With the data economy amplifying demand for all things digital, many organizations are examining their systems and wondering whether their data and technology assets are up to snuff. Can they scale to meet the challenges and the promise of AI?
Companies are looking for straightforward results: a tailored, Uber-like customer experience, an adaptive supply chain like Amazon’s, responsive products like Netflix’s.
Straightforward, yes; simple, no. Even if your organization has AI initiatives underway, the path to AI is long and winding. Unlike other scalable technologies such as cloud and analytics, AI requires a fresh look at every use case. Modernizing your organization’s assets to take advantage of AI means constant renewal.
Guidelines for AI Preparation
Our recent webinar explored the data disconnect that needs to be addressed when prepping your organization and systems to scale for AI. The experts who joined our panel came ready with first-person experiences and suggestions you can put to work. Here’s a recap:
Data modernization isn’t just about technology.
In addition to smart tech strategies, data modernization requires a full-scale corporate shift that’s already underway. New research from MIT underscores the extent of the changes taking place. In the past, organizations put data analytics to use for internal gains, such as beefing up business processes or boosting efficiencies, noted Barbara Wixom, principal research scientist for MIT Sloan Center for Information Systems Research. While internal projects still account for 51% of economic returns from data analytics assets, Wixom said market-facing activities represent 49% of the data-based revenues, according to her research. It’s a dramatic ascendance for data and one that ripples through every business function.
Start by creating a structured process.
Modernization begins with business value. The first step is to create a value chain map that clearly identifies opportunities that result in outcomes that impact the top and bottom lines. What’s the associated data that will help generate those outcomes?
The most important thing is to set your priorities, pointed out Alex Rosenthal, head of the enterprise data and insights office at Guardian Life Insurance of America. He believes organizations should prioritize the potential areas where data can elevate business processes, whether it’s increasing sales or reducing risk. Get that into a single list, he says, and establish your goals as you move to a customer-centric organization. What will help the bottom line? How do you improve CX?
Build a technology foundation based on what you’re monetizing.
If your organization is focused on improving business processes and driving efficiencies, then creating a data lake should be at the top of your IT to-do list, said Wixom. If your focus is CX and you’re creating features and experiences associated with your product, the focus should be on cloud and scale.
Cloud is an enabling component for modernization because of its scalability and flexibility: You can choose how to store, compute, distribute and scale. The level of elasticity and scalability changes, however, depending on what you use the cloud for. Leveraging it for infrastructure, for example, requires less elasticity and scalability than at the application layer. When planning a cloud migration strategy, the two most important variables are cost and optimizing business value.
Make data accessible.
Modernizing data is essential for realizing the promise of AI. If historical data isn’t stored in an accessible way, AI and machine learning efforts are doomed to fail, observed Anurag Sehgal, managing director for Credit Suisse, and head of data analytics, machine learning and AI at the international financial services company. Relevant questions include: How easy is it for data scientists to access all kinds of data that exists within the organization? How readily are they able to experiment and fail fast?
Get the skill sets right.
As Sehgal noted, when it comes to data modernization, a SWAT team is needed: People with business expertise partnering with those who understand the required sources of data. And given the multiplicity of devices, it’s vital to incorporate a design-thinking process to make data more accessible and meaningful.
On the tech side, several relevant skills come into play. Full-stack developers are key, and Python is the go-to technology for processing and modeling. Data visualization is also important. You’ll need to assess your organization’s true IP and determine your market differentiators vs. commodity skills and partner accordingly.
Whether you’re just getting your assets in gear or already have AI initiatives in production, modernizing your data assets is an evolving task. By continuously addressing data and systems needs, businesses can ensure a successful path to AI.