How The Globe and Mail, together with Sophi, uses Snowplow to deliver advanced data applications

Background

Founded in 1844, The Globe and Mail is Canada’s leading news media provider. With a weekly readership of 6.1 million, it’s considered a national institution. Despite its reputation as the country’s most highly regarded newspaper, it hasn’t been immune to the huge challenges facing the media industry. With the shift towards digital content consumption over the past two decades, The Globe and Mail decided to invest in significant innovation in order to ensure that its world class journalism wouldn’t suffer from the decline of print newspaper revenues.

In doing this, it turned to behavioral data. Over the past 10 years, The Globe and Mail developed an incredibly advanced artificial intelligence-based automation, optimization, and prediction engine called Sophi, which is fueled by behavioral data captured by Snowplow.

In many ways, what it has achieved can be considered the gold standard of how a media organization should use data; since its implementation, this new platform has been responsible for driving tens of millions of dollars in incremental revenue.

By combining Artificial Intelligence (AI), Machine Learning (ML), and Natural Language Processing (NLP), The Globe and Mail managed to navigate the ever-changing news media landscape and achieve its commercial goals.

In this case study, we’ll explore the challenges that led The Globe and Mail’s team to develop Sophi, and how they’ve used Snowplow data to deliver advanced data applications and secure the future of this historic newspaper.

Challenge

Back in the early 2010s, the publishing industry was in disarray. Alongside the rapid decline in the print market, the rise of monopolistic ad platforms was negatively impacting the ad revenue of media organizations. To survive, many news outlets and content providers hurried to make better use of their data.

The Globe and Mail recognized the urgency of the situation and decided to make early, significant investments in the formation of a data science team.

They started by implementing a number of packaged tools in order to address specific needs. As they identified more needs across the organization, they implemented more tools; within the first couple of years, they had invested in a whole host of point solutions.

Before long, however, they realized that their setup, despite the significant initial investment, didn’t fully meet their requirements. The packaged tools weren’t able to capture all of the data that the newsroom needed, and the data it did collect was effectively siloed.

Sometimes the journalists and editors would ask questions that we simply didn’t have the data to answer.”

MIKE O’NEILL, CEO, SOPHI

Encountering these limitations, Mike’s team decided to construct their own pipeline to capture and deliver behavioral data. Having built a full set of software development kits (SDKs), however, they came up against different challenges; the maintenance of basic tracking and of the preparation environment, for example, took valuable time from creating products downstream.

They needed a more manageable, flexible solution; one which enabled the data science team to empower multiple stakeholders across the business and provide a boost to revenue.

Solution

This is how Sophi was born. At its heart, it was intended to help The Globe and Mail’s Editor-in-Chief answer the question: ‘What should we do more or less of?’. It would give editors and journalists visibility over how their work contributed to the organization’s commercial success, and would use predictive modeling to maintain ad revenue and drive significant additional subscription revenue.

In order to achieve this, their solution needed to be fueled by clean, highly-structured behavioral data. For Mike’s team, Snowplow was the obvious choice for capturing and delivering this data. It provided the flexibility to track exactly what they needed to, whilst making it easier to comply with changing privacy regulations and eliminating thankless tasks. Recognizing these benefits, The Globe and Mail joined the Snowplow Open Source community and became one of its largest contributors.

We needed reliable, scalable and flexible data collection. Taking advantage of Snowplow was a no-brainer – it meant we didn’t have to reinvent the wheel, and we could concentrate on getting value out of our data downstream.”

LYDIA JIN, MANAGER DIRECTOR OF TECHNOLOGY, SOPHI

The Globe and Mail’s AI platform uses behavioral data captured by Snowplow (alongside other types of data) in some incredibly advanced ways. The most important of these is in the creation of a composite weighted content score, known as the ‘Sophi Score’. This score takes into account the ‘value’ of a piece of content across different platforms, depending on what the platform deems its purpose to be (which can be to generate advertising revenue, subscriptions, retention, email registrations, etc.). Whilst there are many factors that contribute to this score, it’s driven by how users engage with the content of each article.

In this sense, the ‘Sophi Score’ allows The Globe and Mail to evaluate different types of content on a single plane, acknowledging that each article contributes to the business in a different way. This gives journalists a better understanding of how their work supports the commercial objectives of the organization, and enables newsroom chiefs to allocate resources based on established trends (i.e if finance articles drive more subscriptions than sports articles, and require less effort to produce, then a news chief may wish to transfer resources from the sports desk to the finance desk).

The suite of AI solutions also helps editors by taking care of the decision of whether or not to paywall an article. Rather than rely on a fixed meter or guesswork, it calculates and compares the expected ad and subscription revenue of an article, and makes a predictive decision based on what would bring the most value to the business. This has enabled The Globe and Mail to shift its focus towards a subscription-based revenue model whilst also increasing its revenue from ads (and paradoxically boosting reader engagement in the process).

Sophi has helped The Globe and Mail move from a 70% advertising revenue and 30% reader revenue split, to a 30% advertising and 70% subscription revenue split, and it’s not losing market share on the advertising side.”

MIKE O’NEILL, CEO, SOPHI

To further boost conversions, the platform autonomously curates 99% of the content on The Globe and Mail’s websites and mobile apps, and updates every promotion slot on every page every 10 minutes based on real-time data. While editors retain ultimate control over the primary page above the fold box on their home and business pages, the platform curates the rest of the pages, and sometimes differently for certain segments of users. For example, users coming from a certain referral source (like social or from a certain geography) that are yet to subscribe, are presented with articles that have been assessed as optimal for driving subscriptions, whereas existing subscribers are offered articles that have been assessed as optimal for retention.

Results

By harnessing high-quality behavioral data with Snowplow for advanced analytics and deep learning models, The Globe and Mail created an extremely powerful solution for automatically curating and optimizing content distribution. Since its implementation, Sophi Site Automation has increased click-through-rates from the Globe and Mail’s homepage more than 10% and increased the subscriber acquisition rate by 10%. When taken alongside Sophi’s contribution to ad revenue (estimated to be several million dollars), and its automation of time-intensive activities, this AI-powered platform is proving fundamental to the ongoing commercial success of Canada’s most important newspaper.

This hasn’t gone unnoticed outside of its newsrooms – The Globe and Mail and Sophi’s AI platforms have won multiple awards, including Digiday’s 2021 Media Award for Best Publisher Platform, the World Association of News Publishers’ (WAN-IFRA) Digital Media Awards Worldwide 2021 for Best Paid Content Strategy, the International News Media Association (INMA) 2021 award for Best Use of Data to Automate or Personalize, and the Online Journalism Association (OJA) 2020 award for Technical Innovation in the Service of Digital Journalism.

How you can get started with Snowplow

To learn more about how Snowplow can empower your organization with behavioral data creation, book in a chat with our team today. Alternatively, Try Snowplow is our free, easy-to-use version of our technology, which allows you to create your own behavioral data in under 30 mins.

‍