Jacques van Niekerk, Michael Murray & David Bartram-Shaw

Global CEO, President and Chief Product Officer, Director of Data Science, Wunderman Thompson

Supernova Award Category: 

Data-Driven Digital Networks (DDNs) and Business Models

The Organization: 

Created in 2018 when two marketing powerhouses – the world’s first advertising agency and the pioneer of direct marketing became one – WT brings more than 200 years of trailblazing innovation to the creative industry. Today, the company is 20,000 strong in 90 markets around the world, where its people bring together creative storytelling, diverse perspectives, inclusive thinking, and highly specialized vertical capabilities, to drive growth for our clients. The agency offers deep expertise across the entire customer journey, including communications, commerce, consultancy, CRM, CX, data, production and technology.

The Problem: 

Historically WT worked with its clients to help find new customers and create messaging to support brands and solutions. Today, efforts are focused on how it use data and inspire authentic interactions that brands can have with customers. As a result, the organization sought to be less focused on transactional goals and more directed toward driving longer engagement possibilities for its clients. WT manages over 10 TB of data with millions of rows, yet the organization was only training the data science models on a few thousand samples of data due to compute and advanced analytics, machine learning software limitations. Lacking compute and analytical power meant that disparate data sources could not act in concert, limiting the predictive power of WT’s vast data properties.

WT sought a cloud-based, AI solution and machine learning strategy that would provide the flexibility and scalability to maximize the use of its data assets. Analytics experts at WT knew that consolidating these assets would enrich their value. Increasing the volume of data contained would enable data scientists to enlist machine learning and AI to develop more accurate predictive modeling for ultra-personalization of audience segmentation based on respective brand business requirements. To help its clients cultivate more meaningful relationships with customers, WT wanted to operationalize predictive models in a production environment that could scale across the entire organization.

The Solution: 

WT turned to the IBM Data & AI platform to help drive its transformation. IBM partnered with WT to build oneTeam consisting of marketing & advertising industry experts, IBM’s world class data & AI technology platform and advanced data science experts. The team jointly conducted intensive workshops to design and develop the project. Its first task centered on combining its three large proprietary datasets, comprised 10+ TB of data. The team created specially prepared representative assets derived from the data sources for the proof-of-concept. With data accessible through a common data source, they moved to organize the data to support a robust analytics foundation. To standardize formats across disparate sources, data assets from legacy datasets and flat files were loaded into open source into columnar files. The use of AutoAI enabled faster building & deployment of machine learning model with sophisticated training features with higher accuracy & quality.

The Results: 

The outcome of WT ’s data transformation initiative yielded remarkable results. The teams developed a data preparation regime to rationalize data from WT ’s three separate data properties. Comparisons between data points in each source were used to filter out records for desired features and reconciled against one another. The teams were able to subsample tens of thousands of records for feature engineering, applying decision tree modeling to highlight and select the most important features for training data.

The ability to use all of its data, techniques, human insight and understanding, gives WT a best in class capability to find new customers for any of the brands they support. The solution includes all of the data, full-scale, much more advanced machine learning and the ability to run all of that processing on elastic compute inside of various cloud providers IBM’s AI capabilities and WT’s proprietary trusted data sources help provide human insights at the local level who people are, what they buy by category and frequency, and hundreds of other points of context. This solution enables WT’s ability to support brands ability to make better-informed business decisions improving accuracy of predictions for campaign management, audience segmentation and reach.

Metrics: 

WT realized significant uplift over its previous models. Performance was measured along the with increased segmentation depth, response rates rose dramatically, averaging a change from 0.56 percent to 1.44 percent, an increase of more than 150 percent. This new approach to predictive modeling uncovered new customers – new personas – in existing databases WT had previously been unable to reveal, dramatically expanding deliverable customer lists. Net benefits can be categorized into –

  1. Increased response rate
  2. Increased depth of mail file
  3. Scalable pipeline with collaborative and parallel work streams
The Technology: 

WT analysts worked with the IBM Data Science Elite team to develop its AI and machine learning solution using the Cloud Pak for Data for data science and machine learning including a pre-release version of its AutoAI feature.

Disruptive Factor: 

WT disrupted the perspective of COVID-19 data reporting when it swiftly launched a powerful interactive application powered by its Identity Network of human insights and IBM's Watson Machine Learning, to provide a comprehensive view of COVID-19 impacts on communities to help identify and activate steps to support businesses as they navigate the road to recovery.

The Identity Network application connects the health, transactional and demographic data from WT 's Identity Network with publicly available COVID-19 data to create a daily updated, county level view of communities and populations. This initial public view provides meaningful visibility to begin to inform recovery strategy, communications, timing (by location) and activation decisions around programming to accelerate business recovery.

The three areas of the application focus on Health Conditions, Covid-19 and Census (Risk), Health Support within Communities (Readiness), and the Impact of Covid-19 on the Economy (Recovery). The human-insights fueled application was built to inform market level actions and can be broken into further categories, Risk, Readiness and Recovery. WT combined the full range of the Watson capabilities and tool chain with its vast data streams to create a unique view into a critical aspect of the recovery – and in a manner that protects individual privacy and data ownership.

Shining Moment: 

As the COVID-19 pandemic wrought havoc on human health and the global economy, WT used the technology solution described to explore how to help clients navigate the unpredictable route to recovery. Continuing its work the IBM DSE team, WT created the COVID-19 Risk, Readiness and Recovery Dashboard that integrates detailed demographic data & human insights to analyze the health of community and economy on a local level, giving business and community leaders information needed for reopening.

About Wunderman Thompson

Wunderman Thompson is part creative agency, part consultancy and part technology company, our experts provide end-to-end capabilities at a global scale to deliver inspiration across the entire brand and customer experience offering deep expertise across the entire customer journey, including communications, commerce, consultancy, CRM, CX, data, production and technology.