$90 million in maintenance per year for healthcare.gov

The Washington Post says that the programmers who built healthcare.gov are being fired (story). They are being replaced by Accenture, at $90 million per year for maintenance.

This does raise some questions…

  • if the site works, how is it possible to spend $90 million each year on maintenance?
  • if the site doesn’t work, absent spectacular incompetence by the old team, how will it be possible for a team of new programmers that has never looked at the source code to make progress at a more rapid rate, at least in the short run?
  • how much is actually going to be spent if you could the salaries of federal government workers in addition to these contractors? And, rather than building this site, would it have been cheaper to fly all of the Americans who have successfully signed up to non-Medicaid plans via healthcare.gov over to France every time one of those folks needed a real medical intervention? (paying market/private rates in France for the services consumed)
  • should we buy Accenture stock as a way of getting some of our tax dollars back? This chart shows that building Web sites for hundreds of millions of dollars is actually pretty profitable.

[Note that Google was funded with about $25 million that lasted five years, albeit supplemented by some ad revenue (source). Thus if the healthcare.gov project goes perfectly in 2014 it will be consuming cash at 18 times the rate that the world’s most rapidly growing company did.]

12 thoughts on “$90 million in maintenance per year for healthcare.gov

  1. Phil,

    Some time ago, you posted regarding a US citizen going to France and paying privately for a heart procedure that was much less expensive than what they would’ve had to pay for the same procedure here (in the US).
    Will you please provide a link to that post? Thank you.

  2. Mark: I remember the story clearly but I couldn’t find the story either. After much digging with Google, I finally found the story you’re talking about in a comment on http://philip.greenspun.com/blog/2010/02/21/coming-out-ahead-on-health-insurance/ . Here’s the relevant text: “A friend’s parent had a heart operation in Los Angeles at a cost of over $50,000. It was unsuccessful. The family went over to France and had the operation repeated by the surgeon who had invented it. This time it was successful. They paid out of pocket for a private room in a beautiful modern hospital and a lot of excellent French food. The cost was about $7,000.”

    [Separately, it was easier to find with Bing than Google.]

    [Note: this posting moved up by moderator.]

  3. The comparison with the “$25M cost” to create Google isn’t valid. In particular, the phrase “supplemented by some ad revenue” is a gross understatement. Google was pulling in just under $1B/year by 2003 (source). This completely dwarfs the VC investment.

    Healthcare.gov has no comparable revenue stream (other than federal tax dollars, of course!).

  4. I’ve built software for CMS. I don’t know anything about this deal in particular, but I know from my experience that the level of documentation necessary to satisfy CMS that we weren’t screwing them over made everything take a lot longer. I managed the software testers on that project, who spent, gutchecking here, half of their time doing things related to auditability, not related to testing. The full story:


  5. J. Peterson: It is true that Google had a lot of revenue but they were also the fastest growing company in history and growth tends to consume cash, e.g., for necessary capital investments. But the revenue point that you make about Google is an inspiring one…

    Wouldn’t it be more logical to allow multiple private companies to set up health care exchanges and then pay each a fee for every customer who manages to sign up? If you gave me access to the XML feeds from the various government and insurance company servers and promised me $20 for each person who bought insurance through philip.greenspun.com I would start programming tomorrow!

    The Federal Aviation Administration is nobody’s model of efficiency, but they did something like this for pilot weather briefings and flight plan filings. There are multiple private companies, e.g., http://www.duat.com and http://www.duats.com, each of which gets paid every time a pilot interacts with the web site rather than calling up an FAA-paid human (formerly government employees, now farmed out to a contractor). If one of the sites were to be down for any reason, a pilot can simply use the other. Both companies have an incentive to keep the sites working so that they can keep getting paid.

  6. Multiple privates companies HAVE set up exchanges. The company I work for being one of those companies. However, a friend of mine who works for the company has questioned the wisdom of our company doing so given that so many other health insurance companies, all the ones bigger than our quite small company, have opted out from doing so.

    I don’t know the particular details of why we would set up an exchange or why others would opt out. Unlike my friend, I figured it would be wise of us to do so.

  7. > how will it be possible for a team of new programmers that has never looked
    > at the source code to make progress at a more rapid rate, at least in the short
    > run?

    Most of them will probably be laid off by the old contractor and then hired by the new contractor. That’s what happened to me when my employer lost their federal contract.

  8. Google programmers are paid somewhat more than Accenture programmers (although probably less than Accenture “partners”), and are many times more productive. Google programmers also don’t have to deal with monstrosities like HL7.

  9. Why did they go for another large management firm instead of hiring it out to a proven web development firm to handle the project? It makes absolutely no sense to me. Which politician or committee was responsible for this hire?

  10. does the $90 million include power and equipment fees for operation, or is it just paying for managers and programmers? I could see electricity + facilities costing $30 Mil. or so. 30 programmers + managers $10 million. Compliance: $5 mil. where does the other half go?


    sort of related: I had to see a doctor in Paris during a holiday (Bastille Day) and was told the best way for me to do it was to go to the emergency room. The waiting room was full but I saw a doctor in about an hour. At checkout the patient coordinator told me I was not part of the French medical system and I would need to pay in cash. Two orderlies came out and ‘helped’ me find the ATM machine. Cost? Euro 140 for the facility fee + Euro 60 for doctor fee + Euro 12.27 for the medicine. This was 5 years ago. How much would that have cost in the US?

  11. I don’t understand why they didn’t build this on top of something like Salesforce. Base CRM database just works. Scaling is free. Easy to build a custom UI. Cost would be way way way lower…(under $10M?).

    Here’s a cool demo called “Healthcare.luv” that built on top of Salesforce at this year’s Dreamforce conference…the team built it in under a week…

  12. Might have been better to tie at least part of Accenture payments to utilization and then, with skin in the same, payouts might be more justified and also produce a measurable result. Would have been interesting to watch how the focus changed round the conference room table too!

Comments are closed.