The medtech industry is at a critical juncture. Digital solutions put in place during the lockdowns related to the COVID-19 pandemic have evolved into clear customer engagement preferences. Companies that wish to remain competitive are rethinking their traditional commercial models and designing differentiated digital strategies for the future.
Companies in the medtech sector have access to more data—and more advanced analytics—than ever before. As the availability and competencies of AI continue to grow, it’s being deployed to address an ever-widening range of commercial challenges, including microsegmenting customers for providing account insights, creating propensity-to-buy models for account prioritization, recommending next best actions for customer engagement,1 suggesting tender prices for bid support, and helping predict and prevent customer churn (Exhibit 1). Now is the time for medtech companies to reimagine their customer engagement models by incorporating these and other capabilities.
Medtech companies that have begun to explore commercial AI use cases are already seeing significant business benefits and commercial growth. These early-adopter companies are reaping multiple improvements, including a 1.5- to 2.0-fold increase in customer funnel metrics, such as the number of identified leads; 50 percent higher proposal conversion rates; and up to 10 percent increases in incremental revenue, according to McKinsey analysis.
AI use cases are seemingly limitless in the medtech commercial organization, enabled by new capabilities in data and AI—and that’s part of the challenge. While some companies lead the pack in AI adoption, others are overwhelmed by the possibilities and are making common missteps. They may be waiting for the perfect data set or advanced machine-learning tools and overlooking the immediate benefits of more basic algorithms. Some are so busy planning massive, longer-term AI strategies that they miss out on incremental AI wins and experience. Others with overstretched commercial teams are farming out the work to IT and seeing limited results.
AI is a fast-moving space. Companies that wait for the optimal tool, data set, algorithm, use case, or start time risk being left behind; those that begin sooner rather than later will accrue the greatest value. This article lays out five frequent missteps companies make—and what commercial organizations seeking to build a winning customer engagement model can do instead—to achieve near-term benefits and build a foundation for the future.
Five potential pitfalls when exploring commercial AI use cases
Medtech companies can avoid five pitfalls when engaging with AI and analytics in commercial use: waiting for the perfect data or technology, presuming that only the most advanced AI will deliver insights, assuming that data scientists and field reps can’t work together, pursuing only the areas of greatest opportunity, and adopting a “go big or go home” approach to advanced analytics.
1. Waiting for the perfect data or technology
All types of customers have come to expect the seamless, intuitive, omnichannel consumer experiences provided by their favorite online retailer or preferred financial institution. Those experiences can inspire medtech leaders to transform their own customer engagement models (Exhibit 2).
But business leaders can mistakenly assume that progress requires the massive data volumes and advanced technology stacks that digital tech giants possess. Trying to emulate those giants can trigger concerns about the breadth and quality of data in the industry. In fact, the medtech industry has lagged in its accumulation and management of data. Logging valuable data such as client visits or sent emails was a low priority historically. Internal data sets were often ineffectively managed. Sales organizations ran on intuition rather than insight. Thus, the value of analytics was limited.
However, it’s imperative not to let the perfect become the enemy of the perfectly good. Commercial organizations can reap significant value from data and analytics without new data sets or an ideal technology foundation. As one leading US-based medtech player with a broad portfolio of capital equipment and associated consumables found, a “good enough” data infrastructure can enable valuable analytics use cases, including providing guidance to representatives about the customers who are most likely to purchase select products in the next three months—and yielding insights behind that guidance. Rather than wringing hands over whether there were enough data or worrying that reps would never adopt analytics-driven processes, the company acted. It focused on doing a few essential things—and doing them well:
- It created a standard data taxonomy as a backbone for internal data, and it assigned owners to maintain data hygiene and best practices for data storage and compliance in the cloud.
- It purchased from a single outside vendor both treatment and usage data, such as claims and procedures, at the account and healthcare professional levels.
- Most important, it linked the internal and external data sets using a common identifier.
As the initial use cases proved their benefit, the company purchased additional data sets, such as payer mix, patient dynamics, and healthcare professionals’ channel preferences. It combined the data, analyzed it, and fed the resulting insights to its customer relationship management (CRM) systems for its sales reps and customer service agents.
Another medtech company began even more simply. The company, a global leader in imaging, wanted to explore opportunities in the long-tail customer segment. To get started, it identified the data sources that needed cleaning, devised a master data repository, and manually pulled the data. Without building an advanced algorithm or hiring a data science team, it used spreadsheet functionality to develop a data-driven lead-scoring application that provides a clear benefit to sales reps. The company has since integrated more of its data and technology stack, adopted a new CRM system, and launched a cloud marketing platform.
The finish line is always moving. Companies will never have all the data they think they need, and better tools will always be in development. But while some companies wait for the ideal, others are rolling up their sleeves to standardize data taxonomies, get the most from existing technology, integrate internal and external data, and build out analytics use cases that differentiate them in the marketplace. The best approach is to get started with a pilot, prove its value with a minimum viable solution, and use the lessons learned to develop requirements for the data and technology stack.
2. Presuming that only the most advanced AI will deliver insights
Medtech leaders can become enthralled by the potential of sophisticated algorithms to generate accurate predictions. Indeed, there’s a lot to get excited about. But companies don’t need the most advanced machine-learning techniques to derive significant insights. In fact, more basic approaches can yield better outcomes, especially as commercial organizations begin their AI evolution.
One US-based multinational medtech company with a focus on surgical equipment and supplies learned this lesson as it built a predictive next-best-action engine. Commercial success relied on the company’s ability to engage various stakeholders in a hospital, to demonstrate the value and benefits of the medical supplies it produces. Thus, the ability to provide sales reps with recommendations on how to better engage those stakeholders was a high priority for the business.
The company put most of the effort into data engineering and creating an algorithm to give reps highly accurate account-level recommendations on how to improve engagement with customers (Exhibit 3). In the process, it brought together internal sales and marketing data with external data—for example, publicly available health system data. As the predictive model was still being developed, the project team used simple analysis and business rules to create a quantified view of account potential and penetration, an invaluable tool for account prioritization. Combining the “simple” account prioritization tool with the “sophisticated” next-best-action recommendation model resulted in a more holistic solution than the predictive model alone provided. Sales reps could use the account prioritization tools right away as predictive modeling continued. Some reps took time to trust the models’ recommendations, and without the additional analyses, they might have abandoned the solution altogether.
When they define their data and analytics, medtech leaders can begin with the outcomes they seek and then make their technology choices accordingly. High-value insights are often uncovered early in the process, even before sophisticated modeling occurs.
3. Assuming that data scientists and field reps can’t work together
Technical teams tend to lead commercial AI projects in medtech companies. The assumption is that product managers, data scientists, and data engineers have the talent to deliver the best solutions. The business typically is involved in upfront gathering of requirements and, later, in user testing—with minimal involvement in solutions development. After all, what do managers and sales reps know about machine learning? An equally important consideration, however, is how much data scientists know about the challenges of the commercial organization.
Leaving the commercial enterprise out of developing solutions is a mistake that creates technical and change-management risks. Solutions may not address the business context or underlying needs, and commercial users may resist buy-in and adoption. The business can play an important role as solutions evolve throughout development. Full value is captured only when there is close collaboration among experienced sales reps and managers, who bring their business judgment, and well-trained data scientists, who translate judgment into analytics.
Please contact the authors if you’d like to understand more about McKinsey’s diagnostic offerings for commercial AI.
A US-based medtech company, piloting a propensity-to-buy solution, created a cross-functional team with two key roles: user champions and analytics translators. The user champions were sales team members selected for their open-mindedness and eagerness to innovate. The analytics translators served as liaisons between the technical and business sides of the project. The user champions ensured that the voice of the business was heard throughout the 12-week project, and the analytics translators made certain that business needs were translated correctly into requirements for technical teams.
The nature of the solution evolved during the pilot. While the initial goal was to identify high-potential customers, the solution pivoted into a next-best-action use case. The resulting solution provided more value to the sales team and the business—and would never have emerged without the involvement of end users.
Medtech commercial leaders pursuing AI opportunities would do well to bring business and science together from the start. Best practices include creating a cross-functional team with key business, technical, and translator roles—and instituting agile development processes involving business stakeholders at all phases.
4. Pursuing only the areas of greatest opportunity
Medtech leaders may incorrectly assume that AI’s value to the commercial organization correlates to customer size. Bigger, they surmise, is better, resulting in some hesitation to leverage commercial AI solutions, such as lead scoring, for smaller customers.
Large accounts can indeed drive a significant volume of the business and eat up a fair share of resources for sales and postsales service. The top ten to 20 accounts are commonly responsible for the large majority of sales in some product categories. In addition, companies tend to acquire more data on those larger accounts.
But while analytics can reveal insights into the big accounts, medtech companies can generate insights throughout the client portfolio. When one global capital and consumables company created a propensity-to-purchase model to guide sales reps on the customer upsell opportunities and the churn risks, it was surprised by some of the outputs. Big names were at the top of the list of upsell opportunities, but there also were some midsize clients and more traditional firms presumed to be less interested in exploring new products. As the sales reps visited these tier-two and tier-three accounts, they confirmed the opportunities, increasing their confidence in the model.
AI can uncover significant value throughout the client portfolio. Because smaller clients can easily be overlooked, the value of analytical models to those accounts can be even greater. A lead-scoring engine can help to identify and prioritize the most economically attractive accounts of any size.
5. Adopting a ‘go big or go home’ approach to advanced analytics
Once medtech executives understand the potential value of advanced commercial analytics, they may be eager to go all in, involving every country and business unit from the start. They may try to develop global solutions for churn prediction, lead scoring, and next best actions. But beginning too broadly and attempting to solve multiple business problems at once are perilous pursuits. They can create significant development challenges, enhance complexity, increase risk, and ultimately delay delivery.
Limiting the scope of early efforts is critical to establishing important proof points as soon as possible. Medtech companies that have been most successful in building and scaling commercial AI solutions begin with clear business objectives, motivated participants, and a limited scope of products.
One global company with a broad product portfolio in orthopedics was eager to develop a lead-scoring solution for its field reps. After clarifying the business goal, the company conducted a pragmatic assessment of feasibility at the country, business unit, and product levels, including the availability of local resources and the quality of available data, for example. Using the study findings, the company prioritized two countries and two product segments for the lead-scoring solution.
Limiting the scope ensured the focus and collaboration for developing a minimally viable product quickly, with fewer country- and product-specific requirements to integrate. The solution could later be adapted to a particular country or business unit. Such an approach—agreeing on one business objective and starting small—enables medtech companies to derive insights more rapidly and prove initial value, thereby creating momentum for broader AI adoption and business impact.
How to get started
For medtech companies that have embarked on or plan to start a commercial AI journey, the best way to begin is by identifying and prioritizing potential use cases and assessing internal analytics capabilities. Developing this holistic view of the status quo helps companies identify strengths and gaps relative to best-in-class performance. Commercial organizations must therefore examine not only data and analytics capabilities but also design thinking, agile execution, and change management as they integrate data and advanced analytics to create differentiated customer engagement models.