Methodologies All the Way Down – A Structured Approach to Data Products
Written by Drew Lazzara
Director, Analytics & Insights, Global Operations
“We have a dozen data products in production right now, and another half-dozen that will be in production in the next three months.” For Manish Motiramani, director of analytics and insights for global operations at Medtronic, that growing suite of offerings is the validation of a disciplined approach to data strategy development. “I think the biggest success factor – and there have been many – is the combination of a very strong team and robust methodologies and frameworks to guide them.”
Increasingly, data leaders are essential to sharpening business strategy, and the results are dramatic. Gartner’s recent survey of chief data officers found that organizations which utilize data leadership in strategy development are 110% more likely to demonstrate verifiable business value from their data and analytics initiatives and 90% more likely to have higher team effectiveness among business users that consume analytics.
Medtronic, one of the world’s largest medical device companies, embraces this potential. Motiramani and his team are a critical part of informing and driving the overall business strategy, and their data products are the result of an interlocking system of methodologies designed to bring simplicity, performance and innovation to Medtronic’s operations.
Frameworks on Frameworks
“The old Apple campaign was, ‘There’s an app for that.’ Well, we say that there’s a methodology for that.” Motiramani laughs as he says this, but structure and mechanisms for prioritization were essential to a workforce hungry for data insight but lacking in technical know-how. Before his team got started, “there were individuals who had some data, and had built visualizations for senior leadership. But these were not data-focused teams, and they did not have the analytics experience.”
Harnessing this appetite for data requires robust, enterprise-wide structures for alignment. Creating these structures is the real work of strategy-building. At Medtronic, the data roadmap for Medtronic’s operations starts with four foundational pillars:
- Trusted and Real-time Operational Data Foundation with Master Data: Data must be in one place and collected in a standardized way. “Like any multi-billion-dollar organization, we have 30, 40, sometimes even 50 systems that are used for similar functions. We have to consolidate the data from all these systems, because without that full view, you’ll never get to useful analytics that enable decisions and actions,” says Motiramani.
- Democratize Data for All: Data must get to the right people at the right time to enable informed decision-making.
- Leading Indicators That Drive Outcomes: Identify data trends that have the biggest influence on future outcomes. As Motiramani says, “It’s great to see what my spend was last month, but wouldn’t it be even better to see what my spend will be next month? Some of these predictions will be achieved through machine learning, but you can still surface some leading indicators before you productionize your ML projects.”
- World-Class Analytics Organization: This means developing the best technical team to build analytics products, but also having the best user group to interpret information and get the most out of the products. “Every user should be able to interpret our key performance indicators – what does ‘x’ mean, and what can I do with it? We’ve built some insightful analytics, but we don’t just throw it over the fence to our users. We have a very effective data-literacy program and regular office hours. We solicit lots of feedback to make their jobs and our analytics products better.”
These pillars orient Motiramani and his team around core values; they’ve also created the appetite for more data products across the business. Managing this demand requires additional structure. “We have to remind our analysts that just because a tool can do it doesn’t mean we have to. We know that seeking cool things to do rather than identifying business problems to solve is putting the cart before the horse.” To manage this mindset and prioritize the right new products, the analytics team developed the OAI methodology:
- Outcomes: What is the business looking to achieve? Motiramani says, “This has nothing to do with the data or the application; we talk to users about the business problems they face and where they think tasks aren’t optimized.”
- Actions: Once desired outcomes are defined, what are the right actions to get there?
- Insights: What business insights do you need from the data to decisively take those actions?
“So, the OAI methodology leads us directly to the insights we need,” says Motiramani.
Supporting the User Experience
Even with strong methodologies, data products remain ineffective if they aren’t utilized consistently across the business. This is something Motiramani knows from experience. “When our team was formed, we started evaluating existing dashboards. One problem we saw was over-engineering. Some dashboards had more than 15 visuals. And when you have something that complicated, you take a hit on performance.”
To tackle this challenge, Motiramani took inspiration from the original dashboard – the automobile. “In your car, you have only four gauges that give you the most important indicators. You don’t need a gauge for every component under the hood.” And just like a car, these key indicators show you the health of your business and point you toward solutions if there’s a problem. “If a yellow light comes on, you take your car to the garage to see what’s wrong. We took the same approach.” Motiramani calls it the “three, four, five” methodology:
- Three clicks to action: Every dashboard should go from the highest-order metrics to the most granular details within three clicks. There’s even a sub-methodology – MAD. The first click takes the user to “Monitor metrics,” which illustrate only the most critical KPIs. Anything that needs attention is listed in yellow or red depending on the severity. Users can trigger the second click, “Analyze anomalies.” This is where data is sliced to understand the context of the problem. Once a user understands the context, the final click is “Drill into detail.”
- Four Seconds per Click: Any user interaction with a dashboard – the initial render or a slice or toggle deeper into detail – should happen in fewer than four seconds. This ensures speed and performance.
- Five Seconds to Comprehend: A user should be able to understand what they are seeing in fewer than five seconds. This places a premium on simplicity when products and dashboards are designed. No more over-engineering.
The “Explore” Space
Motiramani’s system of frameworks triage business problems and create guidelines for product usability. But there’s still a cost associated with new products, and this places a premium on getting products right the first time.
“Our IT team does our data curation across our multiple data sources,” says Motiramani, “and they came up with a concept called the ‘explore’ space. IT has carved out a space for our team on the enterprise database server,” explains Motiramani. “We ID key systems, do a manual pull of the data and load it into the explore space.”
Rather than wait for months as data is curated, the explore space allows Motiramani to start prototyping visualizations immediately. “We get user feedback on those prototypes from manual pulls and define exactly what the data product should include. We then work with IT to curate that data from the multiple systems. Users can start to see a draft before the data is even curated, and we are able to identify exactly what fields we need to pull from which systems. It allows us to be much more agile and it reduces rework.”
Motiramani is careful to point out that this is still very much a journey. “We’ve come a long way, but I’ll always keep moving our goals a little further north,” he explains. Steadily expanding what’s possible is the direct result of Medtronic’s disciplined approach to data. Each framework builds on a previous one, allowing the power of data to span from business problem to user to innovative solution quickly. Analytics is, after all, a journey, not a destination.
Special thanks to Manish Motiramani and Medtronic.
by CDOs, for CDOs
Join the conversation with peers in your local CDO community.
PODCAST | AUGUST 3, 2021
PODCAST | AUGUST 3, 2021
The Next Big Question, Episode 19
CIO Piyush Chowhan of Lulu International discusses turning his passion for customer experience into value for the enterprise. Piyush shares how to break down the silos between technology and digital business strategy and how to optimize organizational processes and workforce skills to meet the customers’ needs.
BLOG | JULY 26, 2021
BLOG | JULY 26, 2021
Who’s Answering the Next Big Question? VP of Data Amit Sethi of Momentive
On The Next Big Question podcast, we talk to executives about a big, timely business question. Here are highlights from our conversation with data leader Amit Sethi of Momentive, who discusses how technology informed data practices and increased the business appetite for insights, as well as how data leaders can synchronize their capabilities with demand.
BLOG | JULY 19, 2021
BLOG | JULY 19, 2021
C-Suite Most Concerned About Employee Engagement -- Above Revenue & Growth
In a recent survey of nearly 1,000 C-level executives in Evanta communities, 70% said their top business concern was maintaining culture and employee engagement -- followed closely by employee retention and recruitment. Read what else they reported about returning to normal at their organizations.