Major Medical Insurance Provider Drills Into Marketing Effectiveness With phData
Customer's Challenge
The digital marketing and analytics team at a midwest medical insurance provider needed to understand the effectiveness of their marketing efforts.
But without the ability to drill into individual campaigns, marketing channels, and conversions to specific plans, they were unable to act on insights or experiment with marketing scenarios.
phData's Solution
phData’s data science team created the data harmonization and model in Python necessary for granular marketing insights. Harmonization brought disparate data sources together across digital and traditional marketing, enrollments, leads, and more.
The data model, in turn, helped the marketing team determine how much spend per person for a given tactic in a county would impact the total number of enrollments.
Finally, phData created a Power BI dashboard to allow the customer to conduct marketing scenarios based on how much they would plan to spend in a certain channel.
Results
Before, the marketing and analytics team had no way to bring its disparate data sources together, let alone leverage a data model to predict and track marketing effectiveness.
Now, the customer has the right data flow, models, and tools to accurately gauge marketing effectiveness. The organization’s marketing and analytics team can:
- Quickly access a cohesive dataset at the county-month level that reflects marketing spend, marketing impressions, leads, applications, and enrollments.
- Understand which marketing channels most impact their enrollments.
- Take action on these insights and test out a variety of marketing scenarios to weigh expected results.
The Full Story
As a major midwest health insurance provider with coverage in 12 states, the customer needed an effective way to connect their marketing efforts with member enrollment. The insurance provider covers 1.5 million individuals across employer-based plans, individual and family plans, and Medicare and Medicaid plans.
The customer’s digital marketing and analytics team struggled to paint a cohesive picture of how impactful their marketing tactics performed in terms of member enrollment.
With disparate data sources and unharmonized data workflows, the marketing and analytics team couldn’t get an accurate (let alone real-time) understanding of performance by campaign, region, persona, or plan. All of their metrics had to be tracked at a state level because they could not drill down any further.
The marketing team wanted to understand how spending in each marketing channel impacted the number of individuals who would enroll in their Medicare or individual and family plans.
phData stepped in to provide data harmonization, set up the right data model for marketing effectiveness, and created a Power BI dashboard for reporting and experimentation.
Why phData?
The customer chose to work with phData over other vendors because of our team’s proven track record with effectiveness models. From data creation and retrieval to specific reporting requirements, phData has helped dozens of customers improve their marketing effectiveness with data modeling.
Project Overview
phData’s data scientists created the data harmonization and model in Python. Over 500 disparate flat files had to be cleaned, aggregated, and combined to create a view that was needed for modeling.
From there, a linear regression was leveraged to determine how spending per capita in each marketing channel impacted the number of individuals that ultimately enrolled in insurance.
Finally, phData’s data experts visualized the model output in Power BI, making the insights accessible to the entire marketing and analytics team.
Step-by-Step: Data Harmonization, Modeling, and Visualization
phData kicked off the project by creating a process to harmonize disparate data sources. These included data sources for:
- Digital marketing
- Traditional marketing
- Enrollments
- Applications
- Market competitiveness factors
- Leads
Before, the customer did not have a one-stop-shop view where all of this information was joined together at the same level of granularity. Now, the data is in a state where the team can quickly determine:
- How much money was spent in each county
- The number of individuals that responded to an ad
- The number of individuals who applied
- The number of individuals who ultimately enrolled
Next, phData created a unique data model based on the customer’s competitiveness within the county and the types of plans that were offered. The variables for each model was the marketing spend by channel normalized by the total eligible individuals for a given county, along with some additional competitive factors defined in tandem with the customer team.
With these variables supporting the model, technical success was measured by an overall model fit of 85 percent.
From each model, the customer team was able to determine how much spend per person for a given county would impact the total number of enrollments.
From there, phData created a Power BI dashboard to let the marketing and analytics team conduct scenarios based on how much they would plan to spend in a certain channel. This allowed them to better understand which factors most impacted their enrollment numbers and how to adjust certain spending to achieve their enrollment numbers.
Take the next step
with phData.
Looking into data models for your organization or marketing team? Learn how phData can help solve your most challenging problems.
