For Providers getting started with business intelligence (BI), and even those looking to expand their existing BI programs, the sheer variety of tools and packages available in the market can be confusing. I often get asked for help in sorting out various possibilities and the pros and cons associated with each. There are a myriad of vendors offering analytics tools, data visualization tools, data integration tools and pre-built provider-specific analytics that address different areas of need. Many vendors offer on-premise and software-as-a-service options and have varying levels of functionality that can be purchased. Despite all the variations and claims from vendors that their tools are different from everything else, I feel BI deployment options for providers can roughly be grouped into five categories. A fundamental understanding of these choices should help providers better navigate the BI landscape as they formulate their strategy. There may be a few choices that cut through these boundaries, but most can still be convincingly classified into one of these five buckets. My goal is to provide an unbiased view of the options for implementing provider BI and their associated benefits and limitations.

Option  1: Analytics provided by packaged vendors

Providers have the option of buying pre-packaged analytics functionality from their existing operational systems vendors. This is among the simplestoptions to getting started with BI but has its limitations. Since electronic medial record (EMR) vendors  have only recently started to ramp up their analytics capabilities, it is not considered a core competency. This approach may result in only partially meeting the organization’s analytics needs. The data sets addressed by this option are generally limited to those already contained in the underlying application and it is generally not possible or very difficult to add data from other systems. This option does have the benefit of closer integration with existing workflows and relatively quick to implement. I would recommend this option to providers that have very limited IT resources and are only looking for a relatively basic BI program. These providers also need to be sure that their underlying core systems are there to stay for the foreseeable future.This option can become the stepping stone to a different strategy in the future.

Option  2: Software-as-a-Service (SaaS) packaged analytics

There are vendors offering packaged provider analytics as a service. Required data is sent to the external vendor using automated ETL processes at pre-determined frequencies. Until recently, SaaS options were mostly limited to operational and financial analytics. There are a few vendors now that also offer clinical and outcomes analytics using the SaaS model. The obvious concern with this approach is data security and privacy, though top vendors allay these concerns using well developed internal controls. The other area to watch out for is the limitation on customizations allowed by these vendors as well as restrictions on data refresh frequencies. That being said, this is a good cost-effective option for many smaller providers because the vendors are able to leverage economies of scale in data management and infrastructure cost.

Option  3: Data visualization tools

Data visualization tools allow users with little technical know how to create their own graphical and tabular views of data with low dependence on IT resources. There are several vendors in this space and a few of them stand out because of differentiated visualization capabilities. These tools are able to access the underlying application data sets directly and can e deployed with or without a traditional data warehouse. Many provider IT leaders prefer these tools because of reduced need for support staff and user training, as well as favorable pricing compared to mainstream BI tools. However, these tools may have limited scalability when dealing with large data volumes. Performance suffers when directly querying databases that are not built for analytical reporting. Also, in scenarios where reporting needs to be done across heterogeneous data sets from multiple underlying applications, data visualization tools are generally not a good choice.

Option  4: Pre-packaged point solutions

Point solutions typically address a specific area of analytics in depth. Examples include areas like readmissions management, predicting hospital-acquired infections, meaningful use reporting, departmental scorecards and optimization of nursing staff. Though it may seem tempting, it is generally not a good idea to create an entire BI strategy using a basket of point solutions from different vendors. Point solutions typically address data specific to their needs and it is difficult to analyze or reconcile data across multiple such solutions. These solutions can work very well in conjunction with one of the other options. These can also be a good choice for providers that do not have a data warehouse or a broad BI strategy yet, but still need to address specific analytics needs.

Option  5: In-house data warehouse

An enterprise data warehouse built specifically for analytics has always been the most versatile option for providers committed to BI as an organizational priority. There are multiple architectural approaches for building a data warehouse and starting with subject-specific data marts can work better for many organizations rather than tackling an enterprise strategy upfront. Several mainstream vendors have excellent and highly competitive analytical tools that can sit on top of a data warehouse. A well-built data warehouse can support organization-wide analytics and can have a multi-year lifespan. A data warehouse brings a variety of data sets together, thus enabling cross-sectional analytics. With the right tool choices and optimization strategy, scalability to large data volumes and user bases is possible. For large providers thinking of a “big data” strategy that incorporates non-traditional data sets, a robust data warehouse is among the fundamental building blocks. All this versatility comes at a price. This approach takes the longest to implement and has the highest cost among other options. Also, skilled IT resources are required for the build as well as for ongoing maintenance. The data warehouse needs to be populated from complex underlying systems, which is often a daunting task by itself.

There is no one-size-fits-all answer and providers need to pick their BI approach based on a combination of organizational and external factors. I welcome your comments based on your observations and experiences.