In recent years, it has changed from a “content company into a product company”, the Post’s director of product Julia Beizer told the audience at the Digital Media Strategies event in London today.
“What product means at a content company is balancing the competing needs of news, design, development and the business,” Beizer said, “and letting the consumer have a seat at the table.”
At The Post, the process of building a new product in the newsroom is divided into five key steps, she explained.
Define the audience
“Our biggest competitors these days aren’t The New York Times or CNN. They are Hulu, Netflix and any place a user might spend their time,” Beizer said.
After purchasing The Post, Bezos challenged the team to develop a new product to capture a “new national audience”.
The initiative came to be known as Project Rainbow and the app first launched on Kindle Fire in November 2014 before it was rolled out on iOS, Android and Apple Watch.
Looking at the data, the team discovered The Post’s audience who was coming to it for its politics coverage had a wider interest.
“The first thing we did was to think about how we could balance that type of coverage with news from a broader circle,” said Beizer.
The design of The Post’s new app was inspired by a rising social media trend at the time, using a card format the team calls Brights. These were created by a team of designers who were brought over from the print side, which “helped the work resonate”.
“Historically, content companies have thought about what they do and then tried to find an audience for that, but we did it in the opposite way.”
Set goals for what you are trying to achieve
In 2013, the outlet updated its suite of apps with Post Classic, a subscription-based replica of the print product available in PDF format on mobile devices.
When people started complaining that it took too long to download, the team set off to update the app in an attempt to make it load faster.
“Product development helps narrow your focus – what exactly are you hoping to achieve with a piece of software?
“We couldn’t share off a significant amount of time," Beizer said, "so we thought, ‘is our goal to make it faster or just get readers to perceive it as being faster?’”
After deciding on the latter, the product was revamped to preload the first two pages of the newspaper when the reader clicked on an issue.
This gave people the impression that stories were loading instantly, while the app worked in background to make the rest of the content available as the user scrolled through.
Start testing in the early stages
The Post’s app built under Project Rainbow was extended to desktop and mobile web browsers in 2015, to “try and share the mobile and social audience,” said Beizer.
“People come in for an article, they leave and we might never see them again, so we experimented with the user experience to see if we could get them to stick around longer.”
One of the features introduced was a preview of the next story within the article a reader had landed on.
The team conducted 10 tests over a period of four months, seeing a 22 per cent growth in engagement per page.
This approached helped shape how product testing is done at The Post, explained Beizer, outlining key elements:
- start small: “get something out there even if it’s not fully developed”;
- ramp up: “grow your sample size – if one experience is beating another in A/B testing and the difference isn’t significant, you won’t learn anything”;
- be bold: “dramatic tests produce quicker and more actionable results, and we have to be creative about the experiences we choose to take on”.
Measure the success of new initiatives
The Post is betting big on distributed platforms – it was among the first news outlets to “go all in” on Facebook Instant Articles and one of the publishers who adopted Google’s new Accelerated Mobile Pages initiative.
“If these platforms create better experiences for our users, people are going to be there and we want to be there too,” said Beizer.
“But with each new platform comes a new set of metrics that will then change when you shift to another platform, so we need to be clear about what success looks like for the company.”
To measure its success on Instant Articles, The Post started looking at one metric in particular: visits per visitor.
“Do people come back more often when they see the reading experience can be better?
“We saw that a significant amount of users returned more often within seven days,” Beizer said.
Take advantage of what works
An important aspect when investing time and resources into building a new product that turns out to be successful is knowing how to leverage that. “When you dig into your data and find something that’s working, how do you put more eggs in that basket?”
Beizer gave Bandito as an example – a new tool developed in-house to automate headline testing and article optimisation in the newsroom.
Bandito allows Post editors and journalists to create multiple variations of the same story before it is published, alternating elements such as the headline or the main image, which can significantly determine whether a reader clicks on the article or not.
The tool then pushes out all versions of the story and after monitoring performance, it automatically offers people the option that has performed better.
The Post has already seen up to a four per cent increase in the click-through rate on certain stories optimised with the tool and it plans to expand its use to other verticals, Beizer said.
“It helps us measure when our work resonates with readers, but at the moment it's limited only to people who come to our homepage.
“It’s something we will definitely double down on, as we look for ways to bring it to social and distributed platforms,” she added.
Free daily newsletter
- Report: What publishers prioritise in their Facebook strategy
- Tip: Check out this list of free digital resources for newsrooms of any size
- Media24 is training students in South Africa to set up news websites for their schools
- Distributed news: 3 takeaways from a discussion between digital editors and technology companies
- Tip: Bookmark this list of common mistakes to avoid when working with data