Dir, Strategy, Partnering and New Capabilities Novartis Pharmaceuticals
DHC: In recent years, a lot of the conversation about artificial intelligence and machine learning has been about how it will impact the pharmaceutical medical device industry specific to research and drug discovery – using an algorithm to go through a lot of data for drug screening. A key question for the marketing teams ... how does AI apply to the commercial side of the house? Does it apply to the marketing?
Will Jones: The simple answer is yes ... because when you think about the entire healthcare system, more than likely, most of the time, the pharma piece of the healthcare system – which is just one piece of the entire healthcare ecosystem – we're the ones that are typically lagging behind when it comes to using technology to improve the experience of whatever output we're trying to help the marketplace achieve or gain or have. Three-plus years ago when Novartis was first interested in trying to understand this better, we piloted some use cases, and the thing that we initially tried to understand is ... could this new use of technology and way of looking at data, could it help us solve the business. The answer is yes. The first thing you had to get right was – can we first bring the right people in the room? We learned that when we a brought cross-function team together in one place, the automatic answer was "wow," AI/ML can be extraordinarily useful for the business because it brings people together who otherwise sometimes talk over each other or don't even talk to each other at all. Another key thing we learned out of it was the ability to answer significant key business questions that typically have gone on for several business cycles without being able to be effectively answered. We now have a practical way to answer some of those key business questions, because no matter how many smart people we have in a company or on a team, no one can understand all the data as fast as they need to understand it.
DHC: What are some of the other challenges within pharma when it comes to embracing AI and machine learning? It seems like a lot of the traditional methods of targeting seem to be working, right? What is the motivation or incentive to change?
Will Jones: In many ways, you are absolutely right because many of the things that pharma companies are doing today are actually good things. Pharma companies have found really extraordinary ways to use what they have fairly well. Some of the current data sources that have always been in play are still good data sources, and they probably still solve anywhere from 60 to 80 percent of the questions we have in terms of better targeting, and better segmentation, and building better responses and messaging that we can use in promotional ways. Normal data still gets us a long way. Where I think AI/ML really comes into play and where it really kind of just takes off and it leaves some of the practical datasets in its wake is the fact that AI/ML really allows you to bring a lot of data sources into one platform and start to look at data and get more insights and questions from what you thought were answers. You can bring more functions and teams together in a way that traditional data just can't do. For example, in one AI/ML use case, we brought medical, marketing, finance, operations, market access, and HEOR/Outcomes in order to solve for key business questions for multiple brands. Bringing all these functions together allowed us to speak a common language; between all these teams coming together to solve the key business questions was the fact that AI was [is] that common currency or language that we were [are] all using to kind of understand things we never could before. So when we start to think about predictive triggers, biomarkers or lab tests that were very significant, these are significant not just for understanding the marketing plan or understanding first-line or second-line positioning. They're also very important to understand if you are on the clinical/outcomes side because now some of the analysis done with AI//ML was allowing the business to find efficiencies, solve for deeper questions than before, and show us there were new ways to operationalize outputs between marketing and the field force. Some of that same data analyzed by AI/ML was also confirming known answers to some business questions and, in some cases, answers were more robust and rich for every business function that was able to touch it. Of equal importance was the ability to have rich discussions with a lot of cross-functional teams; that really wasn't happening all the time in a consistent way. That was a good byproduct of AI/ML.
DHC: You mentioned earlier the explosion of data in oncology – exponential increases in data. Given we've seen the number of data sources increase – and the volume within those data sets increase rapidly – which data sources are important and why?
Will Jones: So I think how you get your head around this is that there's still some limitations on the data that's collected, just because it's pharma; and so obviously, one of the biggest challenges at times of appreciating AI/ML and the data that it brings into the business is the fact that we still have a lot of data/privacy rules and standards that we have to meet so that we're doing things correctly and safely, along with a standard of integrity and transparency the marketplace truly respects and appreciates. What AI/ML has allowed us to do beyond traditional data is to bring considerably more data together to be assessed – whether it's wearable technologies, claims data, lab data, scans, EHR data, is you can look at a lot of data together and derive understanding from every particular touch point a patient has had with a physician(s). All those different actual points within that patient journey become important because it allows us to appreciate what the patient is experiencing – pre-treatment/diagnosis to current-treatment/diagnosis to post treatment/diagnosis. How can that pharma company build the resources that support that patient journey? In pharma, we are striving to create resources, tools, and information that allow the patient to have similar experiences of the online functionality they experience in the other parts of their daily life. The patient/consumer perceives phenomenal ease of use in everything else they do, except for what it seems like in healthcare – we aim to recreate that same experience.
DHC: Any advice for someone within pharma seeking to get a better understanding of how AI is changing the business and assessing the impact it will have on the industry? Any advice from your experience and what you've seen that works?
Will Jones: Yes, I think this might be the most important question. It seems like the pharma business – outside of all the other major sectors – has been slower to appreciate some of the dynamic that's coming toward us that's happening right now with AI/ML. I think we have to be even more open minded to being curious about innovation, especially AI/ML. Colleagues and peers that are open minded to this [AI/ML] seem to be creating a competitive advantage for learning and applying new ways of doing the business. There is clear evidence that other sectors of business have embraced AI/ML and are starting to see changes to their business and the outcomes that AI/ML is helping to create.Our sector is just beginning this journey, and we should expect exciting achievements as we are beginning our journey leveraging AI/ML. I would encourage anybody to be open to this change and to be able to start to appreciate working with companies that are trying to integrate AI/ML, and start to really build another competency to be a part of this change, because this change isn't going anywhere;it’s here to stay. I think most everyone would agree, whether it's a top-tier oncology company or a top-tier diabetes company, if something can help your teams address the business needs today more efficiently and then scale up what you're trying to do in a competitive and economical way, and drive the business while delivering an exceptional patient experience – I think companies are open to that. That's why I think AI/ML is here to stay. You are seeing many companies with a chief digital officer as a function, or they are hiring for a chief data officer or VP/head of strategic data within organizations. How a company assesses, integrates, governs and strategically commercializes data moving forward is about building a new competitive advantage in the healthcare ecosystem. Governing, integrating, and scaling data across marketing and medical and clinical in efficient ways – this has not been done before, but this is now the future reality of pharma and the healthcare ecosystem. These are all new things that are happening within the last 12 to 24 months, and now pharma companies are trying to get them right.
I think the thing that is driving executive leadership to really embrace AI/ML are very practical things about data – data integration, storage, and governance. The fact that there's so much data coming into organizations and it’s challenging to harness the potential of all of that data, except through some of these new and innovative ways that AI/ML can help harness. In the end, executives know that this is a significant undertaking. It's extremely expensive, and it's not a one-time cost. Human capital, executive vision, and leadership have to be willing to sponsor this long term.
Dir, Digital Innovation and Corporate Program Management Takeda
DHC: A lot of the conversation about AI and big data and machine learning has centered around drug discovery, logistics, and business processes. Does AI and machine learning apply to the commercial side of the business?
Dan Gandor: It does, although I believe it depends on how exactly you’re defining AI. If you’re talking about it in terms of pure analytics or analytical horsepower, then there are use cases in marketing and commercial like finding better targets, finding new targets, targeting, and segmentation. There are automation aspects as well: relationship marketing automation, platforms like Veeva Suggestions, and the engines behind that. If you’re defining AI as natural language processing or natural language generation like chatbots, there’s activity in that area – both on the patient and prescriber side. There are also internal use cases where you have some sales forces starting to have reporting accessible via voice so they can pre-call plan on the drive in the car.
DHC: What are the challenges in making AI work within a large pharmaceutical company, overall?
Dan Gandor: I think, like many sexy, digital things, it’s important to not just do technology X for technology X’s sake, whatever that may be – AI, chatbots, voice, IoT, you name it. Rather, one should be focus on the underlying strategy: e.g., what’s the business challenge; how do we solve it? Therefore, it becomes applying AI or whatever the technology (which may be simpler than true AI) to specifically solve that business challenge.
DHC: What are some of these other data sources, EMR being one, that you think will become more important in the coming years?
Dan Gandor: I think you’re right in saying that EMR and wearable data sources are becoming more prevalent and more useful. Those are situations where the data is too large to be analyzed by traditional techniques and approaches. Although, frankly, I’m still seeing that the industry sometimes still struggles with analyzing and integrating all the commercial data sources we know and want to use today – let alone what’s one the horizon. One doesn’t necessarily need AI to do that, it’s just the blocking and tackling of omnichannel marketing. Step one is get the fundamentals right, then one can point an eye towards being ready to take in massive data sets like EMRs and wearables and things like that, thus opening the door towards new advanced analytic techniques.
DHC: If we think about customer service and AI – the intersection of the two – how is AI relevant to customer service, to the customer experience? Can we predict customer needs before they even state them?
Dan Gandor: Do you really need artificial intelligence to get to that answer? That’s where I’d propose it’s debatable. If you only have four things to tactically put in front of a customer (be it HCP or patient), do you need a whole complex engine behind the scenes to figure out what’s best? Do you need true AI for this, or is it more about a smart algorithm to say, “here’s who should get what, why, when, where, and how,” and then to make sure you trigger that knowledge to the right internal stakeholders (e.g., sales reps)? Or maybe you just transform the website to show the right message at the right time? Again, do you need AI to do that? Probably not. This is back to one’s core definition of AI. Is it true self-learning artificial intelligence? Is it a smart algorithm driven by modular personalized content? We do need to optimize our touchpoints in terms of what channels to use, the timing of the channels, the messages in those channels. I think there’s an opportunity for automation, to help make that more effective and more real-time. Whether AI is actually used for that automation is debatable.
Previous chapters of this e-book reviewed the definition of AI (and machine learning), the landscape of AI companies, the macro trends driving AI, and the use of analytics to automate and empower modern pharma marketers. This current chapter will take a deep dive into the following questions:
These questions and more will be explored in this chapter with results from a new survey, accompanied by additional insight from interviews with pharmaceutical thought leaders.
The Digital Health Coalition (DHC) surveyed a group of 24 pharmaceutical executives across marketing roles in May 2018 and asked them about 13 different technology trends impacting their businesses (and brands). The survey revealed that pharma executives were most concerned with tech areas of focus that have been around for the past decade.
Where does AI fit? Figure 1 illustrates how pharma executives rank AI in 2018 and 2019 on a five-point scale (from low to high importance).
What do these data tell us about AI in the short term? Respondents ranked its (mean) level of importance as 3.29 out of 5 in 2018, and 3.71 for 2019. For context, mobile for 2018 was rated 4.5, increasing to 4.6 for 2019, and social at 4.0 for 2018 and 4.1 for 2019. On the surface, the importance of AI is trending only slightly upward, but noticeably with a larger relative increase than other leading trends.
However, the averages hide some of the trends underway. While only 25% of pharma executives ranked AI as a 5 (on a five-point scale) in 2018, fully 38% ranked AI as a 5 in 2019. In other words, at the far end of the scale – “very important” – we see a significant jump in the segment reporting the greatest importance within the next 12 months. At the opposite end of the scale, only 8% of pharma executives rank AI as having very low importance (1 out of 5) for 2019 – down from 21% in 2018. There are significant shifts away from AI being “not an issue today” to “becoming very important” in 2019.
It is critical for brands seeking to plan ahead for market moves to understand and plan their strategy for where the market is headed – as opposed to placing all of their emphasis on established technology such as mobile and social.
Building on the above research, the Digital Health Coalition surveyed pharma executives in July 2018 to better understand the current state of AI adoption, use, and value realized. Companies participating in the DHC research included Pfizer, AstraZeneca, Lilly, Takeda, Boehringer Ingelheim, Sanofi, Genentech, Roche, Biogen, and GSK, as well as some emerging pharma and biotech firms. The study focused on gaining a better understanding of how companies are embracing AI, and where they are capturing benefits from AI projects.
As a baseline, we asked pharma executives about the level of knowledge their company has about AI, as well as their personal level of knowledge. Executives rated their personal knowledge as greater than that of their organization. For some, that may be because they deal with AI in their current roles. For others, it may be a case of illusory superiority, the cognitive bias where individuals often overestimate their abilities compared with those of others.
None of the respondents gave their organization an “A” grade (using a scale from “A” to “F”) for corporate knowledge of AI. Some 30% graded their company as a “B,” and another 26% gave their company a “C.” The remaining respondents – just under half of the total at 44% – chose a “D,” “F,” or “Don’t Know” to represent their company’s knowledge of AI.
When asked to rate their personal knowledge, numbers perk up a bit (as mentioned). Just under 20% give themselves an “A”; another 41% rate themselves a “B.” Only 30% report a “C,” 11% a “D” – and no one was willing to rate their personal knowledge as an “F.”
It’s notable that only 30% of executives believe their company is a “B” or better when it comes to AI knowledge – representing an opportunity for continued education and awareness.
The gap between knowledge and action is often large. However, our survey revealed that a healthy number of pharma companies are already making moves to use AI in their marketing and commercial strategies. Twenty-six percent report AI is already being used for marketing purposes, and another 30% report plans to use it.
Next, we dove into the priority our respondents were giving AI for marketing and customer engagement in 2018 and 2019. While just under half reported they were currently using or plan to use AI for marketing soon, only 18% reported a high priority for AI in 2018 – with “extremely high” at 11% and “very high” at 7%. However, the story – and relative priority of AI – shifts dramatically for 2019. Fifty-one percent of respondents reported a “somewhat,” “very high,” or “high priority” for AI in 2018, but that same number jumped to 67% in 2019, with the spike happening in the “very high” or “high priority” categories. At the other end of the scale, while 22% reported the priority is “not at all high” in 2018, that number was cut in half with only 11% reporting the same level for 2019.
Two key questions often debated by large companies are related to how any new technology – like AI – is operationalized:
Just under half (41%) of pharma executives surveyed report that their companies do not currently have a team or group dedicated to AI strategy and implementation. However, 30% do – and another 11% report having a team focused on AI strategy (if not implementation). To put this in context with overall industry trends, a recent study by Boston Consulting Group (BCG) with 3,000 global executives (across all industries) in 2017 found that while 60% of respondents say that a strategy for AI is urgent, only half of those 60% report their organizations have a strategy in place (a net of 30%). The study also found that the likelihood of having an AI strategy in place was highly correlated with the size of the organization: the largest global companies were the most likely to have an AI strategy.
In facing the question of whether to “build or buy,” pharma companies have historically embraced outsourcing capabilities, for example, “buying” as needed to bolster clinical trials, sales force, and advertising and marketing needs. Their approach to AI seems to mirror their efforts in other areas. Twenty-six percent of respondents report they have a service provider building out AI solutions and capabilities for the firm. Another 32% report they are purchasing commercially available AI solutions. Just under 20% are investing in AI startups and 13% are actually building AI capabilities in-house. Only 3% report they have acquired assets in the AI space.
As part of the research with pharma executives, we asked them about the companies they use (or invest in) related to AI strategy and implementation. The companies below represent those mentioned by pharma executives. Of note, the “platform” companies (Google, Amazon, and IBM Watson) were mentioned the most often as the partner companies are using to better understand AI.
Why would pharma invest in AI for marketing? All companies want to avoid chasing every “shiny penny” that emerges in the digital and technology space (for obvious reasons). The best and brightest often look for ways to use emerging technology to address specific business issues or problems, or to automate a high-volume business or marketing process. When it comes to future expectations, the greatest potential driver of future (or continued) investment in AI is … better customer insight. How can AI make that happen? In many cases, AI does the heavy lifting to help brands organize large data sets – making it possible for marketers and data scientists to focus on the data (or customer segments) where they need to focus their attention.
Right behind customer insights, pharma believes AI can help improve and optimize customer satisfaction. Although pharma companies have not historically invested in customer satisfaction (or experience) platforms compared with other industries, perhaps they are now watching how other industries are using AI (and data) to help identify key customer touch points and improve interactions with the company and the brand. Continuing down the list, they cite automating routine business processes, increasing revenue, and increasing efficiency as the next drivers for AI implementation within the firm.
We know the key driver to future (and continued) use of AI for marketing is better customer insights. Where are they realizing measurable value and positive gains today? Where and how is AI already making an impact on the business? The top area for value realized is also the top future driver – better customer insights (17% reporting current value realized). Beyond that, the second area is automating routine business processes (14%), following by increased efficiency (8%), and improving customer satisfaction (8%).We know the key driver to future (and continued) use of AI for marketing is better customer insights. Where are they realizing measurable value and positive gains today?
Our surveys of pharmaceutical executives revealed a new view of the state of AI in the industry today, answering key questions, including:
Overall, the feedback from the pharma industry is that while they are still focused on key trends like social and mobile, AI is quickly moving onto their radar – driven by the promise of a better understanding of the customer and a better system to deliver the optimal customer experience. While organizational structures and leadership mandates for next-generation moves such as AI take time to trickle down into the brands, the quick wins are becoming more and more common as teams increasingly include and operationalize AI into their 2019 brand and marketing plans.