Many organizations that strive to become data-driven ultimately struggle to get value from their data. Seemingly promising analytics proofs of concept fail to scale in production, technology platforms aren’t always fully mature, and driving real impact from data often requires fundamental changes to the way people work across a range of disciplines and functions.
For GlaxoSmithKline (GSK), a data strategy grounded in near-term priorities and value creation has proved key in avoiding such issues, enabling the pharmaceutical titan to establish the processes and technical foundation necessary to embrace bolder transformational moves aimed at competitive advantage. The strategy, launched in 2018 and dubbed Value Strikes, began bearing fruit in 2019 with a series of advanced analytics use cases, earning GSK a CIO 100 Award in IT Excellence.
“The Value Strikes program was a way to accelerate our enterprise data and analytics ambition,” says Jen Baxter, senior vice president of tech strategy and performance at GSK. “Each use case, termed ‘value strike,’ leveraged our existing data in order to achieve our strategic priorities while delivering significant near-term P&L or cash value. In parallel, these experiences helped build up our people and technology capabilities across the organization.”
Baxter says the program successfully demonstrated how GSK could use a structured approach to defining opportunities for AI, incubating solutions, and scaling them effectively.
“We are now scaling a few select cases to embedding and systematizing data and analytics across our operational activities,” Baxter says. “This is easier said than done, but we are making great progress and it’s awesome seeing teams and individuals learning.”
A practical approach to data transformation
One of the company’s most successful ‘value strikes’ involved inventory. GSK’s supply chain analytics team deployed a new set of digital and analytics tools focused on inventory reduction opportunities across the company’s supply chain. The new suite of tools included a digital value stream map, safety stock optimizer, inventory corridor report, and planning cockpit.
“The pharmaceutical supply chain — from raw ingredients to customer — is incredibly complex,” says Shankar Jegasothy, director of supply chain analytics at GSK. “We wanted to use our data to create better visibility of our end-to-end supply chain, and then use predictive and prescriptive analytics to guide decisions around inventory and planning. The program aimed to deliver value through successive waves of increasing complexity, from policy adherence to full-scale supply chain design optimization.”
GSK scored further analytics success in payables, where it designed new analytics tools aimed at improving compliance of invoices and purchase orders to agreed terms and at renegotiating contracts to align terms to best practice in local markets and industries.
The company completed proofs of concepts and incubation of both the inventory and payables value strikes within months and have begun scaling them across the business.
“Our approach was to solve some of our most pressing business challenges leveraging data and analytics, not to develop algorithms and technology in isolation,” Baxter says. “We selected each value strike by working with our executive team to identify key business questions that could potentially be answered through advanced analytics.”
Building the foundation
To capture and extract insight from previously siloed data, GSK data scientists developed two platforms: a data provisioning platform that integrates all enterprise data into a single system and continuously feeds algorithms with the latest data in real time, and a visualization platform for each of the use cases to support decisions.
Data scientists developed both platforms with input from business and end users.
“A significant portion of the program effort was spent on capability build,” Baxter says. “New business processes had to be designed or existing ones adjusted to integrate the new ways of working. The tools were designed not as one-off solutions but as continuously developing products that would continue to meet changing business context and carry on delivering value into the future.”
Each value strike project used an agile, sprint-based approach starting with a targeted minimum viable product (MVP). The idea was to prove value at a small scale, gather feedback, and use that to rapidly accelerate the speed of development and deployment.
“It took two weeks to build our first digital value stream map, which was for one stock keeping unit. We then added the rest of the brand SKUs in four days, and finally SKUs for an entirely different brand in only one day,” Jegasothy says.
Putting together a cross-functional team of business experts, data scientists, and technologists proved key to the platform’s success.
“This representation of skill sets means we can problem solve and gather feedback rapidly,” Jegasothy says. “For instance, we don’t have to wait for perfect data to get value — by combining advanced analytics with data expertise, we can squeeze insights even out of sparse data.”
Jegasothy says the Value Strikes program has helped GSK identify a large array of value opportunities within its first year, including overstock prevention, stock policy adjustments, the ability to highlight and address areas of high lead time duration and volatility, and so on.
“The full value potential has been validated with the supply chain organization,” Jegasothy says. “Beyond the financial impact, planners and supply chain managers are newly empowered with data-driven insights and are seeing the art of the possible.”
Data-driven dos and don’ts
Baxter offers five tips for IT or analytics leaders looking to help their organizations get more value from data:
Business value is king. Data and analytics are not the objective. They merely serve as enablers of business priorities. “Make sure you can clearly articulate and track your value delivery, not just your activities,” Baxter says.
Bring the organization on board. Invest in building the capabilities of your team and users and ensure all new processes are fully documented. “We like to say that about 10 percent of the effort is in the algorithms and about 20 percent in the technology — the remaining 70 percent is in behavior change,” Baxter says.
Build trust in the tools. Full transparency on data and analytics is key in building trust in new tools, Baxter says. Dashboards that enable users to explore data in its most granular form, and to explore how that links to high-level insight, have had the most traction within GSK.
Encourage (and manage) development requests. A growing number of feature requests do demonstrate that tools are being used in the organization, but they can quickly overwhelm developers. “These requests should be logged and prioritized by business value and effort so that the program can continue to pay for itself,” Baxter says.
Remember you’re building capabilities, not tools. “These are not static analyses,” Baxter says. “They are constantly evolving, which requires the right people and skillsets to be in place to continually grow and manage.”