A Data-Driven Culture: Making Data a Part of Everyday Decision Making

data-driven cultureHealth Catalyst believes in using data to drive better performance and we encourage our clients to embrace a data-driven culture. By internally deploying the same analytics infrastructure we’ve developed for our clients and utilizing the same improvement methodologies, we have learned a few best practices to overcome specific challenges. We believe some of these best practices are applicable to health systems attempting to pursue data-driven improvements.

A Data-Driven Culture

When it comes to establishing data-driven improvement processes, health systems face significant challenges. They have to implement the analytics infrastructure, best practices, and adoption processes required to improve healthcare quality and cost. At Health Catalyst, the focus is on operational analytics to help us scale by improving internal efficiencies.

Building a technology infrastructure is relatively easy. The challenge is operationalizing data-driven decision-making. Creating a data-driven culture—making data a part of everyday decision making—is much more difficult.

A Hybrid Team Structure

At Health Catalyst, we use our analytics tools and deployment methodologies to drive improvements that enable us to scale our business. Structurally, we utilize a hybrid approach to organizing our analytics team: the analytics team reports up through a single group, but individual analysts are embedded in specific departments. The central nature of the group ensures consistent use of data and scalability of a centralized EDW and analytics platform. Embedded analysts in specific departments empower operational leaders to prioritize specific improvement efforts and allow the analysts to build domain expertise in the department’s unique datasets and nuanced processes. Ideally, departments view the embedded analysts as part of their team, driving trust in the analyses, which, in turn, increases utilization.

The analysts work as part of permanent, collaborative, and multidisciplinary work groups, consisting of data analysts and operational team members. These work groups use data to manage improvement initiatives. This structure ensures constant communication between analysts and operations, enabling iterative development.

Journey to a Data-driven Culture: 6 Ways to Overcome Common Challenges

  1. Prescriptive Analytics, Big Data, and Machine Learning Sound Exciting: However, First Nail the Fundamentals

As an executive, it is fun to know your company is using the newest technologies or analytics techniques. As an analyst, it is tempting to first pursue cutting edge data problems. For both constituents, it is important to remember to first target the low hanging fruit. Traditional business intelligence has been around for decades for a good reason. Ensure analysts are reviewing metrics that drive significant portions of an organization’s bottom line. Understanding where variation and inefficiencies occur will help drive improvement efforts. These efforts will also identify areas where it is appropriate to dedicate additional resources to see if more complex analytical methodologies can be used to find further insights.

  1. Building an Application is Just the Beginning: Focus on Operationalizing Learnings

It is one thing to build an analytics application, it is quite another to get people to use it. Adoption and training often takes as long, if not longer, than creating the application. Work groups should focus on identifying exactly when data from an application should be integrated into an individual’s workflow. Once this is determined, an email message summarizing a new application or a detailed product description will not be sufficient to drive utilization. It is essential to host multiple hands-on training sessions to explain a new application and address any user concerns or questions. The analytics infrastructure also needs to be able to monitor application usage. Monitoring usage enables the team to identify key users or sub-departments where adoption is lower than desired. Usage data can also be used to identify specific unexpected use cases (i.e., understanding which features of the application are being embraced). These insights enable the team to refine the analytics initiative and increase utilization – this is an iterative process.

  1. Operational Team Members are Busy: Leadership Needs to Ensure They Have Dedicated Time to Participate in Improvement Work Groups

Involving operational team members in data-driven improvement efforts is critical. However, simply adding improvement efforts on top of the already very full plates of operations team members is not sustainable in the long run. Executive engagement is essential to create capacity to allow operational team members to lead work groups. Executives also need to create an organizational culture to make analytics and improvement everyone’s responsibility, ensuring idea generation from all team members and broad adoption of recommendations suggested by work groups.

Senior leaders need to:

  • Set expectations that data-driven process improvement is a priority
  • Assist in identifying members of the cross-functional, permanent teams to drive improvement
  • Provide on-going support to ensure team members are given time to work on improvements
  • Regularly review ‘outcomes’ measures to track the progress of improvement initiatives
  1. What is the Point of Operational Improvements: Tying in Financial Data will Provide Purpose

Operational improvements can often seem like a ‘nice to have’ as opposed to a necessity. Tying in operational data to financial data helps put into perspective the operational improvements that need to be made in order to drive revenue or reduce costs (and to drive quality, in the case of health systems, where patient satisfaction data should also be tied in). Tying specific operational metric improvements to a budget that all team members are rewarded on is even better – providing a very clear tie between granular operational metrics and the higher level financial implications. Tying in financial data takes courage and leadership from executives. A culture needs to be built in order to support such financial and operational transparency across departments.

  1. Why Should I Trust the Data: Develop a Detailed Knowledge of the Transactional Systems and Create Transparent Data Transformations

Getting people to trust data, when doing so will mean changing their processes, is hard. Human nature seems to lead people to question the data they see. Without a detailed understanding of how the analysis was performed, one wrong number or any uncertainty regarding the data can lead teams to question the entire analytics process.

We have found that a key to building trust in the data is to share exactly how the data is collected and entered in the transactional system before starting an improvement effort. While work groups and analysts initially view understanding the underlying transactional data as a less exciting piece of the analytics process, we have determined it is critical to success. Demonstrating how the data is gathered and input enables work groups to gain an understanding of possible limitations and flaws of the data, often times leading to data capture improvements that include implementing data quality processes to help avoid the “garbage in, garbage out” problem.

Once a work group understands the data available, it is important to create transparent analytical applications – black box data transformations often lead to confusion and lack of trust.

This detailed level of understanding of the data pipeline is crucial to ensure long-term scalability. End-users must learn to enter clean data in to the transactional system at the onset to prevent constant data cleansing at the analytics layer and the analytics layer needs to be as simple and transparent as practical.

  1. A Tendency to Overburden Team Members with Data Collection: A Cost/Benefit Analysis Is Key When Deciding to Collect Additional Data Points

Organizations engaged in process improvement tend to want to track as much data as possible. This means tracking both outcome measures (e.g., hospital-acquired infection rates) and process measures (e.g., how often clinicians wash their hands). Organizations must also decide how long to collect and track process measures.

Collecting data in perpetuity is ideal for analytics, but not always practical—especially when the data has to be entered manually. Organizations should be very deliberate when deciding to add permanent data-tracking requirements and consider the costs versus the benefits for each one. For example, tracking whether clinicians wash their hands before performing a particular task imposes both hard and soft costs in terms of clinician time and data-entry fatigue.

Performing an overhead value analysis based on the frequency of usage and the value of the data is critical.

Keep an Eye on the Target

Health systems creating a data-driven culture face considerable challenges. Under pressure to improve care and reduce costs; implementation of the analytics infrastructure, best practices, and deployment processes are critical to success. Starting this journey can be daunting, but it is important to start with the fundamentals and not forget the importance of adoption, training, and utilization of the data by operational team members. In order to set up cross-functional work groups, senior leaders will need to free up time for work group team members to lead data-driven improvement efforts. Senior leaders also need to promote a culture of transparency and pragmatism in order to reap the full benefits of analytics and sustain the efforts. The ultimate goal? Making data a part of the everyday decision making process. Building a data-driven culture—constantly turning data in to action—is the ideal that we believe organizations should target.

Loading next article...