Try this statistic on for size: The 500 petabytes of digital healthcare data that existed in 2012 is predicted to reach 25,000 petabytes by the year 2020. That’s an increase of nearly 50 times the amount of data from just eight years prior!
Healthcare marketers may be swimming in data, but what’s important is to not drown in it. The key, in other words, is to not just have the data – but to figure out what you’re going to do with the data.
As will be discussed during today’s Evariant webinar, analytics in healthcare is critical for making effective use of data. Predictive analytics can help drive improved health outcomes for patients and give you the biggest bang for your marketing buck.
But the question is: What, exactly, is predictive analytics?
What Is Predictive Analytics?
From a high level perspective, predictive analytics is the process of deriving insights from patterns and correlations in data and using that information to drive more strategic marketing campaigns that will result in improved patient outcomes.
More specifically, predictive analytics allows you to pinpoint optimal targets, with statistical precision, and determine the consumers and patients most likely to respond to your marketing campaign.
Think of it this way: All of the data you’ve collected about your prospective patients (health records, consumer data, financial data, etc.) is all just background noise unless you can systematically organize the data in a way that tells a story.
Let’s say you build a predictive model for targeting a bariatrics procedure. By using analytics, you know that your best prospects are going to be individuals who will likely end up having the highest response rates for the bariatrics campaign.
Ideally, you’d want to create a few test cells for your campaign. For example, for anyone 55 and older, you would try one message, and for prospects under 55, you would use a different message.
You’d also want a third group, a random sample of all prospects that would get both messages. The goal is to see if the targeted messaging, in this case, by age, results in higher response rates.
Testing will usually lead to increased response rates over time. Without testing within the same campaign scheme, you would have a difficult time knowing which message or channel was most effective.
Once you have the results from that first campaign, you can roll them into the next.
Essentially, you’re increasing the probability of higher response rates for the next version of the campaign because you’ve used the response analytics and the test cell results to make sure that you have the best of the best for the next campaign.
Put simply, you’re narrowing down the sample. The term we use a lot is ‘you’re selecting out those who you shouldn’t target’.
What’s the Difference Between Predictive & Concurrent Analytics?
As will be discussed during the webinar, predictive and concurrent analytics are not synonymous.
Predictive analytics is a subset of concurrent analytics (many variables used at the same time) used to optimize analytics to specifically predict a target or outcome.
You can, for example, use a set of diagnostic, visit history, sociodemographic, socioeconomic, lifestyle, and interest variables (shown right) to predict the likelihood that a patient prospect is a candidate for hip replacement surgery.
Concurrent analytics, on the other hand, is an applied process and technique where at least two types of analytics are used simultaneously. Concurrent analytical elements for the bariatric use case above would include preselects (demographics and BMI), segments/clusters (mosaic group or channel), and predictive modeling/scored data (patient prospects with the highest scores).
An overview of steps involved in concurrent analytics is as follows (it’s important to note, however, that the process of optimizing concurrent analytics is subject to change, based on the specific variables):
- Define outcome/target
- Define and match target and prospect samples
- Clean all data
- Run descriptive and exploratory statistics
- Organize segments and clusters
- Build predictive models
- Apply preselects and filters (including channel)
- Validate preselected sample elements against outcome and target samples
- Test using control groups, A/B splits, etc.
- Conduct response validation analytics
- Change what needs to be changed
The overarching goals of predictive analytics are to reduce marketing costs, increase responders, and improve the success of future marketing campaigns. A very big part of driving ROI has to do with the recurring revenue that comes as a result of having analytics in place, not just the one-off success of a particular campaign.
Recurring revenue is predictable, stable, and can be counted on in the future with a high degree of certainty. It both minimizes customer churn and builds brand loyalty, which are important considerations seeing as how it is three to seven times as hard to sell to a new or non-patient as an existing patient. The idea is to acquire new and non-patients at the lowest possible cost, thereby allocating more resources to patient retention efforts.
In today’s hyper-competitive healthcare landscape, it’s simply not enough to collect data on your patients and prospects – you have to actually analyze, interpret, and apply the data in order to drive the success of your marketing campaigns.
With predictive analytics in place, your organization is setting itself up for increased patient response to marketing campaigns, leading to improved health outcomes and greater patient retention and loyalty.