Hopefully you’ve already read our companion piece “How Predictive Modeling Works for Medicare Marketing,” but if you haven’t had the chance yet, you should read the article here. In that article, we introduced the idea of looking at predictive modeling like you’d look at horse racing, and how you could improve your odds just betting on 10 horses. Bringing this back to AEP marketing, what if it was 100,000 horses?
Modeling helps scientifically determine who is going to respond to your 2018 AEP campaign by:
- Using 2017 AEP direct response leads
- Appending a wide array demographic and lifestyle data elements
- Statistical tools (such as logistic regression) to determine the MOST predictive 10 to 20 data elements
- Applying response model results to all 100,000 prospects
The end result is the model predicts who has the propensity to respond to your specific AEP campaign. Each prospect is scored and ranked (just like the horse race) from 1 to 100,000.
What is a decile?
Since there are many prospect scores in the model results, to make it easier, the scores are typically divided into 10 equal groups based on predictive score, i.e. 10,000 in each decile. Here is an example broken down for a service area of 4 counties:
Building an AEP Circulation Plan based on the Model Scores
As a marketer you can use the response model scores to develop an AEP mail circulation plan based on budget, cost per lead and cost per new member goals, we well as by county, business rules, network and competitive environment, etc.
Based on the model scores an example AEP circulation plan (direct mail and digital) would be as follows:
- Drop 1: All counties top 6 deciles
- Drop 2: Green shaded only
- Drop 3: Green shaded only
- Drop 4: County A, B, and C top 3 deciles; County D top 6 deciles
- Drop 5: County A, B and C green shaded; County D: all deciles
I will save you a lot of time: these five various drops = 250,000 total contacts or 50,000 x 5) which was our goal!
Why use modeling?
Modeling has been proven over time, across all industries to consistently outperform other targeting methodologies. No more spray and pray. It leverages all data and prospect information available to make the most targeted offers to the most responsive prospects for your specific organization.
In addition, modeling maximizes marketing ROI by communicating either significantly less to achieve the same results or communicating the same amount for improved results
Lastly, we have been reviewing “response” modeling but you can also use modeling to predict other behaviors such as conversion, lifetime value, attrition and win back.
By incorporating predictive modeling during 2018 AEP you can surge ahead and win just like Secretariat did at the Belmont Stakes: with other horses barely in sight: