operated from 2011 until August 2016.

This post details the successes and failures of the project, which I eventually closed after failing to find a buyer for the price I thought it was worth.

MusicNewsHq originally was a SEO play; a way to drive traffic to a now-defunct song lyrics website. Over time it became an active hub for following your favorite music artists.

MNHQ was always more popular than the lyrics site it was intended to funnel traffic towards, so I kept it alive even after shutting down the former. At that time MNHQ was costing me virtually nothing to host and was a personal playground for technology exploration.

The site ran by itself and generated monthly revenue, but I simply did’t have enough time to keep improving it.

Quick Stats

244,567 Peak monthly pageviews (~8k/day) $6,562.63 Total revenue generated
~5k Twitter Followers 12k Monthly Emails Sent
1,459 Facebook Likes 977 Pinterest Followers
850k Pages Indexed in Google

MusicNewsHq home page June 2016

This article is indended to be a case-study and portfolio piece for future interested parties.

Table Of Contents

## Background — What is it?

At the core, MNHQ is an “all things music” network.

Aggregation bots perpetually collect news articles, albums, videos, pictures, and concert information. That gathered information is then used to generate fan sites for each known artist.

Based on engagement and views, the most popular of that content bubbles up to the home page and genre specific pages.

That same popular content is then shared via our weekly email newsletter, Pinterest, Facebook, Tumblr, and extensive Twitter network. At present count there are ~200 additional twitter accounts which are genre & artist specific.

Initially, I added only a couple hundred popular artists to the site, since its main purpose was to drive traffic to Lyricful. Once shutting down Lyricful, I decided to take the Long Tail approach, continually adding any music act deemed newsworthy. As of today (June 11, 2016) there are nearly 16k artists online. New artists are automatically added to the network on a daily basis; from sites like Billboard, Pitchfork, Noisey, Amazon.com, etc.

## Previous Sale Attempt

In December of 2013 I attempted to sell the site on Flippa.com.
The highest bid was $405. I kept it.

At that time, there was no monetization strategy in place, the design was fugly, and I did a horrible job of explaining exactly what the site did or how it all worked. I’m surprised anyone bid at all.

Despite the failed sale, I still believed in the concept. I thought it could be something a lot bigger than it was, so I set out to improve things and see where it could go.

When time permitted the past few years have seen many changes, including…

  • A responsive site redesign
  • Systems put in place to make news content more relevant
    • Simple machine learning for spam/ham classification
    • “Storylines” that automatically group similar articles from multiple sources
  • Creation of a “backstage” administration interface
  • Expanded data collection
  • Ramped up social media sharing
  • Weekly emails with relevant content to people who sign up for updates

Page views eventually grew to a peak of 244,567 in December of 2014 (~8k/day), and MusicNewsHq has generated $6,562.63 in revenue to date.

Not a roaring success by any means, but certainly better than $405. I still believe the concept has promise, and in the right hands it could grow even further.

## Growth, Monetization, and Decline

After the failed sale I put significant effort into growing traffic and it worked — for a time.

Monthly Sessions for MusicNewsHq.com

I picked up SEO tactics where I could, optimizing the site for terms like “[artist name] new (album video) [year]”, backlinking via social media, and other various tricks. 2014 was looking promising until an unknown, demoralizing SEO hit took me down around June.

SEO is a difficult, dirty game, and unless that’s all you do it’s supremely challenging to make any heads or tails of why you’re suddenly getting less traffic from search. I consulted expert opinions; nobody could agree on what exactly was happening. Google Webmaster tools also provided few insights. After long contemplation, I suspect a ranking algorithm is responsible for the sudden drop in traffic.

Pushing forward despite the setbacks, I optimized traffic for mobile and tuned the site using Google’s page optimization techniques. Once again traffic started to grow around December 2014 before crashing into a downward spiral in May 2015.

It’s obvious that relying on traffic from Google with my lack of SEO skills is a foolish exercise. Realizing this, I’ve been attempting to boost visits through email, social media, and other channels.

Social media currently drives a small amount of traffic, but thats also an uphill battle I’ve not been able to win. Adding a customized home page for users and weekly email newsletter have seen similar (limited) success. The site currently has around 5k registered users that have been verified to have valid emails.

Traffic has been coasting ever since, with a slight downward trend. (Here’s where a sticky mobile app, or someone who really knows how to drive traffic and pump SEO could benefit from purchasing the site).

Monetization Attempts

I’m sure by now you’re wondering how the site eventually made money. It ended up being a combination of things, as shown in this spreadsheet…

#### What Worked

Google AdSense has generated a good amount of predictable income based on hits. This should be no surprise to anyone familiar with content site monetization. They’re the gold standard for a reason. I’ve tried other banner ad placement systems, but none compare to the conversion and money generation potential of AdSense.

Amazon’s Affiliate Program also converts well, but generates nowhere near the amount of revenue I expected — considering 60% of MNHQ’s landing page hits go to the “new album” page, shown below.

Beyonce MNHQ Album Download Page

This page prominently features 3 links that redirect to Amazon — ‘Download Now’, the album cover, and the ‘Merch’ link.

I played with diverting all Mac and iOS traffic to the iTunes Affiliate program, but that converted horribly. All traffic has been switched back to Amazon.

Sponsored Posts

Advertisers continually want to post sponsored content, but unless it’s on message for the site (online casinos, dating sites, etc) I usually don’t accept.

One exception was when a media company contacted me to post articles about videos on Yahoo Music. I gladly took their money, and learned a bit about how Yahoo was doing business to drive traffic to their sites in the process. (Sneaky, Yahoo). I believe there’s some value in providing that type of service moving forward.

#### What Didn’t Work

The biggest disappointment of the monetization attemps has to be the failure of selling tickets through an affiliate network. I was certain that would be a huge money maker. Another swing and a miss…

Ticketmaster long ago killed their affiliate program for the public — although it’s rumored if you’re a “big boy” in the game (Pandora, Spotify…) you can still partner with them. Being a mosquito in comparison I eventually settled on SeatGeek. On the surface, their affiliate program seemed the most fair, their API was decent, and I thought their site was well designed. Man, was that a bad idea.

SeatGeek Click Tracking

MusicNewsHq sent SeatGeek over 6,880 targeted, unique clicks during the course of two years. Can you guess how many of those clicks converted to sales? Just one.

I’ve gone back and forth with their affiliate manager, thinking perhaps that tracking on their end wasn’t setup properly. He assured me that wasn’t the case. I tuned the site to promote only local concerts to visitors, using geolocation based on their IP address. That didn’t improve conversions either.

Local Concert Calendar - based on visitor location

At this point I believe either SeatGeek is outright screwing me for the traffic I send them or their site simply sucks at converting. Neither of these things I can control. Perhaps someone with better powers of design and persuasion could fix this. It’s a huge untapped stream of revenue I’ve not been able to crack.

Selling Promotion & Banners to PR Agents

PR Email Response

From day one I received inquiries from musicians and PR people wanting their press releases and new material reviewed on the site. I received such an overwhelming number of these that at one point I couldn’t respond to each one manually.

Foolishly and without testing, I knocked together an advertisement system, rate sheet, and auto-responder.

PR agents hated it. Apparently, they’re OK with taking musician’s money to secure FREE press, but there’s no way in hell they’re sharing that money with anyone else. I was oblivious to this fine point of online music promotion. One response in particular hammers home the point succinctly.


## Hosting & Tech

MusicNewsHq is a Ruby on Rails site hosted on 5 servers at Digital Ocean.

MusicNewsHq Server Setup

All public traffic is handled by a load balancer powered by Nginx. It passes off traffic to three Nginx / Passenger servers to fufill web requests.

At one point I decided to upgrade to a 16gb database droplet for a little performance breathing room — which costs $160/m. This was by far the most expensive aspect of running the site. On 6/15/2016 I downgraded that to an 8gb droplet, which cuts the cost in half. This should put the site in “break even” territory with current traffic levels.

I’m using Percona MySQL server, which is a MySQL server tuned for better performance. That droplet also hosts Redis for general caching duties, and some Resque workers to process background jobs for Rails.

The final server in the puzzle is a “utility” slice, which runs the aggregator bots, social posting and maintenance scripts, then finally some more Resque job processing threads.

Aggegration Crawler Bots

Content is perpetually collected by Ruby aggregator bots that crawl the web and various APIs. Currently I’m running 4 at a time, but this could be expanded easily to crawl content faster — provided the hardware could support it.

The aggregator bots collect the following for each artist…

  • “Popularity” as determined by the Spotify API
  • News Articles
  • Images
  • Album Releases
  • Concert Events
  • Music Videos

News stories deemed similar by the aggregators are automatically grouped into “storylines”, which have more weight on the site and bubble to the top as “popular content”. This was done in attempt to provide more relevant, useful information to visitors and increase the quality of the overall site.

Some images from external sources that meet a certain criteria are stored on Amazon S3, to prevent “broken images” if servers we’re reblogging from don’t support hot-linking. This could be enabled for all images referenced by reblogged articles if necessary.

Social Network Sharing Scripts

Popular content is shared based on a custom algorithm that takes into account artist popularity, interaction on the site, and some other “special sauce”. Pinterest and Twitter get updated hourly. Tumblr is updated four times daily, and Facebook gets updated once per day.

Custom Home Page / Email Newsletter

Visitors can “watch” their favorite artists by signing in with social network authentication or their email address. If someone signs in with either Twitter or Facebook, the site will scan their social media accounts for artists we know about, then watch them automatically.

MNHQ uses this information to build a custom home page for each registered user, and then emails the most popular content each week to them.

This social follow process could easily be extended for other services, like Spotify, iTunes, or Google Music.

iTunes and Google Music in particular would be a great source of information, and an easy way to boost engagement, provided that data could be accessed by a mobile app. As mentioned previously, I believe this would be a great way to grow traffic dramatically moving forward.

Administration Interface

Of course, the site also has a complete management “back-end” interface, that allows for creation of new artist sites and management of existing ones. Finally, banner ads can be created/placed on the site, and total user management is provided.

## How the site could have been successful

If you’ve read this far, I’m sure you have some ideas on how the site could have been improved. I’ve got three ideas of my own.

Develop mobile app(s)

This is probably the most effort, but I believe it could prove the most rewarding.

People simply don’t consume the type of information MNHQ provided on the web as much as they used to. Creating a native mobile experience that notified users when their favorite artist had a local concert, new album release, or “breaking” news item could see engagement jump dramatically.

The technology behind MNHQ’s aggregation was proven over many years, and could be accessed from a simple REST API.

Once built, a mobile client could be promoted via email to the existing user base and as a pop-up on the web site, which still sees nearly 1k unique visits on any given day.

Change the scope of information presented, repurpose the tech

Building on those thoughts above, perhaps MNHQ was simply too broad of an offering. Perhaps it’s not one giant mobile app that needs to be developed, but rather a few small ones. I can easily imagine small “notifier” apps — for local concerts, breaking news items, or album releases. Might those be received better?

On the other hand, perhaps MNHQ was too narrow of an offering… The same technology used for aggregating could have been a jumping off point to build a Movie/TV fan site network, or a general “celebrity” site network. Imagine a huge TMZ-esque site dedicated to every newsworthy celebrity, event, or music festival.

Fix SEO, cut hosting costs

Finally, it’s been proven through this experiement that my SEO skills are lacking. The site is holding it’s own ranking highly for two terms, but nothing much else…

  • [artist name] new album [year]
  • [artist name] new video [year]

Someone more knowledgable about SEO could easily improve on what I’ve done, and I’m assuming grow traffic again to levels where the income would outpace the current hosting costs.

Following that same train of thought, if you had access to cheaper servers than Digital Ocean you could move the site there and let it coast. It basically ran itself and would have continued to do so.