Create data products in preference to static reports
Agile - fast moving, focused on business results
Collaborative engagement model with stakeholders
Strong design ethos - using visuals to tell stories with data
Our product managers bridge the gap between technology and the needs of the business. They coordinate the many moving parts of a project, balancing the demands of stakeholders, and organizing the workflow of the team.
Our data science team draws from a wide multi-disciplinary background. They are experts in machine learning, statistics, computer science, computational biology, etc. Our diverse backgrounds allows us to find the best approach to every problem.
Our data engineers are polyglots by design. They need to be able switch technology contexts quickly and are inquisitive and able to dive deep into complex data-related questions. They must be able to deal with any sort of data, small or large, structured or not.
It is critical to look after the interest of the customer in a rapidly changing digital landscape. We define and maintain privacy policies, standardize data collection, while educating the business and ensuring they are in compliance with ethics and standards.
Through design, we are able to communicate complex ideas to a broader audience. As our team builds data products, design works to enhance the user experience with interactive data visualizations, infographics, and help innovate new ideas for various digital platforms.
Experiments & Bespoke Analysis
Experiments provide the foundation for informative analysis. The team aids in the design of experiments around the business while also initiating experiments for model development. Additionally, we are able to assist with point-in-time analysis critical for Overview business decisions.
Statistical & Operational Modeling
A main focus of our team is building models—mathematical representations of real world behaviors or events. They can be used for multiple purposes: gain insights, make predictions, and identify patterns, which help us understand our customers and make the business grow.
Often our models and analyses evolve, and are developed into functional products that serve the business and its customers. We have built a real-time article analytics tool for the Newsroom and interactive network visualization tools for our Risk & Compliance customers, among others.
News Helm provides the Newsroom with unprecedented access to real-time analytics through a readily available, self-service web-based application.
News Helm launched in July enabling section editors to monitor content performance within their bureaus: View Site
Large-Scale Data Analysis
When the newsroom receives large-scale data sets, Data Science is on hand to quickly process data into analysis.
As a preliminary text-parsing exercise that will form part of a future document processing pipeline, Data Science delivered on Hillary Clinton's private emails that were publicly released. This was published as part of a Capital Journal blog post on the Wall Street Journal website: View Site
Consulting for WSJ Interactive Graphics
Dow Jones Interactive Graphics Team builds infographics to complement Wall Street Journal content. Data Science provides statistical advisement when requested, resulting in creative collaboration between the two groups.
The NCAA Madness Machine is a recommendation tool for constructing personalized college basketball tournament brackets: View Site
Several customer data models form the foundation of the Knowledge Engine (KE). The objective of the KE is to enable personalization across all customer touchpoints.
Machine Learning to Map Subscriber Segments to Digital Reading Patterns
An independent research group (LRW) established customer personas for The Wall Street Journal readership. To use these personas in operational targeting and decision-making, a machine learning model was built to match customers to their personas.
Targeting Segments Increases Conversion
To prove the operational value of LRW segments, an experiment was designed and executed through a targeted marketing ad on WSJ.com. The experiment found that Career Driven Leaders respond to targeted advertisements at higher rates, p=0.003*.
The positive experiment results allowed DS&E to release the model into production. DS&E organized, coordinated and ran the first experiment using segments and targeted ads which provided insights into structuring future experiments.
SPA: Subscriber Predictive Attrition
SPA delivers accurate, actionable, standardized measurements, and predictions of future subscriber churn across News Corp properties.
Paywall Optimization Through Experimentation
To infer the impact of moving WSJ.com behind a hard paywall, a controlled experiment is planned for August. The experiment will evaluate the effect on ad revenue and new subscriptions.
Coming soon September, 2015.
Operations Management with Predictive Modeling
In an effort to determine which print delivery contractors were over and under performing, Data Science established a benchmarking model to score contractors.
The model outputs weekly reports to teams managing paper delivery.
WSJ+ Impact Analysis
The WSJ+ Loyalty Program launched for subscribers in 2015. To assure the program's effectiveness, Data Science delivers quarterly analysis on the program's impact on retention.
CAS: Customer Appraisal Score
CAS determines the potential monetary value of WSJ subscribers utilizing both historical usage and subscription revenue data, as well as future predicted behavior.
LEAP: Loyalty, Engagement, Affinity, Patterns
LEAP is a unified, customizable and extensible framework that summarizes a large set of metrics into four consolidated customer-centric KPI's known as Loyalty, Engagement, Affinity and use Patterns (LEAP) scores. LEAP scores are directly actionable in audience development, customer insights, and personalization campaigns.
Professional Information Business (PIB)
KEPLER: Risk & Compliance Tool
KEPLER provides a highly visual, responsive, and interactive user experience to the R&C (Risk and Compliance) dataset. This product also contains the tools and pipelines to regularly ingest R&C data updates. View Site
NEWTON: kNowledge Extraction Workflow from Text and News
NEWTON is a data processing and predictive pipeline that analyzes large volumes of news and other unstructured content (> 300K articles per day). With NEWTON we can semi-automate the identification of key relationships in FACTIVA content and generate hypotheses (or leads) to help prioritize content to evaluate, keep or reject.