The Department of Health and Human Services is under new management. Sylvia Burwell has taken over for Kathleen Sebelius. Burwell’s new lieutenants include Andy Slavitt and Kevin Counihan. Slavitt helped HHS recover from its disastrous launch of healthcare.gov, repairing the site so that it could function normally and allow the program as a whole to work. Counihan was in charge of Connecticut’s exchange, which was among the best performers anywhere in the country.

This year, healthcare.gov and the state exchanges all need to raise their game. While 8 million people signed up for private health plans via Obamacare last year, an estimated 8 million people eligible for subsidized private plan coverage remain uninsured. This isn’t a surprise. Experts, like those at the Congressional Budget Office, have always predicted it would take several years before the law reached all of the people it could. Even so, it's hard not to be disturbed by polls showing widespread ignorance of how the Affordable Care Act works and how to take advantage of it. A recent McKinsey & Co. survey  found that most respondents who remained uninsured were eligible for subsidies but didn't know it.

Amazingly, that included people who had shopped on the exchanges. Somehow, hundreds of thousands of people managed to use healthcare.gov or one of the state versions, go through part or all of the shopping experience, yet never realize that—thanks to federal tax credits—they could pay a fraction of the sticker price for whatever plan they chose.

This is obviously a big problem, one Burwell and her team have said they recognize. In recent interviews, they have singled out improving the consumer experience on healthcare.gov as one of their top priorities. But what exactly would that mean? It’s not an easy question to answer, because there are two obvious ways to improve healthcare.gov—and doing one will inevitably make the one harder.


The first challenge is making healthcare.gov simpler. To that end, job one is giving visitors a quick idea how much they're likely to pay.  Last year, one of the first questions that applicants faced was “Do you intend to file taxes for 2014?” Many low-income people who don’t pay taxes clicked “no” on this, unaware that they were effectively disqualifying themselves for subsidized coverage. Kate Kozeniewski, who as a certified ACA “navigator” and program coordinator at Resources for Human Development, a national agency based in Philadelphia, helped thousands of people sign up for insurance, said that “If I could ask anything in the world, it would be that they somehow indicate that if you click 'no' you're not going to get any subsidies.” That's because the subsidies come through the tax system, and, as Kozeniewski explained, “You have to file taxes to get a tax credit.″

Another, better publicized example of the old site’s complexity was its lack at launch-time of a simple shop-around feature—an option where a user could answer just a handful of questions, including annual household income,  and quickly get a complete listing of available plans, priced with the subsidy incorporated. Not only did this absence confuse users; it also contributed to the traffic problems that crippled healthcare.gov in its early weeks. HHS eventually did add a shop-around feature, but the site didn't clearly steer visitors to it as a first step. “People didn't get onto that on their own,” Kozeniewski told me. “It wasn't in the most prominent place on website, and at first it wasn't available at all.”


That many or most visitors to healthcare.gov never found the shop-around doesn't surprise Ning Liang, co-founder of e-broker HealthSherpa. HealthSherpa got a lot of attention in November 2013 when, with healthcare.gov still dysfunctional, it created a separate website that let people see the prices of ACA plans, estimated subsidy included. Within the course of a few months, HealthSherpa got certified as an online broker and worked with HHS to develop its own dedicated interface on healthcare.gov. Since February, it has been enrolling people in subsidized ACA plans.

Liang notes that healthcare.gov has to be many things to many people. It's a portal to small business plans, private individual health plans, dental plans, Medicaid, and CHIP.  The biggest challenge going forward is “to integrate the site so it doesn't feel like four different companies' worth of projects—so people can experience it as one website.”

HealthSherpa strives to provide that unity, and to put visitors on a fast track to signup. You can't spend a minute on the site without knowing how much you'll pay for available plans (if you report your income accurately). The home page is dominated by a prompt to enter your zip code. When you respond you see price quotes immediately—which change before your eyes as you fill in your age, additional household members and income. The order in which the plans are listed changes when you estimate how heavy a user of medical services you're likely to be.

When a user clicks “buy” on any plan description she is transferred to HealthSherpa's dedicated interface on healthcare.gov—and unlike on Healthcare.gov, the information she has entered so far comes with her. HealthSherpa keeps streamlining that interface, and claims to have reduced signup time to 3-5 minutes for solo buyers and 10-15 minutes for buyers of family plans.

The interface also helps users avoid the kind of errors that Kate Kozeniewski helped clients correct. If an applicant applies for a subsidy and says he does not intend to file a tax return, for example, he is prompted either to state an intent to file or uncheck the “apply for subsidy” box.


But simplicity can’t be the only goal for improving healthcare.gov. An ideal website would also help users to make an optimal choice. That’s not easy to do. Buying insurance is an incredibly complex decision, even for experienced consumers. Somebody who wants to make the smartest choice has to take into account all kinds of information about the plan—monthly premiums, out-of-pocket costs, provider networks, drug formularies, and so on.  Many people aren’t familiar with any of these terms.  Most probably are not familiar with all of them.

The authors of recent University of Pennsylvania study, which tracked the experience of well-educated young adults on healthcare.gov, developed a set of recommendations in response to participants' complaints that key information was not spelled out up front or delivered in context.

Pop-up boxes could offer definitions of key terms, like premium or formulary, where they appear. The site could have clear warnings about key caveats consumers might miss—like the fact that Cost Sharing Reductions, which lower out-of-pocket costs for lower income buyers, are available only with silver plans.

The problem—as you may have guessed by now—is that the imperative to provide more information is hard to reconcile with the imperative to simplify the process. Liang, who says he has spoken to thousands of users over the phone, is somewhat wary of providing too much information, at least at first glance. An alumnus of the product analytics team at Twitter, Liang said that Penn study findings evoke a creative tension “between user interface researchers, who talk to users, and the analytics team, which just measures everything.  We try to strike a balance: talking to thousands of people about their needs and incorporating what we learn that way, versus analytics, measuring what percentage of users drops off at every step in shopping process.”

I put the streamline-vs.-inform question to Charlene Wong, the lead researcher in the Penn study. She notes that HealthSherpa does meet one of the study's core recommendations, in that it “uses the type of decision support tools that the young adults in our study were asking for”—chiefly by asking their expected level of medical use. Wong further notes that HealthSherpa's 'Total Annual Cost' quotes also meet a need highlighted by study participants. Still, HealthSherpa does not provide all of the information that the Penn study identified as essential. For example, it does not highlight the fact that free preventative services are included in all plans—not just the more expensive ones.

Liang acknowledges that the desire for in-context definitions and explanations is “natural for an educated person.” Going forward, he'd like to “progressively offer more options to power users.”  He foresees inviting users to “enter as much information as they have an appetite for”—and rewarding them with results more tailored to their needs, via algorithms not yet created.


In the end, there may be no perfect solution. But perhaps there doesn’t have to be. As Tom Baker, a co-author of the Penn study, points out, the Netherlands today has a system that looks a lot like what Obamacare is supposed to be, with people shopping around for subsidized private insurance. There, however, the exchanges are private—though “the government is involved on the back end.” 

In the U.S., too, could healthcare.gov eventually fade into the background as private exchanges compete to offer the easiest experience, or the most information-rich, or algorithms that best base recommendations on a shopper's past medical usage patterns? The ACA's core premise is that competition among private insurers will control costs and improve healthcare delivery. Perhaps competition among brokers will prove as beneficial—and the market will figure out how to get essential information to all comers.