Pharmaceutical Management Science Association

With advances in machine learning and artificial intelligence techniques, marketing tech and ad tech solutions are emerging that promise to transform the way multi-channel marketing and customer engagement are carried out in practice. Many vendors and consultants are proposing machine learning and AI-based techniques and solutions that promise rosy scenarios for marketing ROI. But do we really know what these techniques are and how they fundamentally differ from traditional approaches? What is their state-of-the-art, advantages and limitations? How can they be effectively employed for omni-channel marketing, attribution and customer engagement?

In this seminar, we will focus on:

• A primer on machine learning and AI techniques with example applications that span across industries
• State-of-the-art in attribution technologies, challenges and the future
• Challenges in optimal media-mix allocation for omni-channel marketing and evolving approaches
• Advances in ad tech solutions for effective customer engagement

Seminar will involve discussion, case studies and interactive sessions.

When we develop physician level communication strategy, we consider direct mail, dimensional mail, third party options, rep level interactions, and of course e-mail, which serves as a primary delivery mechanism for hyperlinks, downloads, e-details and content. As the true work-horse in our omni-channel approach, it is critical that we move away from an ad hoc approach to integration of e-mail into our targeting, and take a more analytic approach to the frequency, content, and sequencing of our e-mail communication.

The needs of the audience depend greatly on their specialty, the nature of their practice (and support staff), access, writing behavior as well as channel propensity. Using various statistical methods, we will create a dynamic roadmap for email best practices that takes into consideration the therapeutic area and the market share of a brand, along with its stage in its product lifecycle. This physician and therapeutic level insight is critical to inform how often, when and to whom do we email, to ensure that a given campaign yields the highest possible return on investment.

In order to truly move towards harnessing AI – Artificial Intelligence, organizations must first take a step towards building their capabilities in a different AI – Augmented Intelligence. We have focused on enabling an ecosystem of standardized business context, analytical building blocks, and business process understanding through our EoCTM platform. With a holistic view towards their business context and the reusable analytical capabilities, organizations can now augment their current business processes with additional intelligence by leveraging advanced analytics and machine learning techniques. Additionally, we present two examples for how we have helped organizations leverage cross industry best practices. The first is analyzing switching customers in the consumer product industry and how insights can be applied to patient switching. The second, is a general increased use of patient data to enable patient-centric marketing which helps organizations build better marketing capabilities for both physicians and payers.

The ability to target physicians with the next best channel combination and next best message has been the goal of many pharmaceutical companies. The suggested approach uses a process that is based on machine learning and is rooted in the net present value based construct successfully used by CPG companies.

In the program, value states are created, and migration of customers among value states are observed. Statistical machine learning is used to understand the relationship between promotional activity both in terms of channels used and messages delivered and the migration of physicians across value states. A champion model is selected and deployed to score the universe of physicians. Finally, the results of the chosen predictive model are compared to actual deployment in the most recent time period to create in the next best action recommendation. The process requires significant computing power and is based on AWS as well as contemporary predictive analytical tools.

The advances available to brand marketers and marketing operations professionals are monumental. The advent of tracking, tagging, and analytics have placed customer-centric marketing at the fingertips of every marketing professional. Marketers are now able to use these three to create a website, test this website against others, track the engagement, tagging items on the website, analyze, and then update to deliver and meet their customer preferences and then deliver content, messaging, or enable recommendations. Marketers now have the ability to evaluate and run real-time analytics and make quick decisions based on measurement. These insights enable more effective and focused personalized marketing that lead to better engagement and greater campaign outcomes. Pharma marketing is at a unique crossroad with technological advances beckoning marketers to learn and leverage these tools to empower their customer-centric marketing.

This session will address the following questions:

1. How important is it to create responsive marketing campaigns across digital channels that are engaging?
2. Is it valuable to interpret data regularly and make adjustments/refinements to audiences or strategies?
3. How may data be evaluated and what insights may we draw?
4. How may a more customer-centric marketing strategy be enabled? What are the tools and methods to enable more personalized, targeted marketing?
5. What are the benefits of improved engagement and how to maintain the momentum?

Historically pharma and healthcare companies, like other industries, have focused on impression and reach based measurement for marketing ROI. However, a critical element is repeatedly missed in typical ROI or attribution studies – patient or HCP engagement across social media and specifically the quality of that engagement.

Agencies and marketers have always found it difficult to keep a track of the numerous digital engagement triggers that platforms provide and across campaigns related to drug launches, self-health diagnoses, therapy guides, etc. Additionally, Pharma has kept refocusing on search and display as primary channels, which don’t measure nuanced engagement metrics.

However using patient or HCP engagement as a guide to a company’s digital marketing strategy, especially for social media, is critical in creating competitive advantage and transparency. Therefore, besides targeted audience selection and segmentation, an effective Omni channel strategy needs a unified full funnel measurement framework – a bespoke analytical system that maps and compares all channel performance on a single measurement philosophy.

A unified measurement method, thus, allows customized benchmarks for each channel & content type to be harmonized – creating a single source of truth through an advanced analytical model. This approach provides real competitive ability to evaluate campaign performance across reach, passive engagement, active engagement, conversion and advocacy across multiple channels and benchmarks. This session will address bringing this full funnel framework alive, and address key challenges, like defining & measuring engagement and a popular digital marketing phenomenon called ‘Walled Gardens’.

For many pharmaceutical companies today, communicating to healthcare practitioners has become more challenging. Sales force numbers have been on a declining trend while rep access to physicians has been getting more and more limited. As a result, the need for an effective integrated multi-channel communications program has never been more urgent.

While many pharmaceutical companies have taken note of this reality and are communicating across multiple channels to get their message in front of physicians, just 20% use cross-channel data and attribution to evaluate all marketing touchpoints.1 Employing analytic strategies that evaluate multi-channel campaigns within the individual channel silos will not tell the full story and may lead to inaccurate insights on the value of each channel and therefore, result in a less than optimal multi-channel campaign.

In this presentation, we will share an analytic framework for data integration and analysis that will help pharma companies move from the traditional HCP targeting approach based on prescribing volume to a next generation: more preference-driven, individual-centric way of communicating to HCPs. We will show how through a deep analysis of engagement patterns by content and channel and how these patterns influence prescribing, we can leverage insights from the analysis to create promotional strategies that deliver and incent high-value, emotionally-resonant brand engagements.

The presentation will include a case study and we will take the audience through the framework starting from an overview of the methodology to the data used and finally, to the analysis itself along with the results and strategic recommendations for implementation. It will include lessons learned as well as data/analytic-driven innovative ideas to embed these new segmentation schemas into machine learning/artificial intelligence tools to implement optimal personalized multi-channel engagements.

One of the obstacles with developing a multi-channel marketing strategy is that the underlying data is far from ideal. Indeed, there are two problems. The first one is sampling bias. We claim that insights gleaned from the sample apply to the entire universe when the sample is not actually representative of the entire universe. The second problem is distortion. The data does not accurately depict what it is supposed to. It is fraught with missing, ambiguous, and inaccurate information that emanate anywhere along the capture-transmission-encryption line. Forging ahead with the analysis as if the distortions were non-existent is foolhardy. It may lead to insights that are outright wrong, with devastating consequences on the credibility and reputation of the analyst.

The first problem is well known and has received ample treatment, warranting no further discussion here. The second problem, on the other hand, is as significant as the first one but has received short shrift treatment as it is largely dependent on the industry at hand. This session focuses on distortion in claims data.