Today, when we think of advanced analytics, we think of AI solutions like self-driving cars, automated personal assistants or cancer diagnoses by image recognition. But when many medtech leaders go to work, all they see is reports. For commercial leaders and field managers, “analytics” is limited to slicing and dicing some data, at times with better visualizations and dashboards. Medtech needs to up its analytics game, but what does that entail?
What’s often lost is that there’s a whole spectrum of analytics techniques available to medtech decision makers, and while innovations like personal assistants are great, it’s the middle of the spectrum where most of the value is created, and where typical companies can derive the most competitive advantage. And the good thing is that many of these middle-of-the-road analytics techniques are well-established and already work for many industries with proven results.
Medtech companies can thrive using these techniques, which fall within five levels of analytics maturity:
- Augmented data: Most organizations start their analytics maturity journey with better access to data through dashboards, reports, ad hoc query tools and visualizations. A typical commercial leader needs regular, timely KPIs for sales, margins, productivity, etc., but the amount of data available is staggering and continues to grow, so it’s no longer sufficient to produce more and more reports and let users figure out what to do with them. Embedding alerts, clearly showing outliers and warning signs, or in some cases spelling the insight out as a suggestion are a few ways to augment the data to bring attention to what’s most important. Leading companies also enrich their data by assigning scores to customers, opportunities or other entities as a way of providing quick and uncluttered insights for the field and executives.
- Optimization: The next level of decisions can’t be made by analysts visually slicing and dicing the data, even when it’s sufficiently augmented but requires an optimization. Optimization is a mathematical model that minimizes or maximizes an objective function without violating constraints. Whether in SAS, Excel or specialized software, an optimization model can optimize many operational and strategic processes. Take the size and deployment of a sales force: Our findings published in an Economist study show that analytics-driven optimization in this area can lead to revenue improvements of 3 to 5% relative to a typical non-optimized deployment. Another model can optimize marketing mix, yet another can minimize inventory in all nodes of a supply chain or determine the best pricing or discount strategy.
- Prediction: Predictive modelling makes another step forward, enabling machines to sift through lots of data to spot dependencies and leading indicators, and predict key events of interest. Predictive models use statistical methods and a variety of techniques from linear regressions to decision trees and neural networks. With the rapidly decreasing cost of computing and the availability of open-source code for many types of models, many companies now use predictive modelling for a variety of business problems, such as predicting customer churn, calculating the lifetime value of a customer, and assessing rep engagement and turnover. In many cases, insights that prevent undesired events or speed up desired ones may bring companies tens of millions in extra revenue. One company has built a sophisticated model to predict IDN mergers and developed a program to proactively align pricing and contracts ahead of these mergers in high-risk areas to avoid price erosion.
- Recommendations: Once machines can predict an outcome, they can also be trained to suggest a response or share what has helped in similar cases in the past. Prescriptive models can differentiate between potential actions and determine which will lead to better outcomes. “Next best actions,” which recommend sales tactics to the field, and “proactive launch,” which monitors product launches and suggests adjustments to strategies or tactics, are two of the many applications that leading healthcare companies deploy to optimize sales.
- Automated execution: While these four levels of analytics provide insights to humans, they depend on humans to take action. When a company can automate the decision and the action itself, it can derive the most value from analytics. Keep in mind, we’re not talking about replacing the entire commercial function with robots. Automation doesn’t have to be super advanced to bring outsized returns. Think about automatically shipping another set of spare parts based on seeing a depleting inventory at the warehouse. What if you could send an automated welcome package to a new customer, or an event invitation to a customer who is about to consider switching to a competitor? These applications are achievable for most medtech companies and can make a big difference.
While popular business literature is all about sexy and advanced applications, the real competitive advantage for medtech companies is in maturing from humans sifting through data to increased efficiency via optimization, predictive modelling and, ultimately, automating commercial and operational decisions. These types of analytics use cases have been well-established in medtech and other industries and are relatively easy and inexpensive to implement, yet they create very sizable revenue gains and cost savings. Advancing analytics maturity is the most consequential improvement that commercial and operations leaders can make for the success of their company.