Experts at the International Institute for Analytics gathered to talk about their analytics priorities for 2017. Their advice: Formalize the analytics function and push the analytics envelope.
This is the second year the IIA has crafted a list of five analytics priorities in an effort to provide its constituents with a strategy for the year ahead. Last year, the analytics priorities focused on aligning the analytics strategy to the business strategy. This year, the advice focuses on defining roles more clearly and striking the right balance between experimentation and deployment.
Dan Magestro, a former analytics practitioner who worked at Cardinal Health, JPMorgan Chase & Co. and Nationwide Insurance, and who now heads up research efforts at the IIA, led the discussion. He was joined by Robert Morison, an analytics consultant and researcher at the IIA, and Tom Davenport, IIA co-founder and the president's distinguished professor of information technology and management at Babson College.
The IIA's five analytics priorities are as follows:
Priority No. 1
Embrace the "new analytics era" of enterprise AI.
The current enthusiasm for artificial intelligence (AI) is unprecedented, according to Magestro. The technology is more accessible than ever, and the practice of AI in the enterprise isn't a stretch for companies with established analytics programs. He called it a new analytics era for three reasons:
- Big data was the beginning. CIOs whose IT departments are already well-versed in big data analytics have laid the foundation for AI. "Unlocking the full potential of AI requires big data for training AI models, immense data-processing capabilities and advanced statistical methods," Magestro said.
- It will take a village. AI programs will require new internal partnerships and "top-level strategy" if they're going to stick. A computer science function, for example, might need to be integrated into the business, which isn't a traditional analytics characteristic.
- Automation will leave a mark. Software robots are increasingly performing tasks traditionally performed by humans, and analytics practitioners and data scientists won't be immune to the trend. "I think it's positioned to explode," Magestro said. "Organizations need to embrace it."
Davenport added that automating analytical work can be straightforward and will benefit the organization. "Some machine learning algorithms are really just automated regression analysis," he said. "[Automation] may be threatening to some analytical practitioners who fear the loss of their job; I think it's a huge productivity aid [to them]."
Priority No. 2
Fully explore the cloud for analytics development and production.
Magestro said he expects companies to continue turning to the cloud for both analytics development and production, where companies can not only build prototypes at a faster pace, but scale those prototypes faster to better serve the enterprise.
CIOs and their IT departments will have to work out the "considerations and architectural issues that come with moving analytics work to the cloud," he said, which will include security concerns.
Priority No. 3
Formalize the analytics function.
The analytics function in companies is often ill-defined, which has "wide-reaching implications," according to Magestro. Formalizing it by clearly defining a data analyst's role can affect retention rates and help analytics talent grow. Plus, as centralized analytics teams decentralize -- a trend the IIA included in its predictions list for 2017 -- clearly defined roles will make it easier to move talent around.
But formalizing the function will also affect another part of the company -- the business. "People are calling themselves data scientists who were previously analysts, and business people have no clue as to the different types of analysts," Davenport said. "It's high time for clarity."
The biggest hurdle may be working with the HR department to formalize the roles and responsibilities, Davenport said. "But I think it's critical to not only do it, but to get it embedded into the formal job classification system," he said.
Priority No. 4
Create a better balance between innovation and production.
Magestro said finding a good balance between experimentation and analytics work that reaches the production phase is the most challenging analytics priority. Exploratory data science projects that are never deployed, while important, don't provide a hard return on investment.
Plus, "it's a retention issue," Magestro said. "There are ways for talent to be leveraged more effectively when their work is seen as both innovative and there being a production component to it."
Attaining the right balance requires focus on three areas: building a platform or architecture that supports both experimentation and production; organizing analytics talent and defining the roles it will play; and defining how an analytics team will spend its time. Magestro said the last piece is the most important to strike the right balance because it's a way to prescribe when to innovate and when to productize. "The balance can only pay dividends," he said.
Priority No. 5
Focus on functional analytics.
Context matters, Magestro said. As the line of business realizes the power of analytics, it will push to have analytics and analytics talent work more closely with decision-makers. Doing so will "ultimately provide stronger context" when building out the analytics, leading to stronger outcomes and better decisions, he said.
"One thing that will drive this is more analysts embedded in business units," Morison said. He added that teams may be able to provide analytics as a service to the business by leveraging reusable components, such as APIs, to speed up model development.
Five analytics predictions for 2017
IT resolutions for 2017
A CTO's hope for the year ahead