AI is changing the way we do business. In my last post, I sat down with Pratap Khedkar to discuss AI’s role across the industry. Now that we’ve been introduced to AI and pharma, we can get more specific.
All aspects of the pharmaceutical organization are striving for increased efficiency and effectiveness by leveraging AI. As we pointed out in our previous post, there’s been more excitement in R&D because of how much data is available from clinical trials and consumer activity. But because of the focus on R&D, many commercial uses of AI have been underserved. One of those areas is product launch.
To delve into AI and product launches, I spoke with my colleague, Dharmendra Sahay, who leads the integrated analytics practice at ZS and has extensive experience optimizing product launches for clients across therapeutic categories.
A: What used to happen is that analytics was executed by analysts offline. They come up with recommendations and you then ask, “How do I take this recommendation and implement it”? In the context of a launch, the analyst would produce a dashboard, for example. And the dashboard gives you some KPIs. But then you can have a lot of questions on why something is happening or how two or more factors relate. The analyst then performs a one-off analysis and sends it to the user.
Today, all of this is getting automated and integrated. With all the primary and secondary data in one place, covering both patients and physicians, and sophisticated machine learning algorithms leveraging this data, one can discover not just the fact that one region is doing better than the other, but also get a little bit more nuanced.
What is happening across regions? What is driving performance? Is it access? Is it promotion? Is it messaging? So, you learn not only how you’re doing but you also get much more actionable insights about where you should be doing something differently and what should be done. And one integrated approach covers all of this.
The equation has changed from a multi-phased, manual process to an algorithm and an app that can give me the information I need now so I can be focused more on taking the action as opposed to analysis. And doing it this way, with learning algorithms getting smarter over time, you can improve in a systemic manner. As I go further along in my launch, my solutions start to get more and more sophisticated because of all the learning of the prior cycle.
Q: With this approach, I gather that you can bring together several varied constituents with a shared understanding of the current state. Can you expand on how this might work in practice?
A: Yes, that’s one of the greatest benefits of this new way of thinking. Where it’s been effective is if you get cross-functional people in one room, and then you have a system like this, and they're all interacting with the system. Then they can each, at the end of the meeting, have very specific action items and plan how they will coordinate those actions. So, in the few cases where this has happened, there are lots of examples of how quickly they were able to decide. Usually with a lot of data you get analysis paralysis. If something unexpected happens, then it’s a few weeks, or a few months before you can act. But with this approach, you bring everyone to one meeting and you decide your plan of action. In one case, for example, right in that meeting, the national sales leader was able to call the district manager and give them specific feedback on what they needed to do differently with samples from that morning onward. And patient services were able to adjust their copay strategy in that geography. So, we’re looking at two key benefits here: speed of surfacing key insights and a focus on time-critical action steps by supporting a cross-functional, common view during a launch.
Q: What does this mean for the future of product launches? How will pharma companies be leveraging these services?
A: For launches, companies have historically established these capabilities from scratch. Because every launch has a new team and this new team is building a whole new capability, that includes the analytical function. But a lot of data is common across launches and therefore the algorithms can be common. And when you re-use them, they can learn how to manage a launch better. So, there could be a couple of scenarios in the future: First, for some clients who are either small, medium or just don't want to build a full-scale capability, they may be able to subscribe to an insider launch analytics service, so they can keep their focus on acting, rather than building the solution.
Second, for the clients who have solved for this capability already and have all their data in place, they may be only buying underlying algorithms and front-end applications. Or they may prefer to develop the application for optimal experience within their organization, but then license the algorithms. They’ll be able to avoid reinventing the wheel by leveraging investments they’ve already made. But at the same time, the system in place will get smarter over time as it learns from multiple launches.