“We’re taking analytics from a planning perspective to a planning, execution and evolution perspective,” says David Kreutter, VP, US Commercial Operations at Pfizer Inc. As a result, analytics has become “much more operational than it’s been in the past.”
The ways that Pfizer Inc., the global pharmaceutical company, uses analytics is changing, in no small thanks to the pressure on physicians to prescribe generic rather than brandname drugs.
David Kreutter, VP, US Commercial Operations, is accountable for Pfizer’s U.S. commercial operations, including business analytics. He says that “in terms of customer analytics and commercial operating analytics, Pfizer U.S. has a strong legacy of management science in operations research support.” With a team of 40-50 economists, statisticians, operations research colleagues and scientists, Kreutter says the group historically has focused on promotional tactics: understanding the effectiveness of strategies used in the field, in conversations with physicians and other players who influence whether a Pfizer drug is prescribed or not.
Today, that’s a little different. Today, Kreutter says his team closely tracks how sales representatives present material and how presentations are received. It’s information, he says, that’s critical.
Kreutter spoke with David Kiron, executive editor at MIT Sloan Management Review, and Rebecca Shockley, the business analytics and optimization global lead for the IBM Institute for Business Value, about how the company is generating daily data reports, why precision is overvalued and what the coming generic version of the company’s popular Lipitor drug has meant in terms of attention to analytics.
Kiron: How has the strategic role of analytics changed at Pfizer in recent years?
It has changed significantly. Part of that change is in response to the cost constraints that we’re all facing, “all” meaning pharmaceutical companies.
There’s no great surprise here — a number of very, very successful drugs in the marketplace are disappearing this year and over the next couple of years, which is just astounding in terms of the number of dollars. From a healthcare perspective, it’s absolutely great for patients to have a generic version available of Lipitor [a Pfizer cholesterol medication that reduces the risk of heart attack]. From a business challenge perspective, though, taking billions of dollars out of a corporation in one fell swoop is a little bit daunting.
Part of our response is how do we reduce our operating expenses so that we can maintain our operating margins? The question of capital allocation becomes even more critical than it normally is in an environment of constrained resources. You don’t have the same operating buffer, if you will, that you had before.
What I’m seeing, from an organization perspective, is more of a focus on understanding what the data are telling us in order to use resources in the most efficient and effective way possible. Historically, I would describe our use of analytics at Pfizer as a kind of intellectual journey. People would have hypotheses or strategies that they would want to pursue through numbers. They would quantitatively analyze them, but for the most part, unless there was a glaring difference between the hypothesis and the analytics, people would pursue their strategies as long as they were compliant with our legal and regulatory requirements.
That’s pretty much going away. Because we’re at a point where we can’t ignore any data telling us about the effectiveness of our business strategies. The stakes are just too high, and the resources to allocate aren’t the same as they were before.
Kiron: How have you seen that change play out?
There’s certainly an uptick in focus and utilization. There have seen some clear demonstrations of how we can use analytics to link the goals of teams to actual execution at a field force level. For instance, we are able to understand the “why” of things a little better. When something is not working, is it the strategy that’s not working? Is it the execution? Both? We’re taking analytics from a planning perspective to a planning, execution and evolution perspective, so it becomes much more operational than it’s been in the past. I think that’s the key part.
Shockley: Tell us some more, if you can, about how you’re using analytics to understand your customers.
Well, we have data on almost every physician in the U.S. who prescribes Pfizer medications, and that data comes from IMS. What’s really changed over the past year or so is the amount of data we collect directly from our customers — meaning either the healthcare providers who do the prescribing themselves or the people in their offices who would either influence them or execute on their behalf, like nurse-practitioners and physician’s assistants.
As we’ve evolved from a paper-based interaction model to a digital-interaction model and a multi-channel model, we’re getting a huge amount of information from our interactions with our customers.
A lot of it is activity-based. When physicians visit our website, we know what they’re clicking on, we know what they’re clicking through to. We don’t have any greater data on how those clicks translate into prescription writing, but we’ve got more data from which to try to discern patterns, which we can use in a predictive way. That’s really what we’re trying to focus on now: can we detect patterns early-on, or at least much earlier than prescription writing, that will allow us to adapt more quickly to our customers’ needs as well as to the competitive environment?
Kiron: Lots of data, a lot more quickly.
Yeah. Here’s an example. We segment our customers much like any other industry would segment customers. Lipitor is Lipitor, so we don’t change the legal and regulatory approved messaging around that product, but we will change the presentation based on how physicians approach the care of their patients and how they absorb information. We tailor the presentation to be aligned with that customer segment’s needs and preferences.
When we were a paper-based organization, we had no way of knowing whether a sales representative actually showed the recommended content to the recommended segment. There was just no way of understanding that.
Now that we’re in a digital operating model, we have an opportunity to ensure greater compliance controls and detailing effectiveness. For example, we can ensure that representatives are using the approved and up-to-date materials and are interacting only with healthcare providers that belong to approved specialties for our products. We can also actually track, literally in real-time, the utilization of content and then the behavior. We can see if representatives are delivering the recommended content to the recommended segment. If our strategy is to deliver certain messages in a certain order, we can see if the message was delivered that way. For example, if we know that a certain segment of doctors in South Florida have a heavy proportion of elderly patients, they will often want to hear about drug-drug interactions first (since their patients are on many medications). We can track if we executed against that strategy, and we can track if that strategy had the impact, the literal prescribing behavior, that we anticipated. It’s a huge level of insight into the basic operating model, and really helps us to figure out, if we don’t have the impact we hoped for, if our strategy was right but the execution was flawed, or if the strategy fundamentally needs to be rethought.
Kiron: How do you get that information in real-time?
By real-time, I mean daily. I don’t mean literally as the representative is talking to a physician. Let me give you a little bit of context. When we get IMS data at a prescriber level, there’s about a 6-8 week lag between what literally happened, the physician writing the prescription, and us getting the data. Two months.
When representatives talk to physicians in person, those representatives are now using tablet PCs. As they click the screens with their styluses to illustrate points, those clicks are recorded. That’s how we’re able to see things like the order of presentations, the messages within a presentation that were presented, if the physician found it engaging. Representatives synchronize their tablets on a daily basis, and we get a data stream back to our data warehouse. Our customer data master now has all of that click-stream data for each representative and each doctor.
Just to put that in scale, in Pfizer’s Primary Care Business Unit, we have about 4,000 representatives. Each of them sees about seven or eight physicians a day, and details about two to three products in each of those calls. And they’re out there five days a week. We’re collecting a lot of data on a daily basis now.
Shockley: It’s interesting that presentations are now made on tablet PCs in the field. Do you feel like the needs of your customers have changed?
Yeah, they certainly have, and on a couple of levels. At one level, there’s the simple fact that pharma, as an industry, is no longer the sole source of information about our products. Physicians don’t need a sales representative to deliver a package insert for a product: they can go on the internet and get it themselves.
As well, their autonomy in determining their own medical practice has changed significantly. The demands of payers around generic utilization and cost containment have really started to reduce the degrees of freedom that our customers, the physicians, have.
But those payers are our customers, too. Whether you’re on the formulary for CVS Caremark or Walgreen’s or whomever very much influences the utilization of medication. And of course the patients themselves are very influential. They’ve got access to much of the same information that physicians have, and that dynamic, the patient-physician dynamic, has changed dramatically. Patients have become more empowered and more accountable for their own healthcare. The whole industry, the whole healthcare ecosystem, is in flux.
Shockley: What’s been the most difficult thing in getting to where you are today?
From a capabilities’ perspective, the hardest part has been the talent pool. Hiring the right people. The pharma industry in general isn’t state-of-the-art in the use of analytics, so when I think about talent and capabilities, I don’t look at what is Merck doing and how do I access that talent, I look at consumer packaged goods, financial services, telecom, and ask, how are they advancing their business? We haven’t even caught up to where they were five years ago. What are companies in those industries doing now? What talents have they accessed to drive that? How do I access that talent?
The challenge is contextual knowledge. Pharma is not telecom. It’s a different operating model. So how do we round out our staff people so they can apply their brain power and capability set in a business-contextual manner? We want to make sure that the value they’re bringing can be realized — we don’t want to wind up with theoretical answers that aren’t executable in our environment.
The challenge also is that I want to keep them engaged. If their primary motivation is to do cutting-edge research and analytics, and to interact with like-minded individuals, how do I make sure that I’m meeting their needs so that we can retain them and keep them productive?
Shockley: Are there particular skills that you think are critical to have within your analytics group that you have particular difficulty finding?
That’s a hard question for me to answer, because based on how we currently use analytics to drive insights, I think we have the capabilities we need. But that’s kind of self-limiting. What I don’t know is, how do we take it to the next level? How do we provide the next level of insights? I’m not talking about a marginal improvement. I’m talking about a frame shift. How do we provide next-generation insights to the organization?
Shockley: So talent is a challenge. What about cultural issues? What’s been hard culturally in getting to where you are today?
What’s been difficult is the temptation — and maybe it’s an appropriate temptation — to ask, in a cost-constrained environment, if we can really afford to spend this much time and money on data and analytics.
I actually think that the spending question is misplaced. The question isn’t how much money do we spend on data and analytics. It’s how much value are we getting from data and analytics. If the literal amount of value that you get exceeds the cost, then finance theory would say, “Yep. Do it. Keep doing it.” If it’s not, it’s not necessarily a question of do we cut it. It’s a question of what can we do to get more value out of it. But the reality is that there’s a pretty intense focus at the moment on cost.
The other challenge is how do you literally ascribe value to analytics. Invariably, someone will say, “I would have done that strategy anyway. Whether you analyze it or not, I would have done it.” And yeah. I don’t know: maybe yes, maybe no.
Demonstrating the value of analytics on an on-going basis is part of building the organizational support. I do think there are two schools of thought around this. One is the top-down and the other is bottom-up. I think they’re both valid, but I just think they require different approaches.
Shockley: Has Pfizer taken one approach or the other? A combination of both?
I would say it’s a combination of both. The reality is that all these strategies are executed bottom-up, because demonstrating value at the brand-team level is absolutely critical. I don’t think top-down would be possible because I fundamentally believe that if you mandate analytics or you mandate anything, at least in our organization, it’s the kiss of death.
Kiron: I understand, though, that the analytics operation at Pfizer is in the process of becoming an integrated service for the company. Can you talk about what that means and how that will work?
Currently, the analytics department is a shared service across the U.S. business. From an organizational perspective, analytics sits within Commercial Operations. It’s not a pay-by-the-drink system, but based on utilization patterns on an annual basis, our services are charged out to the four U.S. business units.
Now, all of that is about to change. The overall direction going forward is that analytics will be integrated, meaning that secondary market research, market analytics and management science will be integrated into a single analytical function. And it will be integrated on a global basis.
This gets really tricky because while I have an integrated analytical function in the U.S., it doesn’t really exist like that in some of our other markets. There, people who do this do it as part of a broader job. This will be a way to leverage the sophisticated capabilities that we have in the U.S. through a consolidated delivery center to the less well-endowed organizations around the globe. Can we deliver better insights more cheaply than we’re currently delivering them? That was the rationale for the consolidation of analytics, but analytics was only one of a number of shared services that were consolidated on a global basis.
Shockley: You’ve been in analytics for a long time. Are there any other observations you’d like to share?
Well, I think the focus on analytics has really been one on sophistication and precision. I want to pick up on the precision part because I think there’s a fundamental belief — and I think it’s something that’s untrue — that a precise answer gives you a better decision.
We obviously can analyze things much more precisely than we can execute them because it requires humans to execute these strategies, and each human applies some personal judgment. The beautifully precise answer gets diluted as the strategy gets executed. As I mentioned before, we can now at least track that that process.
But I also think that some of that focus on precision is misguided. Because by definition, we’re trying to analyze the future, not the past. As I mentioned at the beginning of our conversation, analytics is not a descriptive exercise; it’s a predictive exercise. Therefore, by definition, there’s uncertainty: We don’t know everything about the future. Maybe some of our focus should be on helping the organization understand the bounds of uncertainty and the actions we can take within those bounds of uncertainty.