A couple of weeks ago I talked to Stacey Broadwell on the TechRecruit podcast and we ended up talking a lot about the lifecyle of building software, and ultimately how frequently software projects fail.
It's no secret in the industry that software is hard to make and projects often fail. Whether that number is 68%, 90% or 71% that end up being "unsuccessful" is up for debate but the number is probably well over half.
So if you're a recruiter it's good to keep in mind that when you are talking to hiring managers they are likely working on a project that is already behind schedule and overbudget and they need help the day before yesterday.
I thought it would be interesting to make a list of some major failures that happened in the last decade since we are closing out 2019 and starting on 2020 this week!
A project that was initiated in 2010 to automate and replace Canada's government payroll system. Originally planned to be a Peoplesoft implementation by IBM (off the shelf software), it was plagued by problems that are still ongoing.
It had "failed to properly pay nearly half of Canada's workforce of public servants, representing 153,000 people. The report added that the system, whose original 2009 budget was $309-million, had already cost taxpayers $954-million and could rise to $2.2 billion by 2023 in unplanned costs."
It was estimated that the government spent over $840 million on this online insurance exchange rollout, but due to a myriad of issues fewer than 1% of users were able to enroll in insurance during the first week of operation.
UHIP - Unified Health Infrastructure Project was supposed to be a new public assistance program for the state of Rhode Island. Initiated in 2016 and estimated to cost $119 million, the project has not completed yet and estimated to cost a total of $487 million to fix the current issues.
The US Department of Defense is in the process of replacing it's Electronic Health Record system (EHR) at a cost of $4.3 billion dollars but according to recent memos it is:
"neither operationally effective nor operationally suitable,"
Surprisingly, this is a new system meant to replace a $2 billion dollar system that was only rolled out in 2004. The new system is expected to be complete in 2023, but if the latest reports are any indication that won't happen.
Interestingly the VA (Department of Veterans Affairs) is also working on a new $10 billion dollar EHR system which sounds doomed to failure. Why the federal government needs so many separate EHR systems is beyond me.
German grocery store chain Lidl apparently developed a new inventory management system with SAP at a cost of $500 million Euro only to completely scrap the project in 2018 after rolling it out in 2015.
Another Electronic Health Record system - this time for the coast guard. It started development in 2010 and was supposed to cost $14 million, eventually the documented cost was $67 million before the project was completely scrapped. In 2016 the legacy system was retired and now the Coast Guard has reverted to using paper records.
National Grid contracted Wipro to build a new ERP system at a cost $140 million, but eventually spent an additional $600 million and two years fixing the system. This was just one part of a larger $393 million program to modernize aspects of National Grid's IT system according to this article.
I moved from Ohio to Minnesota a couple of years ago and when I went to renew my license I was in for a surprise. In Ohio this was a 20 minute process and your license could be issued immediately. In Minnesota they were in the middle of a rollout of a new vehicle license system.
I had to wait 3 months for my car to be re-titled and 2 months for my new license to arrive in the mail. In the meantime they invalidated my current license (stamped holes through it?) and gave me a slip of yellow paper I had to present along with my mangled license for months every time I wanted to grab a beer.
I was told by BMV workers to "call in a month or two to make sure they didn't lose the title". The project apparently cost the state $90 million before they ended up calling in an outside contractor to bail them out.
It seems that government IT projects fail at higher rates than those in the private sector (according to one article only 6.4% of government projects with a budget over $10 million are "successful") and some of that may be between the mis-alignment between incentives at that level. But, failures in the private sector are also widespread (as I cited in the opening paragraphs).
Building software is hard. It's not like building a house where everything is pretty much established and there are basic rules and plans to follow. Often times when you are building software because of unique data structures it would be comparable re-inventing the process of plumbing and wiring electricity every time you build a building.
Software projects often have unique requirements, situations and constraints. Even implementing off-the-shelf software in the context of an existing organization can be plagued with difficulties.