top of page
  • Writer's pictureKarsten Schmidt

Deploying effective Artificial Intelligence methodology in commercial operations


Undoubtedly, Artificial Intelligence (AI) is a field which is growing at a very fast pace and will be changing our way of working from now on dramatically. Today, we want to take a look at the impact that AI will have on the effectiveness of field forces in Life Sciences. Therefore, we have invited Rasim Shah, Director at Okra.ai, for an interview.


Karsten: Thank you for accepting our invite to this interview, Ras. To start, would you like to tell our readers a bit about your background and what inspired you to work in the AI-space?


Ras: After 20 years in Healthcare, both client (Merck, CareFusion) and vendor side (IQVIA, M3), spanning both commercial and medical roles I realised change was on the horizon. Big data was the main topic of conversation and I had seen many efforts across the globe to capitalize on this emerging trend. In 2017 however, the noise around Artificial Intelligence was growing and I began researching the various applications in line with big data and personalisation. I myself have suffered from a chronic illness since the age of 7 and have been hospitalized a few times without warning. My thought process was simple, could data and AI really warn me ahead of time if something was wrong, is personalized medicine hype or reality - and that was where the journey began for me to start working within AI.


Karsten: During the pandemic, field forces had to learn to embark on omnichannel customer engagements due to the lack of HCP accessibility. This was not an easy task. Now that we are leaving the worst part of the pandemic behind us, I can imagine that it will also be challenging to tell them that Artificial Intelligence is the next hot topic that they must embrace to become more effective. What are ways to overcome the initial skepticism prevailing in commercial teams?


Ras: AI has been a hot topic for many years and the concept of using technology to augment decision making is widely accepted in Pharma. In fact, the concept of Next Best Action (NBA) for commercial, where teams are essentially using predefined rules to suggest and dictate the activity of sales teams has been around since 2015. Today in 2022, this concept of NBA is still confused with AI, so the first challenge is NOT getting individuals to embrace AI to be more effective, it actually centres around educating individuals on what AI is, what it relies on and what good looks like.


AI for life sciences is a very specialised area, with multiple, complex data sources being integrated to support decision making, so it's essential that explainability and transparency are provided to foster confidence between human and machine. To overcome skepticism, commercial teams have to first embrace that the rules of engagement have changed. This new era of engagement requires data and AI, and AI capability is not something everyone in the market is providing so it's easy to invest in the wrong sales pitch. Commercial teams must realise value fast and in doing so must reimagine their success parameters or key performance indicators (KPI’s). With AI, impact is NOT simply increasing email sends, driving click rates or increasing frequency of calls. These traditional KPIs are no longer fit for purpose. With AI we must focus on HCP needs and engage where the patients are. This means dynamic opportunity and priority understanding and drive engagement where the need is greatest. From there we can use the best channel and content to optimise the interaction.


Companies must involve users early in the change process. Automation of data analysis is the future of our ever-changing world and requires humans to be involved. For too long commercial teams have relied on traditional consulting approaches to problem solving, leaving entire teams frozen in time and in effect moving backwards as others pave the way. I have personally seen companies develop or buy generic AI solutions and then try to leverage them across different functions like commercial and medical. This does not work. There is a fundamentally different approach that we use focussing on modular productisation, leading to a suite of AI capabilities that combine to meet the different needs within the enterprise.


In summary the skepticism that I see largely originates from one of 3 things: 1. When the understanding of the subject, in this case AI is not clear so it's easier to revert to what you know. 2. When the internal strategy is not aligned - for example building Data Science teams competes against using genuine external capability and 3. When investments have already been made in alternative technology and the cost of change is too high therefore delaying adoption.


Karsten: Can you share some more insights of how AI should be designed for commercial teams and how it works? Also, what data need to be collected and processed?


Ras: Today, with AI we don’t have to wait years to generate evidence that can influence decision making. Commercial teams must make use of all the available data in both the public and private domain. Only then can they truly generate impactful insights in a dynamic way. AI enables us to combine environmental data, disease prevalence, CRM and much more and dynamically predict the number of patients that need treatment at a specific location. When you combine this information with physician need, you will start to drive engagements that will deliver impact. As I stated before, without AI we will be frozen in time and unable to connect the dots at the speed required, and these connections are not made using a smart computer alone, it's equally important to control how data is gathered, joined and modeled before feeding it to algorithms. At our company we have been able to extract features - which means pulling data from multiple data sources and across multiple domains. This includes prescription data, CRM, demographics, clinical trials, clinical outcomes and somehow, make sense of what it all means. Currently this data is not collected to train algorithms, but rather captured for transactional and workflow purposes and many will have come across them in a scattered way; but today AI technology can synthesise and translate those scattered features and make sense for AI algorithms to learn from. The output of the models are largely predictions and suggestions but commercial teams must also have reasons. This is the WHY, and for me the most important part of commercial AI.


Let’s start by thinking about intelligence. Generally, we trust human intelligence, because we are familiar with ourselves and with the human brain. We trust what is familiar and what resembles us. As a result, we trust that humans can provide clever solutions. But what about artificial, non-human intelligence? How can you trust the intelligence of a computer when you do not share the inner workings of its digital brain? This ability to explain is crucial in building trust. And more than anything else, this is why our technology has been able to get a high traction with those who use our systems. Because through being able to offer explanations we can trust, we stay connected with one another and in control. Producing great explainability takes a lot of work. It’s not just about translating complex algorithms into simple language. It’s also about working out the real cause for a recommendation.


Illustration: Applying AI with explainability in commerical operations


Karsten: I can imagine that it must be quite cumbersome and time-consuming to set up commercial AI for an affiliate considering that different data often sits in silos and needs to be integrated and aggregated.


Ras: For traditional technology providers this is most definitely the case. But as an AI, data driven company we have found that the key to successful adoption of AI is two-fold. First, a quick rollout from start-to-Minimum Viable Product (MVP), used by actual commercial teams. Secondly, ensuring it is easy to go from that MVP to large-scale deployment at speed with minimal client resources. We have put a lot of effort in shortening the time to having a first live system, from an industry-average of 6 months to our standard of 8 weeks. We have streamlined the process of rolling out to new clients including onboarding and integrating data. We are now able to roll out our system into major markets with minimal maintenance effort - the equivalent of half a day per month vs one week per month for traditional solutions. Most people who talk about scale in Pharma today, talk about people and teams (offshore) - the more the better. Unfortunately, this is no longer the way to scale. Our approach has been to redefine the meaning of scale and talk about automation which means less people and not more and frees up the time to focus on impactful actions!


Cloud-native solutions where the architecture enables rapid scalability from the start is critical, instead of procuring and maintaining hardware. A modular approach is also key where updates and scale of individual components can be made independently, and this is something our commercial suite benefits from and allows us to roll out new features and address security issues very effectively with minimal downtime.


Karsten: What is the level of adoption of AI and resulting recommended actions you have seen amongst your clients?


Ras: Adoption is growing at speed, but for simplicity I can split Pharma clients into early and late adopters. For me there are two categories of early adopters: category 1 are clients that have quickly realised (through experimentation) that a rules-based approach does not work and cannot scale, so they have sourced genuine AI automation to support their efforts. Whereas category 2 are clients that have seen the limitations of the rule-based approach adopted by others, and have actively chosen a future fit, data driven AI solution from the outset.


Late adopters archetype are still battling internal data science efforts to build solutions whilst also navigating the debate of ‘what Global wants’ vs ‘what Local needs’. Concerns around data quality also slow their transformation progress. Many clients believe they need to find the end of the rainbow when it comes to data quality and volume. This is a never-ending challenge as data is growing and evolving every day. Clients that believe their data is good enough, soon find out that data capture for AI is different to data capture for PowerPoint reports, so AI engagements uncover and expose data gaps early in the process - which is a good thing and enables a future state that is required for AI at scale.


Karsten: There are of course nuances on how commercial teams operate in different geographies due to distinct healthcare environments. How can an AI approach still be applicable amongst different countries and regions?


Ras: An AI approach must be data driven, outcome driven and explainable. These three factors allow us to navigate the distinct nature of any healthcare system and quickly understand WHAT is driving specific outcomes (for example sales) and WHY. There are millions of data points that we leverage across each country which allows us to understand and validate the engagement landscape prior to delivering any recommendation to commercial teams. From this point on we work closely with users and commercial leads to further enhance the system and customize it to specific needs based on the outcomes clients wish to drive. An AI system cannot remain stagnant, it must learn from every interaction and continue to enhance outputs in partnership with clients.


Karsten: Thank you, Ras, for sharing these interesting insights related to AI in commercial operations. I think that our readers have now gained some more clarity between the difference of Next Best Action algorithms versus AI.


In summary, we can say that the objective of AI in commercial operations is to support decision-making by integrating multiple, complex data sources. The confidence between human and machine can only be achieved when AI-driven predictions and suggestions are provided with explainability and transparency. When field forces understand where needs are greatest they will prioritise their engagements accordingly whilst applying best channels and content that will optimise customer interactions. Key for successful AI-adoption in commercial operations is to shorten time to market with minimal viable products which are then easily scalable.


What will help drive AI-adoption for commercial teams from your perspective?


Please feel free to comment below. If you find this article interesting, please share it.

For further reference please contact info@xeleratio.com or rasim@okra.ai


 

Xeleratio Consulting Ltd.

We help Life Sciences executives improve sales performance with innovative best-in-class Business Excellence tools and methodologies . Expertise in Business Excellence has been gained with over 12 years of working in different global and regional roles in the Life Sciences industry.


 

Please feel free to reach out and request a free strategy session whenever convenient. See all articles on Xeleratio Consulting Blog

留言


bottom of page