- New Industry Study Reveals Chief Data and Analytics Officers Lack the Resources to Deliver on Enterprise AI/ML Innovation Goals
- 95% of business leaders are betting on AI to increase company revenues but underinvestments across data science spanning people, process and technology limit AI/ML transformational potential.
SAN FRANCISCO, March 14, 2023: Domino Data Lab, provider of the leading enterprise MLOps platform trusted by over 20% of the Fortune 100, today released a new industry report, “Build A Winning AI Offense: C-Level Strategies for an ML-Fueled Revenue Engine.” The study determined that while data science is now a key revenue and innovation engine for the enterprise, most enterprise data and analytics leaders are inadequately resourced to deliver on what business leadership wants from AI and ML innovation.
The survey of 100 Chief Data Officers (CDOs) and Chief Data Analytics Officers (CDAOs) at companies with $1B+ in revenue, conducted by Wakefield Research, paints a stunning picture of the mounting revenue expectations put on these leaders and their teams, the organizational imbalances data execs say their leadership must correct, and the toll that underfunded, understaffed and under-governed data science practices take at many large organizations.
Data Science Teams Unprepared to Deliver on AI/ML Innovation Despite Corporate Revenue Expectations
Under pressure, the majority of CDOs and CDAOs (67%) are shifting their organization’s data posture from defensive; i.e., data management, compliance, governance and BI modernization, to offensive; i.e., driving new business value with analytics, ML and AI applications. As such, it’s no surprise that nearly all (95%) say their company leadership expects investments in AI and ML applications will result in a revenue increase.
Yet, while business leaders increasingly look to data science to be a key revenue engine and a driver of innovation, resources such as budget, people and preparedness are not aligned with these corporate priorities. Indeed, data science is not funded to live up to leadership expectations – less than a fifth (19%) say their data science teams have been provided sufficient AI and ML resources to meet leadership’s expectations for a revenue increase.
“Data science executives need proper resources, empowerment and support to achieve revenue and transformation goals,” said Nick Elprin, co-founder and CEO of Domino Data Lab. “Boards and the full C-suite must invest in CDOs and CDAOs and put them in charge of people, process and AI/ML technologies, or risk existential competitive pressures.”
Put Me in, Coach: CDOs and CDAOs Are Ready to Take the Reins, and Budget
Many CDOs and CDAOs believe they play second fiddle to IT on a variety of AI/ML issues.
- 64% say IT makes most data science platform decisions at their company: IT departments lord over data science teams, yet underfund initiatives that can positively impact the bottom line.
- Virtually all CDOs and CDAOs (99%) agreed that it is difficult to convince IT to focus their budget on data science, ML and AI initiatives rather than traditional IT areas, such as security, governance and interoperability.
- Nevertheless, more than three-quarters (76%) of CDOs and CDAOs see driving new business results with AI/ML as at least one of their top three priorities for 2023.
Unleashing the Full Potential of Data Science: Overcoming Pain Points Beyond Funding
People, process and technology are critical pain points that data executives believe stand in their way to outperforming competitors with data science. To build a winning data analytics offense, CDOs and CDAOs believe that their organization must not only modernize their internal team structures and elevate the roles of CDO and CDAO, but also gain centralized support.
- They are nearly unanimous (99%) in saying that centralized support was mission-critical for their organization’s data science, ML and AI initiatives, such as developing or expanding a Center of Excellence, or implementing common data science platforms.
- Almost all (98%) said that the speed at which companies can develop, operationalize, monitor and continuously improve AI and ML solutions will determine who survives and thrives amid persistent economic challenges.
- Though AI innovation is at a premium across industries, teams are flying blind, and struggle to measure AI/ML impact. 81% say their teams’ current toolsets are less than fully capable of measuring the business impact of AI/ML.
Lagging Capabilities Result in AI Risks with Negative Impact Today
- Rising governance and responsible AI risks: Respondents unanimously (100%) said their organizations have experienced negative consequences due to challenges developing and operationalizing their data science models and AI/ML applications – 43% have lost business opportunities while 41% admitted they have made poor decisions based on bad data or analysis.
- High stakes / dire consequences: 44% of CDOs and CDAOs believe failure to properly govern their AI/ML applications would mean losing $50 million or more.
- Startling lack of governance tools: Shockingly, despite high awareness of the risks, 46% of data execs say they do not have the governance tools needed to prevent their data scientists from creating risks to the organization.
“Being model-driven is essential for success, but CDOs and CDAOs often lack the authority to lead IT and other stakeholders towards these goals,” said Kjell Carlsson, Domino’s Head of Data Science Strategy & Evangelism. “This study clearly demonstrates that they both want and need to take the reins and get on the offense, and the rising tide of data regulations and governance needs makes them perfect for the job.”
The AI/ML Divide Is Real and Growing
In today’s climate of rapidly rising data sovereignty regulations, hybrid- and multi-cloud capabilities for training and deploying models wherever the data resides are more important than ever. The study revealed just how important those capabilities are, and how fast the divide between companies is growing. Companies without AI/ML platforms enabling hybrid- and multi-cloud model training and deployment were found to lag behind those that do by an average of five years.
To access the full study: download it here
