Six months ago we started a publishing experiment. We’ve been creating websites and content at Sound Alliance for more than a decade, with sites like inthemix, FasterLouder, Same Same and Mess+Noise, but after our success with a tried and true formula we asked ourselves if we’re repeating things over and over just because that’s the way you’ve always done it.

That’s exactly what we did earlier this year when we came up with the bones of what we thought the future could look like. Given audience trends, the desktop experience wouldn’t be the focus. Given tech trends, it had to be easy to adapt to technology that hasn’t even been invented yet. Given advertising trends, we couldn’t just squeeze more ads onto the same page to compensate for decreasing yields. And given an increasingly competitive market, it needed a fresh attitude to help it stand out from the crowd.

The result was Junkee, a pop-culture destination for 18 to 29 year olds that’s serving as an experimental publishing model. It is genuinely mobile-first, designed to work better on mobile and table than it does on desktop; the content has attitude and is written by a unique family of writers; has an open-source approach to technology, just one high-impact ad per page and a revenue model that’s built around the new area of native advertising.

The results have been well above our expectations, with Junkee on track to deliver around 220K unique domestic browsers in September according to Nielsen Online.  We’ve also learnt a lot of lessons in those six months about what has and hasn’t worked. Here are some of them:

Reducing the number of ads can work as a business model
It’s the most basic publishing principle: need more revenue? Increase the number of ads. Instead we went the opposite direction and only allow one ad per page on Junkee. It’s a high-impact IAB Rising Star that allows advertisers to essentially create a mini-site within the banner. For the first six months, we had three sponsors exclusively – Telstra, Rekorderlig Cider and Deakin University. They took a risk on a new title, and it paid off as we delivered them 534% more ad impressions than we’d promised. The click-through rates were also above average, 0.51% on the single mobile adhesion banner, and Deakin University reported an interaction rate of 25% on their desktop banner ad. All positive signs that when you remove the ad clutter, the audience takes more notice.

User behaviour is changing faster than ever
Based on our other properties, we originally forecast for Junkee to do three pages per unique on mobile, and six on desktop. We were wrong. Facebook accounts for roughly half of all referral traffic to Junkee (split almost evenly between mobile and desktop referrals), and most users are reading content, then flicking back to Facebook to continue scrolling. The actual numbers per unique are closer to half of what we predicted, making us revise our forecast impressions for the next six months.

Long–form content does have a place
People only want to read lists on the net, so you’ve got to keep it short and sweet, right? Wrong again. Our research told us that young people will share content if it makes them look funny or knowledgeable, and that’s guided our content strategy. Some of our most successful stories have been long-form essays like The Case Against Free. Junkee does not have historically “sticky” content like forums and photos, which is why we were pleasantly surprised that the average session duration in August was an extremely healthy 4:06. It’s heartening to know there is a time and place for long form content. Just like there’s a time and place for lists.

You don’t have much control over what goes viral
There’s a combination of high- and low-brow content on Junkee, reflecting the audience. Sometimes it’s an essay like Why The Creators of Breaking Bad Don’t Deserve Our Money, and other times it’s list of animals being cute or ugly. Then sometimes stories go really crazy. In July, Editor Steph Harmon wrote a relatively harmless piece about a new product called NeverWet, teamed it with an intriguing picture and compelling headline – Someone Invented Magic And It Is Freaking Us Out – and it quickly became a viral piece of content viewed over 1.5 million times in under two months, with over 237K Facebook likes.

But you can control what happens next
When more than a million people came to Junkee in its fourth month, we were able to customise the site so they read our other stories. In doing so, we were able to push another story that has been flying under the radar into viral territory. This time it was a think-piece about the growing meme of gender swapping. Through clever placement off the back of the tsuami of traffic coming to read the NeverWet story, we were able to generate over 180K reads in its own right.

Native advertising has so much potential
Native advertising, where high-quality brand funded content is placed directly in stream, is the primary revenue model for Junkee. We’ve so far worked with brands as diverse as HCF, Intel, J&B, Stoli Vodka, and Contiki to create custom content that our audience seems to enjoy. We’ve found that native content tends to have an almost 30% higher engagement rate (ie. the likelihood of someone liking, commenting or sharing it) than the average Junkee story. This is the first experiment with native advertising for almost every marketer we’re working with. HCF reported “one of the most successful months on record among the key under 30s segment. Products tailored to young singles sold more than any other product segment in June 2013 on top of a 2% increase in completed quotes”. While it can’t all be attributed to their native campaign, it is a promising sign for this brave new intersection of content and marketing. It’s also been so successful on Junkee that we’re bringing native advertising onto all our other sites.

Real world connections are vitally important
Junkee may be online, but it’s the connections made in the real world that are helping it grow. Every two months we throw a Junkee Do in Sydney or Melbourne, and have a Half Birthday Party coming up (any excuse for a party, right?). By engaging our writers, readers and advertisers in the flesh, Junkee has been able to cultivate the early and important feeling of a community that every successful online title needs to have.

It’s daunting to throw out some of the rules and start from scratch, but it’s genuinely satisfying when you realise you don’t need to keep doing the same thing just because that’s all you know. We’re going to continue to learn a lot over the next six months, with more wins and a few mistakes on the horizon. But gee we’re having a lot of fun in the process.