Why Marketing Matters for HealthCare.gov

Open Enrollment this year performed far better than expected. Overall, HeathCare.gov states maintained about 95% of last year’s enrollment, even though the administration cut the length of open enrollment in half and slashed outreach by 90%. So does marketing even matter?

Yup. Based on the evidence we have, without the Trump Administration’s efforts to undermine enrollment, national enrollment would have exceeded 12.9 million enrollments or roughly 1.1 million additional people would have enrolled.

The first thing to know is that advertising is important, but it is one of many factors important to enrollment decision making. Another important factor is the cost that people pay for health insurance. Contrary to the Trump Administration’s misleading statements about “soaring” premiums, most Marketplace consumers saw lower net premiums this year than last year. For many Americans, prices were much lower than previous years due to the indirect effects of cutting payments for cost-sharing reductions aka “Silver Loading”. According to HealthSherpa, overall their consumers paid on average 13% less this year than last year. If that price decrease is consistent across HealthCare.gov states and state based marketplaces, it will mean that consumers saw the largest net-price drop in the marketplace’s short history. Therefore, it shouldn’t come as a surprise that low prices were a positive factor for enrollment this year — especially among younger consumers who are especially price sensitive.

If the Trump Administration had notified consumers about these historic savings by, say, advertising, enrollment would have easily outpaced previous years.

If advertising is important and cutting the advertising budget by 90% is negative factor (and one that disproportionately hurts the young and healthy), how can we know what really happened?

In past years, we measured the causal impact of advertising, so we could understand how many people enrolled who wouldn’t have without advertising, calls, email campaigns, etc. For years, CMS measured the impact of advertising on enrollment using industry best practices — randomized control trials (and when required RCTs combined with modeling). A randomized control trial tests a single variable e.g. whether or not a person received an email by placing people randomly into treatment (e.g. email) and control (e.g. no email) groups. This year, the administration denied this had ever been done and presumably halted measurement entirely.

The good news is that there were a number of large scale natural experiments this year. One silver-lining to the Trump administration’s massive cuts to outreach this year is that the cuts were so significant that it created significant variation — making it far more likely that we can learn something from the enrollment totals. However, none of these natural experiments are perfect: for example, there are important differences between HealthCare.gov states and State Based Marketplaces. But by combining multiple views of what happened this year and comparing the findings to the controlled experiments from past years, we can see if they tell the similar story.

Let’s look at three natural experiments that played out this year:

HealthCare.gov States vs State Based Marketplaces: The Trump administration cut outreach by 90% and navigator funding by 41%. The 13 State Based Marketplaces did not. However, last year Health and Human Services spent about $30 million on national Television that benefited both HealthCare.gov and State Based Marketplaces. HealthCare.gov states and SBMs that didn’t buy any television of their own would have experienced this cut most acutely.

  • SBMs outperformed the FFM by 5% this year. It the FFM had seen 100% of the enrollment it saw last year — about 450k additional enrollments would have occured.
  • However, this isn’t an accurate picture of the year over year performance. One of the first acts of the new administration was to cut outreach in HealthCare.gov states before the final deadline. This “ad killsuppressed about 500k HealthCare.gov enrollments last year (OE4).
Credit: ACASignUps.net
  • Factoring in these OE4 Outreach cuts, enrollment this year was just 90% of last year — with SBM states outperforming HealthCare.gov states by 10%. Assuming the HealthCare.gov states had performed at about the same level as SBMs, that would translate to about 9.6 million enrollments for HealthCare.gov — aout 900k additional enrollments which would have been on par with the all time high.

New vs Returning Consumers in HealthCare.gov states: Outreach is important for both new and returning consumers, but the Trump administration disproportionately cut outreach for new enrollment when they cut TV and the majority of digital advertising. Both TV and digital advertising are critical to reaching people new to the HealthCare.gov — how else does someone learn about it, especially when the President is claiming “ObamaCare is dead”?

  • The number of returning enrollees was 1.5% higher, with 6.28 million reenrolling this year vs 6.12 million last year. This is an astonishing number, because the marketplace had 500k FEWER potential reenrollees. Importantly, the schedule for renewal was the same as last year and critical channels for renewal (email and phone calls) continued uninterrupted. CMS has not released the final number of people who actively renewed yet, but going into the final week this number was slightly higher than last year.
  • In contrast to the strong reenrollment, new enrollment was 18% lower this year (albeit with half the time) — 2.5 million new enrollments this year vs 3.0 million last year. Again, one of the first acts of the new administration in January of 2017 was to cut outreach in HealthCare.gov states before the final deadline. Factoring in OE4 outreach cuts, new enrollment this year was 29% lower than it should have been last year. Assuming new enrollment should have held steady — roughly 1 million new enrollments were lost.

Short vs Long Open Enrollment: HealthCare.gov cut Open Enrollment in half, but State Based Marketplaces selected their own deadlines. Three marketplaces kept the same deadline as last year.

  • State Based Marketplaces had varying lengths of Open Enrollment this year, some were as short as 45 days others were as long as 95 days. Plotting performance this year vs last year by the numbers of days of Open Enrollment, you can see the variation in the HealthCare.gov and state marketplace performance.
  • Factoring in OE4 outreach cuts to the final HealthCare.gov enrollment total last year this variation becomes even more pronounced.

Combining all three factors: California and New York together represent over 59% of SBM enrollment, both released new enrollment numbers, both maintained the same enrollment schedule, and both invested in outreach specifically including millions spent on TV. Together they represent what HealthCare.gov performance in OE5 might have been. Remember both New York and California did not have the benefit of $30 million in National TV this year from HealthCare.gov advertising efforts, as they did last year.

  • Again, new enrollment in HealthCare.gov states was 18% lower this year — 2.5 million new enrollments this year vs 3.0 million last year. Factoring in OE4 outreach cuts, new enrollment this year was 29% lower than last year.
  • In contrast, California and New York together saw new enrollment that was 4% higher than last year.
  • If the HealthCare.gov states saw new enrollment at the same rate that California and New York saw this year, 104%, it would mean 3.1 million new enrollments this year — nearly 700k. Factoring in OE4 outreach cuts, it would have meant 3.7 million more new enrollments — nearly 1.2 million more.


All three of the natural experiments point to a similar conclusion — roughly 1 million lost HealthCare.gov enrollments. While these changes are not controlled trials, what we see here is a strong correlation between the administration’s cuts to advertising, change to the enrollment schedule and a drop in enrollment.

With that in mind, we know the following:

  • SBM states outperformed HealthCare.gov states in total enrollment by 5–10%
  • States with longer Open Enrollment Periods saw higher enrollment than States with shorter Open Enrollment Periods
  • HealthCare.gov new enrollment was 18–29% lower than last year
  • HealthCare.gov returning enrollment was 1.5% higher than last year
  • California and New York saw new enrollment that was 22–34% higher than the HealthCare.gov States

If we put all of this together, the evidence suggests that that there were both positive and negative forces on enrollment this year. Generally speaking, the positive forces disproportionately impacted returning consumer enrollment and the negative forces disproportionately impacted new consumer enrollment. The positive forces disproportionately benefited State Based Marketplaces and the negative forces disproportionately hurt HealthCare.gov states. The two most obvious factors that meet these criteria are the shortened HealthCare.gov enrollment period and the cuts to advertising.

In October, I estimated based on the controlled experiments from last year that at least 1.1 million enrollments would be lost due to cuts to outreach cuts. Of those, I said that nearly a million new enrollments would be lost (and 200k additional renewals would be lost in the 12 weeks after the end of open enrollment). Keep in mind, we saw advertising consistently drive enrollment year over year — so there is good reason to believe that it would remain consistent this year.

The simplest explanation is that cuts to advertising and the change to the enrollment schedule were the cause. What we see in the final numbers is further evidence that that the actions of the Trump administration cost a minimum of 1.1 million enrollments.

The one thing this approach cannot measure is whether or not the new positive forces this year (e.g. lower costs), paired with a real HealthCare.gov outreach and advertising program, might have had a synergistic effect that would have generated even higher enrollment e.g. if a business has a lower price AND advertises the lower price it will see more sales than if it just has a lower price. Performance in California and New York supports this case — particularly when cuts to National TV are considered in light of their strong new enrollment performance this year.

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.