• Sustainability
  • DE&I
  • Pandemic
  • Finance
  • Legal
  • Technology
  • Regulatory
  • Global
  • Pricing
  • Strategy
  • R&D/Clinical Trials
  • Opinion
  • Executive Roundtable
  • Sales & Marketing
  • Executive Profiles
  • Leadership
  • Market Access
  • Patient Engagement
  • Supply Chain
  • Industry Trends

Harnessing the Value of AI In Customer Engagement


By using correct data and pertinent ML algorithms, customer engagement leaders can identify deep insights into customer behavior and make the right interventions to transform the customer experience.

Ganes Kesari

Ganes Kesari

Gregg Fisher

Gregg Fisher

Pharmaceutical firms are leveraging artificial intelligence/machine learning (AI/ML) to improve customer experience through customer targeting and personalized engagement. This is commonly bundled under terms such as “recommendation engine” or “next best action” and is being deployed in a variety of engagement contexts, including:

  • Patient support program (PSP) interactions between support specialists and patients
  • Non-personal campaigns for HCPs and patients
  • Personal sales force engagements with healthcare providers (HCPs)

The objective in each case is to target customers most likely to respond, at an opportune time, with personally relevant messaging delivered via content formats and channels most likely to generate a response.

In this article, we share some real-world examples of companies using AI/ML. We summarize key success factors inspired by our clients on maximizing the success of customer engagement AI/ML initiatives.

Here are three key areas where AI/ML is delivering a notable impact on customer engagement.

Adherence via personalized patient support interventions

One of our clients, a mid-size biotechnology firm specializing in rare diseases, runs a highly successful patient support operation. They have been using AI/ML to build a set of tools to support the day-to-day decision-making of their patient support specialists to personalize their interactions with patients.

Their patient support specialists needed to peruse over a dozen dashboards to make decisions on which patient outreach to prioritize on a given day, something that was taking time away from their primary responsibility –connecting with patients. According to the head of data science, “Our decision support tools have been able to improve productivity for our staff and increase the consistency of decision-making. They have also made it easier for the leadership to seamlessly change priorities and run real-time micro-experiments to finetune their approach.”

This company built its decision-assist toolkit with a tiered approach. The first tier incorporated a set of agreed-upon metrics and business rules to decide whom to reach out to. The next layer leveraged various data sources and AI models to improve outreach, initially focusing on channels and timing and then outreach content.

Improving patient conversion

A key challenge for commercial teams is getting patients to fill the prescription they have received from their doctor. Patients may hesitate to begin a new medication for various reasons, including concerns about side effects or lack of information. The impact of poor patient conversion is significant in terms of health outcomes (when patients don’t receive medicines they need in a timely fashion). It also represents a substantial loss of revenue for pharmaceutical companies, retailers, and even providers.

Eduardo Cornejo, a former Senior Director of Digital Innovation at Sanofi, and current affiliate of The Stem’s consultant network, had an opportunity to lead an initiative to pursue this opportunity for the Dupixent brand. According to Cornejo, Sanofi outsourced its patient conversion activities to an external service provider, employing over 500 nurses to handle outreach to thousands of patients. Before the use of AI/ML, these nurses would reach out to patients to encourage them to fill their prescription without the benefit of insights drawn from numerous commercial data sets. The AI/ML program allowed Sanofi to personalize its outreach to patients, including channel selection, messaging and timing, drawing on data from retailers like CVS, external social media data, and internal marketing and commercial data.

“The opportunity for AI/ML was to outreach to patients when they want us to … with the content, at the time they want us, and media they want us to versus the ways that the brand had been doing before the project,” Cornejo explained.

Improving sales force targeting and messaging

The best sales reps know how to become vital partners with their HCP customers, helping them improve patient outcomes and make informed treatment decisions for their patients.

AI/ML next best action helps make the behavior of the strongest reps more scientific by supporting targeting and interaction decisions through ingesting and analyzing varied data types. For example, using AI/ML to analyze patient claims or EMR data has helped some of our rare disease clients identify ultra-rare disease patients to help physicians make a rare disease diagnosis that would have taken years.

Other clients of ours are using channel usage data (for instance, are they “digitalist,” “traditionalist,” or “hybrid” channel users) and learning style data to assist in channel and content formats selection for customers. Companies are also making use of data around product adoption status, content consumption behavior, and message acceptance to inform the right messaging. AI/ML can help quickly find meaningful patterns across these data types that unassisted reps might miss or would historically take too long to uncover.

“AI can help guide diagnosis along the patient journey and increase the level of care physicians provide patients,” says Tom Gaschler, the Head of Data & Insights at Takeda, a Stem client. By providing personalized data-driven recommendations, sales representatives can truly position themselves as trusted partners to HCPs.

The above are three examples of customer engagement AI/ML innovation happening within the pharmaceutical industry. Now that we’ve seen ways that artificial intelligence and machine learning can help, let’s understand what it takes to bring such solutions to life and tackle the most common challenges that leaders grapple with.

While AI can deliver tremendous enterprise value, implementing it isn’t straightforward. Many organizations make the mistake of applying AI in areas where it provides suboptimal value. When they pick the right initiatives, leaders find it challenging to make the build-versus-buy decision. Even with groundbreaking AI solutions, most organizations struggle with end-user adoption and quantification of enterprise value from AI.

From our experience advising pharmaceutical organizations on AI engagements, four critical success factors can help tackle the above challenges.

Pick impactful initiatives by prioritizing business outcomes. Today, we’re hearing much chatter on AI-driven large language models. Most organizations are actively trying to discover business areas to apply tools like ChatGPT. The flaw with this approach is that it prioritizes the tool even before choosing the business problem.

“When our users bring up AI in a conversation, what they are really asking for is support with data insights for better decisions,” said Gaschler of Takeda. “We must navigate the conversation to help surface the core business problem faced by the sales team,” according to Gaschler.

Once we distill the questions they want to be answered about the patients, we must collect the needed data and apply pertinent analytics to discover actionable insights. Irrespective of whether this solution is built with descriptive analytics, statistics, or AI, it is sure to deliver value when it solves a pressing customer problem.

Make the build-versus-buy decision through internal assessment and vendor evaluation. Today, there’s a growing pool of AI/ML talent to hire from and several off-the-shelf tools to manage the journey in-house. On the other hand, the AI/ML ecosystem is flooded with vendors who can tailor custom solutions to meet specific needs. Deciding whether to build or buy can be a crucial factor influencing project success.

When leaders decide to build AI solutions in-house, they need to ensure they are prepared to staff for the right roles and onboard business and technology teams across the organization. When going the ‘buy’ route, assessing vendor capabilities specific to the chosen problem is essential. Check their track record not just with implementing AI in similar contexts and capturing business value as well.

Address fear and uncertainty by educating users on data literacy. One of the most common roadblocks that AI journeys encounter is organizational culture. End users are often unwilling to change or are troubled by job security fears. The executive at a mid-sized biotechnology client pointed out that this impacted tool adoption and business outcomes.

The leadership must play an active role in storytelling their vision and educating how AI will be a crucial enabler for their business growth. The aforementioned mid-size pharma client involved with using AI to improve patient support outreach brought representatives from the patient support teams in focus groups they formed while conceiving the AI initiative. Their functional leadership signaled to the respective teams the criticality of this initiative.

Explaining that their job wasn’t at risk, they helped users upskill on data literacy - the ability to understand and communicate with data. Thanks to these efforts, the patient coverage of this AI solution doubled to 40% in less than three months. The team is building upon the momentum to get the number closer to 100% by the next quarter.

Measure AI value from roll out and scale across the organization. Organizations that struggle to report the value generated by their AI investments often lack a clear definition of target business outcomes, associated KPIs (leading and lagging indicators), and the linkages between them.

Gaschler’s team measures the value of their next-best-action solution through leading indicators such as sales rep usage of the AI decision-support tool, the rep’s acceptance of the recommendations the tool makes on which customers to target and which customers interactions to take next , and the reps’ reasons for dismissing a particular recommendation offered by the tool. Takeda is also evaluating the measurement of more outcome-oriented metrics such as prescription volume and increase in product market share that can be attributed to use of the AI decision-support tool. Such measures are often called lag indicators since they take a while to show up.

In our AI advisory work, we’ve seen that the success of value measurement depends on how clearly the end outcomes are identified and mapped to target metrics. Doing this early in the project with explicit stakeholder buy-in ensures acceptance and stakeholder alignment.

Another challenge is the lack of well-defined tests and control groups to judge the impact of the intervention. This can be done by designing experiments comparing different geographies, reps, or customer groups depending on the nature of the intervention.

We have examined how AI/ML can help pharmaceutical firms transform customer engagement and the key challenges leaders must tackle while considering them. When it comes to AI, there is much conversation about tools, technology, and people. The four critical success factors discussed highlight that AI’s most significant challenges aren’t technical but involve people.

Leaders have a critical role in identifying how AI can help realize their mission and story-tell this vision to the entire organization, with alignment of the workforce across the levels to orchestrate AI initiatives, measure value, and tackle change management issues.

Related Videos