The goal of reinventing a siloed organization into an integrated, collaborative entity is what’s kept Chris Kakkanatt, data science senior director at Pfizer, occupied — and awake at night — for more than a decade. The idea has had a lot of airtime for years, but with the proliferation of data and advances in machine learning, that goal becomes even more urgent for every organization.
During the first day of Transform 2020, in dialogue with Dataiku’s COO Kurt Muehmel, Kakkanatt shared how Pfizer has transformed 170 years of technical debt into collective intelligence across the organization. The approach involved breaking the task down into three primary areas.
“The first area we tried to really emphasize is how do we empower colleagues across the globe to master data sets,” Kakkanatt said. “There are different technologies out there, different skill sets — so how do we empower every one of our colleagues, no matter if they’re a clicker or a coder, to be able to leverage the technology?”
Enabling every function in an organization to access and action data, regardless of technical ability, requires removing barriers. Kakkanatt used the analogy of Google Maps. If you search for directions on any given destination, Google will highlight one route, but also serve up two or three alternate routes, giving the end user the ability to play with the results, depending on the situation.
“By creating interactive visualizations, you allow people to really interact with the models,” he explained. “Plug and Play methodology is what we’ve seen as a gamechanger in terms of people moving away from their own silos, and saying, hey, maybe I should explore different areas. We find that it really brings out the curiosity among among people.”
The second step in eradicating silos for Kakkanatt is transforming how business colleagues in different areas engage with one another. Pre-COVID, that involved actually bringing people together physically, and co-creating in real time, using what-if scenarios — and trying to answer those in the room in real time, pushing analytics to the point of decision-making.
“So in other words,” Kakkanatt said, “a data engineer is able to say, ‘Oh hey Matt, why don’t you take a look at my data flow and understand what I’ve done, and here are some of the metrics that you requested.’ But the actual business analyst is able to go in and understand how the metric was calculated and see it didn’t include a certain inclusion criteria.”
The third area, which truly underlies all of this, is leveraging AI and machine learning for speed and smarter decisions. However, Kakkanatt emphasizes that his team began by being very selective about which projects to apply machine learning to. Pfizer used a variety of business functions and business questions to understand how best to apply the technology across the organization in a collaborative way.
“We didn’t try to use machine learning AI for every single project,” Kakkanatt said, “but started testing, [using] different lighthouse projects to figure out, where’s the right fit for these types of initiatives. Don’t try to use machine learning and AI for every single project.”