When he was demanded to cough up Diply’s secret recipe for generating 1 billion monthly views and maintaining an Alexa top-20 ranking for North America, Chief Technology Officer Gurminder Kandola spilled the beans.
Except there’s not really a secret, he says. London, Ont.-based Diply is poring through the Google Analytics data on its content just like every other publisher is doing. It’s just going that extra mile and taking action on what the data is telling them about their audience. That data interpretation is done through a combination of a technical team and in-house machine learning tools.
“When you’re examining analytics at a large scale, you need tools to help you understand it further than from what Google Analytics provides,” Kandola says. “From a technology side, it’s how you understand what that user is interested in and that they’re engaging in the content to find it useful.”
Kandola joined the Diply team in May 2016, shortly after another stint at Toronto-based social commenting platform Viafoura. He’s tasked with leading Diply’s product and technology teams, helping to build the social content brand up to compete with the likes of Buzzfeed. He’ll also be a judge at the upcoming Developer 30 under 30 awards hosted by Plastic Mobile, for which ITWC is a media sponsor.
The firm also recruited former Twitter Inc. vice-president Kirstine Stewart in September 2016 to help it woo big brand advertisers.
As the parent company name GoViral Inc. infers, much of Diply’s success has come from marketing its content on social media – namely Facebook. It’s been developing more video content towards that end ever since Facebook’s algorithm started favouring that variety of content. That pads out the bite-sized articles on topics ranging from inspirational to DIY to style and parenting.
If, like some people interviewing Kandola, you wonder how a site running articles like “14 Ridiculously Cute Ideas for Crocheting” and “How to Make Batarangs and Get One Step Closer to Being Batman” is ranked as Alexa as being the website with the 11th best reach in Canada and 15th in the U.S. and the U.K., here’s the advice Kandola shared:
1. Know your audience
Diply’s team spends just as much time analyzing their audience as they do writing new content, Kandola says. That results in a team that understands exactly what their audience is looking for and how to craft the perfect experience for their reader.
There’s more to understanding audience behaviour than just knowing if a person is reading an article about making banana bread, he explains. There’s the context of whether that person is watching it from a smartphone or a computer, whether they’re doing so at home or while on the go, and whether they watch a video before reading the text.
It’s Kandola’s job as CTO to be the source of truth when it comes to measuring audience interaction.
“I have to understand what they’ve viewed, how they viewed it, and where they came from,” he says. “So you can sit there and say this particular type of group tends to like a ‘black Mercedes’ and you can give them ore of that type of content.”
Diply’s recommendation engine uses a topic extraction capability to understand what the text on the page is about, and automatically provides more links the reader is likely to be interested in.
2. The device matters
Speed is everything when it comes to web content, Kandola says. Users on mobile devices are the least patient.
“If a user came over from Instagram and our site doesn’t load fast enough, we just lose that user,” he says. “It’s important for me to highlight what’s important to you on that little screen.”
Once it does load, the content displayed on that small smartphone screen better make the best use of the real estate that it can.
3. Never stop A/B testing
Diply is always conducting A/B testing behind the scenes, trying different scenarios to see what will stick with readers. This helps the team form a generally good idea of what the audience is expecting, Kandola says.
For example, for a good DIY template, you might wonder if it makes sense to put the video at the top of the page, or the list of materials a person needs to complete the project. Instead of guessing, you could set up an A/B test trying both layouts and then seeing which one gets better engagement.
4. You’ve got five seconds
That’s literally how long it takes before readers’ attention starts to fall off these days, Kandola says. So it doesn’t make a lot of sense to focus on all long-form content.
“Publishers focus too much on the art of creating content,” he says. “They forgot it’s both art and science and they need to practice them together.”
Publishers should view themselves as technology companies and stop trying to predict what their audience wants on an annual basis, he says. Instead, try to learn something about your audience on a daily basis.
“Ask the machine what patterns it’s recognizing and try to do something about that,” Kandola says. “We are doing what other people are doing. We just analyze the content in more depth.”
5. Viral content can’t be programmed
“There’s no one algorithm that will tell you what’s going viral,” Kandola says. “If you only concentrated on those things, you could have fatigue from your users.”
Instead, try to offer your users a customized experience and make sure the content is offering real value. At Diply, that can mean an article that leads to a good laugh, or a DIY video that helps you make dinner that night. When a piece of content does succeed, go into depth to find out the reason why.
“Publishers don’t ask ‘why’ a lot,” Kandola says. “They’re still running traditional IT departments that need to start building out data teams that are using data to help make these decisions.”
Half the challenge often comes from convincing a writer that there’s something to learn about the audience from data. It takes a mind shift that involves an embrace in experimentation and a willingness to let data help you in crafting a great story.