Tuesday, March 19, 2024

Data Science: Integrate It.

Datamation content and product recommendations are editorially independent. We may make money when you click on links to our partners. Learn More.

This article is part of a series on Data Science. See also Accenture’s Relying on Traditional HR Will Lose You the Analytics Talent Race.

Most business leaders know that analytics is too valuable to be decoupled from the daily business of an organization. To achieve its promise, data science needs to be an integrated partner that works with the business to perform as a single unit. Unfortunately, many organizations are still pursuing data science in a silo, treating it like an independent think-thank or a back-room function. Too often, separate teams have separate goals and priorities – which inevitably lead to lost value.

When we encounter this issue with our clients, often it plays out in one of two ways:

The Communication Gap –  When teams are siloed, there’s inevitability a communication gap and a lack of understanding around shared requirements and business goals. For instance, a product manager at a gaming company sends the data science team a short email requesting a churn prediction model, without giving any further specifics.

The data science team doesn’t question the short and generic briefing, and months later, delivers a churn model that the PM says is unusable. The PM wanted predictions that pointed to players ending their online sessions, so they could trigger incentives to keep playing. But the director failed to say so in his briefing, and the data science team didn’t have the opportunity to ask for detailed information on how the director intended to use the model; thus, their model provided predictions about customers who had already played their last session.

Misaligned Expectations – When the requirements are clear and the data science team understands how the business will use a particular model,  miscommunication can still persist. For instance, say the CFO of a CPG company asks the data science team to predict customer value – which he defines as Gross Revenue over a four-year horizon — to help establish the company’s financial projections.

The data science team delivers a model that is 98 percent accurate and awaits the the CFO’s jubilant response. But the CFO is not at all pleased. He expects the model to be 100 percent accurate. While such accuracy is possible in accounting, it’s not a realistic expectation for a predictive model – which the data science team assumes the CFO knows and therefore does not articulate at the start of the project.

The Reality: Separate Teams, Separate Outlooks, Lost Value

These examples illustrate several obstacles that can prevent data science from delivering real business value:

·  Delayed involvement – Data science is often brought in to analyze the results of a strategy that’s already been defined or an action that’s already been taken, which is formula for missed opportunities. Introducing data science earlier can ensure that the right data is available to support strategy, analysis, and action. 

·  Limited communication – Because of a lack of integration, the data science team often receives requests from people in the business who struggle to define their needs. But precise communication is imperative to ensure actionable results from data science projects.

·  Undefined intentions – As the gaming company learned, without a clear understanding of how a model will be used, the data science team may develop something that does not meet the business’ needs. Data science will always provide an answer, but the business use case needs to be clear from the beginning.

·  Success is not defined – It is crucial for the data science team and business stakeholders to agree in advance on the definition of success. Each team should share its definition of success – the case of the CPG company, 98 percent or 100 percent — and agree on what is achievable.

The Fix: A Common Purpose with Shared Goals

Organizations can address these issues by applying the following three-step approach:

1.  Rethink business engagement – Data science resources should be included in projects from start to finish. Make them part of the team. In many successful companies, data scientists physically split their time between sitting with data science colleagues and with business stakeholders. This fosters collaboration and understanding and provides the foundation for a long-term relationship.

2.  Focus on outcomes – Data science needs to focus on outcomes rather than insights. Every data science problem should start with agreement on the desired outcome and a trackable metric.

3.  Track value – Set up a Value Realization Office to measure, track and showcase the benefits of analytics, ensuring that priorities are driven based on capabilities that deliver the most value to the business.  Tracking value means proving value.

The result? One team, one dream, one outcome. The business and the data science teams work hand in hand to embed data and intelligent technology into their core processes and deliver tangible value for the enterprise.

About the Authors:

Robert Berkey is a managing director at Accenture Applied Intelligence, where he leads the Strategy & Transformation offering globally.

Dr. Amy Gershkoff is a data consultant; she was previously Chief Data Officer for companies including WPP, Data Alliance, Zynga, and Ancestry.com.

Brandon Joffs is a managing director at Accenture Applied Intelligence’s Strategy practice.

Subscribe to Data Insider

Learn the latest news and best practices about data science, big data analytics, artificial intelligence, data security, and more.

Similar articles

Get the Free Newsletter!

Subscribe to Data Insider for top news, trends & analysis

Latest Articles