Understanding the 3 Primary Types of Healthcare Site Selection Models
From The Buxton Co
As the healthcare industry continues to shift to a consumer-centric strategy, a growing number of providers are implementing site selection models to guide their real estate decisions. Real estate site selection models come with many unfamiliar terms, which can feel daunting to first time purchasers.
In this blog, we’ll explain the three most common types of healthcare site selection models so you can have more informed conversations with potential vendors.
Healthcare Site Selection Model Type #1: Service Line Model
What is it? A service line model is a theoretical model developed based on factors that are believed to influence performance for a specific healthcare service. The model may use variables such as area draw, supply of providers, service line demand, core consumer profile, and more. Because the model is based on theory, no patient data is required but can be incorporated into the consumer profile if provided.
What is the output of the model? Service line models produce an index score that tells you how a site’s potential for that service line compares to the average potential for the service line in the geographic area the model is calibrated to assess. If a score of 100 represents the average level of opportunity for the service line in that market, then a 120 score indicates the site has 20% more favorable characteristics than the average site. The site score sheet usually lists the individual variables that make up the overall score so you can see how it was derived.
Is it for me? Because service line models don’t require any of the healthcare provider’s data, they can be used for organizations of all sizes and are quicker to implement than other model types. Service line models are popular with health systems who offer many specialties and need separate models for each, providers who would prefer not to share data with an analytics vendor, and small specialty healthcare provider networks who are just beginning to grow their location count. Service line models are calibrated for specific markets or regions and are not suitable for providers seeking a single model to use for national expansion decisions.
Healthcare Site Selection Model Type #2: Benchmark Model
What is it? A benchmark model is developed “from scratch” based on factors that are proven to correlate with an organization’s unique performance. The process begins with identifying the profile of the primary type of consumers you serve and calculating how far those consumers travel to visit your locations. From there, other factors – such as healthcare provider supply, healthcare demand, favorable cotenants, and more – are tested to identify the combination of variables that best explain performance.
What is the output of the model? A benchmark model produces an index score that compares the expected performance of a site to your existing sites based on the variables included in the model. The site score sheet usually lists the individual variables that make up the overall score, so you can dive deeper into understanding the site’s dynamics.
Is it for me? If you have 21-50 locations that 1) offer the specialty being modeled and 2) have been open for at least a year, then a benchmark model is likely a great fit for you. This model type is popular with retail health and other specialty health providers. Some health systems choose to use a benchmark model for their primary care locations but use service line models for other specialties since the service line models don’t have location count requirements. Benchmark models are appropriate for healthcare providers considering national expansion and can take into account regional differences.
Healthcare Site Selection Model Type #3: Predictive Model
What is it? A predictive model, also called a forecasting model, is the most complex of the three site selection model types. It uses linear regression to provide a forecast for a specific performance indicator, such as number of visits or projected revenue. The predictive model includes many of the same variables as the benchmark model but goes beyond the benchmark model by examining how your locations affect each other, which is referred to as “cannibalization.”
What is the output of the model? Unlike a service line or benchmark model, a predictive model produces an actual performance forecast, such as 7,500 annual visits. Index scores and/or raw values for the underlying variables are usually provided.
Is it for me? Producing a statistically reliable forecast requires a large sample set of good quality location performance data. Healthcare providers with 51+ locations that 1) offer the specialty being modeled and 2) have been open for at least a year typically have the data required to develop a predictive model. For most health systems, a predictive model isn’t the best fit, but it may be a great fit for a retail health or specialty health provider. These models are appropriate for healthcare providers considering national expansion and can take into account regional differences.
If you are looking for a partner who can help you make the right healthcare site selection decisions, consider working with Buxton. Explore our solutions for health systems and specialty health providers to learn more.
Buxton is the leading customer analytics firm that helps organizations identify who their customers are, where those customers are located, and the value those customers have to the organization.
2651 South Polaris Drive
Fort Worth, TX 76137
If you have an opinion on the retailing or retail real estate industries, take this opportunity to share your thoughts. Articles should run between 400 and 800 words. Topics can, be general in nature, consumer observation or specific to retail concepts or practices.
Articles will be posted for at least one week and will then be placed in the Editorial Archives. All articles submitted will be read and considered but we cannot guarantee publication. Each published article will carry the submitters byline (if desired) and is a free service to our community.
Article ideas and suggestions are also always welcomed. Contact PVS@PlainVanillaShell.com