January 22, 2018
3 Key Insights From the MM&M NextGen Data and Analytics Workshop
If you value data and analytics but aren't completely comfortable in mining your brand data for actionable insights, you're not alone.
Data from a March 2017 survey by Dun and Bradstreet and Forbes Insights indicates that most organizations don't have the skills in-house to meet their data science needs. This is the case even though companies consider data and analytics expertise to be a must-have. More than half of the companies surveyed (55%) must hire external vendors to meet some or all their data science needs. And 27% of survey respondents consider the lack of in-house skill a major obstacle to their data and analytics programs.1
Fortunately, there are many programs to keep healthcare marketers up to date with the latest in data analysis and related technologies. Recently, I had the pleasure of speaking at one such workshop, "NextGen Data & Analytics," organized by Medical Marketing and Media (MM&M). During my session titled "Artificial Intelligence—Super Hearing for Healthcare Marketers," I shared two case studies that illustrate how actionable insights about healthcare marketing content strategy insights can emerge from deep analysis of brands' digital marketing data. In these cases, we applied an AI solution that Arteric software developers created with commercially available AI algorithms.
As an adjunct to the main program, MM&M interviewed workshop speakers about their thoughts on various trends, challenges, and issues facing marketers who are seeking to extract insights from their brand data. My thoughts are included below.
What are the biggest challenges in applying data and analytics to healthcare marketing today?
As healthcare marketers, we are currently facing an embarrassment of riches in data and in tools for analyzing the data. We have data streaming in from almost all of our key communication activities. A major challenge for pharma and biotech marketers is that the data are often siloed in agencies, departments, and databases across our ecosystems.
Frequently, valuable data are unknowingly abandoned or allowed to rot in storage because of a lack of integration into a single accessible data store. Marketers may not know that the data are available or may not know how to budget for the acquisition and storage of the data. Marketers often have no clear pathways for providing the raw data to agency and internal partners for deeper analysis. Although many analytical tools and models have become cheaply available, there is currently a critical shortage of people with the talent and vision to apply those tools.
It will take several years for agencies and marketers to develop the internal capabilities and capacity to leverage the explosion of opportunity that the wealth of data creates. Given the lag in talent and process, it is essential to aggregate and store the data today so that we have a sufficient pool of data once we achieve the capabilities to leverage the data through artificial intelligence and machine learning.
What are some of the best-use cases for artificial intelligence in healthcare marketing happening now? What do you expect to see in the near future?
We have found that AI is most effective when applied to narrowly defined business problems, specifically those that involve lots of scale and weak but important signals. Thus, when we identify business problems that 1) have a lot of data associated with them, 2) have modest time pressure, 3) can be narrowly defined, and 4) will have a big impact on customer interactions with the brand, we investigate if AI or machine learning will be a part of the solution.
Conversational interfaces such as chat bots; screenless interfaces such as Siri, Google Home, and Alexa; and natural-language search interfaces, are driving the explosion in the use of AI and natural language processing by healthcare brands.
Today, Arteric is leveraging latent semantic analysis as well as language classifiers and translators to identify weak but important signals hidden within hundreds of thousands of customer interactions with our customers' healthcare brands, to identify new customer segments and create personalized content. We are also using these data to build reference libraries of utterances and natural-language searches that will be used to inform conversational interfaces in 2018 and beyond.
The challenge today is to aggregate and preserve data that may not appear to be useful but will be critical to next-generation AI, machine learning, and analysis-driven content generation.
As drugmakers face continued pressure on their business models, how can data/analytics improve how they communicate their stories and key messages?
For years, we have been building personas and segments in an effort to abstract our customers' key characteristics. Although this abstraction is good for communicating at scale and is a powerful tool for customer acquisition, it is not authentic. And, more profoundly, it does not enhance customer retention. Personas and segments are a metaphor for the customers' needs and wants. Personas lack the context and language required to connect with people on an individual experiential basis.
Our customers, the end users, expect us to know them and respond in a way that reflects a deep personal relationship with them and direct knowledge of their personal needs and wants.
Over half of online customers (59 per cent) now have an expectation that their health care system will meet the same level of customer service received from delivery companies such as Amazon.1
To retain our customers, we must do things that don't scale and we must communicate in a personalized and authentic voice. We must connect in their context and in their language—language that adapts to the cultural, geographic, and situational context. That is the move from the persona to the individual. Each patient or healthcare professional has a story and has come to expect that the brands they interact with will know their story and anticipate their needs.
The hidden insights, which are the very weak but extremely important signals required to make this leap, are buried within hundreds of thousands or millions of rows of data. Arteric's experience is that we can mine these insights by using cheap, off-the-shelf AI and machine learning APIs if we narrowly define what we are looking for. We work with Google, Microsoft, and Cortical.io to mine hundreds of thousands of interactions, looking for these insights and how to evolve the brand's messaging.
In one case, we analyzed 55,000 searches for a blockbuster brand and discovered approximately 250 opportunities to deeply connect with customers. When we analyzed those opportunities, we discovered that the brand ranked for 110 searches but successfully answered the customers' questions only 4 times. The analysis took 80 person-hours and was powered by data that the brand possessed in house but wasn't leveraging. We responded by creating a content strategy to leverage the missed opportunities. We scaled this approach to analyze 250,000 searches for another customer. That analysis identified a Spanish-language market segment that the brand wasn't targeting but had in-house market research to validate. Scaling the approach was only possible using AI and machine learning. Our technology achieved in 2.5 minutes what had taken us 80 hours previously.
Data, analytics, and AI are essential tools for doing at scale what was previously impossible. The key frontier that healthcare marketers should be exploring is how to move from the abstract to the individual, through either personalization or conversational interfaces, and how to connect with people by using the language that people use every day. Healthcare marketers can do these things right now, by using cheap technology and readily available data.
You can read summaries of all of the workshop's sessions by downloading the MM&M eBook. Arteric's search analysts and I are also happy to answer any questions you might have about extracting the most value from your brand's marketing data. Contact us at 201.558.7929. And if you'd like to see how we apply artificial intelligence to accelerate discovery of marketing insights, read this blog article or listen to a podcast published on Life Science Marketing Radio.
1. Fullerton L. Over half of patients expect the same level of service from their healthcare provider as Amazon claims study. TheDrum.com website. http://www.thedrum.com/news/2016/05/12/over-half-patients-expect-same-level-service-their-healthcare-provider-amazon-claims. Published May 12, 2016. Accessed January 22, 2018.