Starting in late October, Zients declared that the website, which is supposed to enable residents of 36 states to sign up for insurance through President Obama's health care law, would be functional for a “vast majority” of users by the end of November.
The reason the date was significant is that consumers typically finalize health insurance decisions after Thanksgiving and the deadline for signing up for insurance that becomes effective at the start of 2014 was Dec. 15 (the administration has since delayed this deadline to Dec. 23). This date is especially pressing for those who have had their insurance cancelled as a result of requirements imposed by the health care law.
On Sunday, the Department of Health and Human Services said in a report and on a conference call that it had met its goal for fixing the website.
In reality, there’s no reason to believe that this is true.
HHS is claiming to have made more than 400 software fixes, raised system capacity to 50,000 simultaneous users, increased speed, reduced the error rate to less than one percent, and increased the stability of the system.
But it’s hard for outside observers to declare this a success based on this blitz of numbers, because at the start of the repair process, HHS never provided the public with specific goalposts against which the site’s progress should be judged.
HHS never said, for instance, “In the next month, our goal is to complete these 400 software fixes” or “the site should only be down x percent of the time.”
What information HHS did provide its new report isn't very impressive if the comparison is with a typical commercial website rather than against the basket case that was healthcare.gov in October.
For instance, an HHS chart – which Zients boasted about – shows system uptime now at 95.1 percent (excluding scheduled maintenance), which compares to 42.9 percent a month ago. But, the industry standard is for websites to be available for users 99.9 percent of the time. Anything below that is considered a failure and 95.1 percent is a disaster.
A 2012 study by web monitoring firm Panopta that looked at the performance of 130 major retailers' websites from January to August 2012 found that the lowest uptime rate was 99.34 percent.
Another study by web performance firm Pingdom that looked at retail websites during the 2011 holiday shopping season, found that nearly half of the websites (such as Amazon and eBay) were up 100 percent of the time. The lowest performing was Foot Locker, which was at 98.573 percent.
A 95.1 percent uptime means that over the course of a year, a website would be down for about 18 days. Alternatively, imagine what a disaster it would be for sales if, during the holiday shopping season, Amazon’s website were down for about a day and a half, excluding scheduled maintenance.
Another goal that the administration had offered was to make it possible for 80 percent of those intending to sign up for insurance to get through the entire application process.
On the call, Centers for Medicare and Medicaid spokeswoman Julie Bataille was a bit vague about whether that goal had been met.
Five weeks ago, Bataille said, just 30 percent of users could get through the process.
“Today, we’re now more in the zone of about 80 percent of users being able to do that same process successfully,” she said.
What does “in the zone of about” mean? Furthermore, she emphasized that not all Americans will feel comfortable signing up by the website, and could seek help by phone or in person. That suggests that the vague 80 percent number includes all forms of signing up for insurance, not just through the website.
And once again, comparing it to the private sector, imagine if one out of every five people trying to buy airplane tickets on Orbitz could not do so.
Additionally, the progress report makes no mention of the back end problems that have plagued the system. Specifically, insurers have been reporting a high error rate on the so-called 834 forms, which transmit enrollment data on individuals from the federal government to the insurers. These errors have involved mixing up spouses and children, duplicating some enrollments, inadvertently canceling others.
If unfixed, insurers won’t have accurate information on who is supposed to have enrolled in insurance by Jan. 1 and won’t be able to smoothly process a large number of payments, which are essential to insurance taking effect.
“We continue to work with issuers in particular with issues around the 834 transaction forms,” Bataille said on the conference call when asked about the remaining back end issues. “A number of fixes that went in place this weekend in particular will significantly address some of the highest priority things we know were a particular concern with those transaction forms.”
She added, “We need to see increased volumes of traffic and transactions, which we anticipate seeing over the course of the weekdays, to make sure that we get information from issuers and see the success rates that they are having with the fixes put in place this week.”
Translation: We won’t even know the extent of the problem until insurers start getting flooded with applicants, which hasn't happened yet due to the problems with the front end of the website.
Given how much of a black box the development of the federal health care web system has been, ultimately whether one believes that the site has been fixed depends on whether one feels confident giving the benefit of the doubt to the Obama administration.
But after the events of the last few months, why would anybody trust administration officials?
These are the same people who, days before the launch of the website, even as it was clear it wouldn’t be ready, were telling Americans that they were on track to have the website up and running on Oct. 1, promising a top-flight consumer experience.
As detailed in the Sunday New York Times, early on, the administration tried to blame overwhelming demand for the website problems, even though it was clear the problems were more fundamental.
Though the administration touted a major “tech surge,” it was actually a mere public relations stunt.
“The announcement conjured Images of an army of software engineers descending on the project,” the Times account reads. “In fact, the surge centered on about a half-dozen people who had taken leave from various technology companies to join the effort."
Then there's the evolving story of the 834 enrollment transaction errors. Back in October, Bataille first claimed on a conference call that the problems were “isolated.” After shocked insurance industry sources pushed back against this account, explaining that they were significant, Zients then said resolving the errors was at the top of the punch list of planned fixes.
Yet the Times reports that, “At the outset, the team had made what officials call a very intentional decision to focus their repair effort on making healthcare.gov work better for consumers. That has meant putting off some ‘back-end’ fixes for insurers, who use the site to receive applications and bill the government for subsidy payments.”
So even though there has been progress made on healthcare.gov two months after its launch, the site is still far from delivering on its original promise to make shopping online for health insurance as easy as buying plane tickets on Expedia or Orbitz. And the less visible back end problems may still create significant obstacles for those hoping to be covered in the new year.