The world’s most sophisticated companies are overwhelmingly counting on data science as a key to their long-term success, but flawed investments in people, processes and tools are leading companies to fail in their best efforts to develop, deploy, monitor, and manage models at scale.
According to a new study commissioned by Domino Data Lab, provider of the leading Enterprise MLOps platform trusted by over 20% of the Fortune 100, and produced by Wakefield Research, 97% of C-level executives polled say data science is crucial to maintain profitability and boost the bottom line. However, nearly as many say that flawed approaches to data science strategy, execution and staffing make achieving that goal difficult.
Barriers to scaling data science
The study unravels a set of findings that show how and why companies struggle to scale data science, despite their best efforts to do so.
One of the key observations is that short-term Investment thwarts growth expectations. While 71% of data executives say their company leadership expects revenue growth from their investment in data science, a shocking 48% say their company has not invested enough to meet those expectations. Instead, they say organizations seem focused on short-term gains. In fact, more than three-quarters (82%) of those polled said their employers have no trouble pouring money into “splashy” investments that yield only short-term results.
The study also finds companies struggle to execute on the best-laid plans to scale data science. Over two-thirds of data executives (68%) report it’s at least somewhat difficult to get models into production to impact business decisions— and 37% say it’s very to extremely difficult to do so. Nearly 2 in 5 data executives (39%) say a top obstacle to data science having a great impact are the inconsistent standards and processes found throughout their organization.
Companies also face shortages of skilled, productive employees and the tools they need. Nearly half (48%) of data executives complain of inadequate data skills among employees, or not being able to hire enough talent to scale data science in the first place (44%). More than 2 in 5 data execs say their data science resources are too siloed off to build effective models (42%), and nearly as many (41%) say they have not been given clear roles.
Further complicating the issue, 37% of data science executives name outdated or inadequate tools to build and manage models as a key factor leading to reduced data science impact on the business. This may explain why a third of data executives (33%) say not improving models can result in loss of productivity or rework, the study reveals.
“We found that while executives have enormous expectations for revenue growth from their investments in data science, they are not making investments in the right places to truly unleash the power of data science,” said Nick Elprin, CEO and co-founder at Domino Data Lab.
“To properly scale data science, companies need to invest in cohesive, sustainable processes to develop, deploy, monitor, and manage models at scale,” he said.
The study also explored what keeps data science leaders up at night. The results deliver a stark warning for companies cutting corners with data science. For example, a shocking 82% of those polled say their company leadership should be concerned that bad or failing models could lead to severe consequences for the company, and 44% report a quarter or more of their models are never updated.
Respondents name several shocking consequences of model mismanagement, including bad decisions that lose revenue (46%) and faulty internal KPIs for staffing or compensation decisions (45%). Besides, security and compensation risks (43%) and discrimination or bias in modeling (41%) were also reported by decision makers.
Towards data science maturity
The researchers of the report have comes up with a maturity index that provides a framework for achieving success with enterprise data science initiatives. Top findings of the index are shown below in the diagram.
The key findings of the index includes:
- Discipline Delivers Rewards: A powerful 82% of High Maturity companies have had data science make a great deal or fair amount of impact on sales or revenue, including 50% who said a great deal, while just 14% of Low Maturity companies can say the same.
- Maturity Moves Markets: On average, an impressive 69% of data models at High Maturity companies impact business decisions, compared to just 49% at Low Maturity Companies.
- Meaningful Data Science Requires Discipline: Executives at 65% of High Maturity companies say their companies treat data science as a first-class discipline, the same as finance or marketing.
“These obstacles are evidence that doing data science is hard, and progress requires a level-headed assessment of an organization’s “data science maturity” and associated resource needs for achieving the successful creation, deployment, and maintenance of production models at scale,” the company wrote in a blog post today.
Gartner predicted earlier that, through 2022, around 80% of data and analytic initiatives will not deliver business outcomes. This shocking statistic implies that most organizations won’t derive value from data science, despite their huge investments.
This can be explained by the narrow focus that most organizations take towards data and analytics. They spend efforts building technical skills and executing projects transactionally rather than architecting organization-wide capabilities for data-driven decision-making.
As the industry continues to face this challenge, McKinsey found that a small set of enterprises, the data and analytics leaders, excel in using data for decision making, despite the gap between such leaders and laggards is widening.
Of course the maturity level varies from sector to sector and also among organizations in the same sector depending on the top management and company culture. The International Institute of Analytics (IIA) published a report with a ranking of analytics maturity scores across industries. For example, organizations in the Financial Services industry had scores in the range of 3.5 to 4.0. Whereas, those in the Insurance industry scored lower, in the 2.3 to 3.1 range.
To conclude, organizations that understand the concept of data science maturity and take steps to evaluate their capabilities can make a great start in their journey towards business value from data.
