Better Analytics for Newsrooms with GADash

An in-progress project from the NYT’s Interactive News Team

The process of finding information about traffic patterns can be time-consuming and complicated for the busy, multi-tasking newsroom worker. Tools like Google Analytics can be quite complex—it’s often difficult to understand the information they present, and to use that information well. By nature, these tools are broad in scope to fit different types of websites, so we can often use them to quickly get answers…but do they fit the right questions? And if not, how do we ask better questions?

At the New York Times, the Interactive News team wants to build tools that make the world of analytics more accessible to newsroom teams. Earlier this year, members of the team began working on Google Analytics Dashboard (GADash), an app-specific display of key figures, traffic patterns, and charts for some of our custom story presentations, as well as general page reports.

Why We’re Making It

“How many people meaningfully interacted with this story?”

It’s a great question to ask. But the answers usually come with so many data-oriented caveats, explanations, and delays—“I’ll look it up and get back to you”—it starts to deter people from asking the question.

In the New York Times newsroom, colleagues use Chartbeat, Google Analytics, WebTrends, Localytics, a few in-house tools, and more to answer these types of questions. Many analytics packages have one-size-fits-all answers ready to go, using datapoints like pageviews or sessions. These provide general answers, but ones that reflect the quirks of the provider or datapoint. Explaining those quirks often undermines the question, and as independently presented figures, these measures can’t provide a full picture of audience experience.

Customized analytics displays help journalists view a story’s performance across metrics of specific interest to the newsroom. They can hide ill-fitting e-commerce jargon and complex figures like sessions, pageviews, and conversions. By describing things in newsroom terminology, we can establish a baseline for recurring questions while also guiding the newsroom toward more complex concepts around reader engagement and app performance.

GADash helps the newsroom quickly see how readers are finding and engaging with news applications and individual stories. And it can focus on custom presentations that also better fit newsroom concepts and evolving questions—for example, dates! Default reports, based on classic ecommerce models, often offer weekly, monthly and quarterly timeframes. But for news, a common timeframe is the date of publication to…some point where traffic begins to exhibit long-tail patterns. Our custom charts automatically set their timeframes from publication date to “complete” signals (which depend on the specific piece).

Intelligent timeframes and other news-oriented defaults help greatly in reducing repetitive work we’d otherwise continually do in Google Analytics itself. Consequently, once newsroom staff is back in Google Analytics, reports and figures there become more accessible and more easily understood. We can answer questions about an article or news app’s performance quickly and consistently. Newsroom staff also augment their mental model of reader experience and analytics figures with improved confidence and context.

Another key benefit is for our developers—our work helps them create app-specific dashboards that include custom reporting and performance monitoring. This enables forward-stepping feedback loops for app changes, based on data, with an emphasis on user experience and user performance.

Defining Useful Metrics & Views

When viewing a specific article or page, the user first sees broader metrics like traffic counts (below), then expanding into traffic over time, navigation patterns, and internal/external referral sources.


Broader summary metrics about traffic


Metrics about demographics and referral sources

We’ve put effort into pushing beyond the basics and into charts that convey complex information very quickly. For example, we’ve received requests from newsroom teams to better understand patterns around devices use and time of day.


A visualization of the popular screen sizes with which people visit our pages

Over the summer of 2015, we traveled to the Portland OpenNews code convening with the goal of extending the beta of GADash by adding a couple of visualizations, improving front-end performance, streamlining API endpoint data, and taking initial steps towards making the API open-source. (The API wrapper does the heavy lifting of turning verbose Google Analytics API queries into tidy JavaScript data, ideal for client-side tables, figures and visualizations.)

We’re continuing to work on GADash, with a plan for open-sourcing the data-extraction API with a helpful guide and examples as soon as we can.

Connections to the Coral Project

I (Tara) work primarily on The Coral Project, but also had the chance to collaborate on GADash with the folks on the Interactive News team. The ways in which an organization like the New York Times thinks about incremental implementation and reporting goals are directly in line with some of the user needs we’ve heard articulated.

Many of the news organizations that we’ve talked to haven’t had the opportunity to better understand community engagement on their sites as a metric. Basic community engagement metrics, such as number of comments or community members, don’t provide enough insight into the depth or quality of conversations on the site, as well as all the factors that could lead to that quality. Without meaningful analytics, newsrooms are less able to make informed decisions about how to engage with, moderate, and sustain community.

For our first iteration of core Coral products, we are building an analytics tool that takes existing comment log data to give indicative reputational scores to publishers. In addition to reporting patterns around daily activity, we’re excited about using analytics as a way to build relationships and trust between publishers and community members, by highlighting the best commenters and contributions. As I begin to think through my approach, it’s great to be learning from other developers who care about bringing more meaningful, custom analytics to the newsroom.





Current page