The graph below is one of my favorite graphs to show when talking about analytics. The graph shows barriers to analytics adoption in a healthcare organization. While the graph comes from a 2010 study, these finding have been consistent in all my discussions with providers, payers, and pharma who are working to develop a data analytics program within their organization.
Barriers to analytics adoption
The study shows the number one barrier to analytics for organizations is data. This is partly access to the right data, or data quality, or the need for new data sources. But I also hear often that folks aren’t sure what data they have and what data they need for the questions they want answered.
The next five barriers to analytics are organizational – a culture of not sharing data, not understanding how to use analytics to drive business decisions, lack of leadership, lack of skills. These issues are symptoms of a lack of, what I call, an “organizational construct” for analytics.
Often, tech vendors (like the ones I worked for, IBM and Atigeo) focus on the technical challenge – the platform, the tools, the data management, the algorithms. The frustration for me, though, was that as a tech vendor, I was never able to help folks with the organizational construct.
Yes, we were sales consultants, helping the customer develop sound business use cases, pointing out the good habits of successful data-driven organizations, and showing what a successful organizational structure looks like. But we were not helping organizations actively deal with the organizational issues. We’d sell the solution and expect teams to have the business and technical analytics strategy and expertise to understand and do what we take for granted. And, of course, this runs into the resistance to change and fear of risk every modern organization has.
In healthcare and life science, we are still asking: How do we facilitate and accelerate the adoption of analytics across an organization? How do we remove barriers to analytics, resolve the issues that folks resist, reduce the risk of action, and provide a choice of paths with value at every step?
The need for a successful data analytics program
Part of the impetus for creating 777labs was to design a successful data analytics program for deploying data analytics solutions at healthcare organizations. A successful program covers both the technical AND the organizational aspects of analytics, helping with the development of the organizational construct in which to embed and successfully deploy the technical environment.
Based on our experience selling data analytics to healthcare and life science companies, we’ve developed a program that takes a stepwise approach to adopting and deploying a successful data analytics program across an organization. The program considers organizational awareness of data fluency (Listening), engaging leadership to iteratively design a data skills and tools strategy (Strategy), a stepwise approach to adopting data-driven applications and solutions (Delivery), and development of organizational data analytics fluency (Fluency).
We believe a strong part of a data analytics program is actively listening to the organization. We often see organizations deploying data products, only for them to go unused or abandoned. And there can be many reasons for this, such as a mismatch between data product and user needs, or lack of skills or understanding to how use data products.
Actively listening to the organization, via periodic surveys, will provide a view of the data fluency in the organization, identify where leadership needs to adjust the message and resources, and expose what questions are not being answered with data-driven solutions.
The dimensions of our surveys cover the user, the data product producer, the organizational data ecosystem, and the level of data fluency in the organization. The questions help discover the skill level of the organization, understand how well employees use data-driven tools, gauge how well the organization shares insights and data, and validate the direction of leadership in building a data-driven organization.
For 777labs, we stress that a data analytics program won’t succeed unless you have a clear idea of what you want to achieve with a data analytics program. Therefore, we’ve developed a methodology to help stakeholders better define goals and uncover needs across an organization. We then help the organization turn this into a plan.
There are a few issues that might make it hard to develop a coherent strategy, so we take an iterative and experimental approach, laying out multiple paths to success, rather than one monolithic and static strategy. The development and iteration of the strategy is tied to the listening, described above, but also taking stock of the current data and application environment. And developing a strategy involves many different groups with different goals. Finding common ground is important, yet might require multiple rounds of discussion and showing of proof-points. Hence, the strategy needs to consider the long term (where we want to go) and the short term (how can we test for value at every incremental step).
One key insight we have gained regarding adoption of analytics is that often the goals are too grand. Enamored with technical sophistication, vendors focus on the big picture, missing that the customer is struggling with the basics. For example, why do vendors push Machine Learning (now being marketed as AI or Cognitive Computing) when many organizations are still trying to visualize their data in simple descriptive graphs. Vendors are promising to help organizations run, when all organizations need now is to crawl. Indeed, we often say vendors want organizations to run a race like Usain Bolt, when organizations are struggling to get to the starting line.
Therefore, we have structured a step-wise delivery of data analytics. The steps are designed to start simply, but to form the foundation for the next steps. Early steps are meant to be delivered quickly (often in less than 30 days), for a low cost, using simple and easily accessible data sets. This allows the organization to ease into analytics, show value from the start, and not break the bank. And, yes, there is a strong component of finding, managing, integrating, and enriching data (the first barrier to analytics).
The last part of the program ties these all together. Knowing your organization, developing a strategy, and building things cannot happen in a vacuum. The users and the leadership need to demonstrate data fluency so that the whole organization can build a culture of sharing data, apply analytics to drive business decisions, ensure the connection between leadership and data-driven decision making, and recruit and develop data-fluent employees.
Listening helps the organization to understand, over time, the needs and improvement of data fluency across the organization. Strategy forms a live and public document that helps employees and leadership understand the direction the organization is taking with data-driven decision making. Delivery of data products provides the vehicle to make an organization data-driven. And the focus on building competency in data, to build data fluency, helps the organization become data-driven at all levels.
Data is a big barrier in adopting analytics across an organization. But it is the lack of an organizational structure that will lead to failure, no matter how sophisticated the technology or algorithm. A successful data analytics program takes into account organizational awareness, an iterative strategy, a stepwise approach to adopting applications, and the development of data fluency.
What do you think of this? How does this match how you have been deploying analytics at your organization or at your clients’ organizations? How might the process outlined here help you with your vendors or with your data analytics efforts?
Let us know.