I’ve never seen a large scale programming project, and I haven’t seen any project of this scale (has anyone?) succeed at the rollout date. I haven’t seen a contracting company say that no, they can’t meet the deadline demanded by the customer. What happens is a POS is delivered, months or years after the deadline, and then being fixed over the subsequent years. Ask Apple about their implementation of the SAP system, the millions thrown away before they gave up.
Yes, I have.
Both were SAP rollouts. One was for a company I worked for and the implementation process took 7 years total. But hardly a hitch.
The other was ExxonMobile’s implementation. Which I studied and continue to cite as a best practice scenario. Took less than two years, but involved keeping SAP vanilla and being utterly draconian about ANY customization. Where a customization was requested, they really forces the business owners to change themselves to fir the system.
There was a third success at this scale as well, mostly custom build with pen source. 10 year incremental effort, still underway. But still successful.
There were three utter failures as well. All big bang style implementations. One was a disaster of an ERP (PeopleSoft) implementation that helped destroy a 16,000 person company. Another straight custom dev for a very complex business model and the third a hybrid Buy/custom build, also for a very complex model, that tried to integrate 20+ COTS products.
It all really boils down to Gall’s Law,
I’ve worked on large scale websites with healthcare plan selection. Also time absence management.
Each of the 50 states have different regulations/excuses and these can change every year. There are legal problems if you do NOT have the site operational (and correct) in time for healthcare plan selections (I am referring to annual enrollment PRIOR to Obamacare/HUACA).
As with NASA, “FAILURE IS NOT AN OPTION”.
We met the dates and were correct.
Sounds like there was no budget for testing (site navigation, load testing, etc.) or else upper management didn’t want to hear bad news, or else forced “sign off” signatures knowing they were going public with a faulty system.