Some interesting facts about payroll project
- They had about 80,000 employees to pay
- Because of the multiple awards and hideous complexity of the awards (think "hand washing allowance"), there are approximately 24,000 different possible combination of items on a pay slip.
- Most of the mentioned $450M (or $1.2 B projected cost) is for the clerical support to process pays, not the IT cost
While the IT project was a horrendous screw-up with bad design choices (e.g. award processing is done in SAP AND in the front end WorkBrain scheduling/time keeping software) equally bad was the management of change, business process.
They did many things wrong
- changed the organisation, payroll system, scheduling system, time sheeting system all at once
- went live without adequate testing
- wrote a poor contract with IBM, allowing them to eventually bill twice the original contract value.
- did not change the major weakness in their business process, which was to process pays before they received time sheets, meaning that they were forever having to go back and adjust pays, sometimes up to 2 years in arrears
- decided to integrate WorkBrain and SAP even though it had never been done before
However, while I am no fan of SAP, this implementation does produce the correct result if given the correct input data. The huge cost is mostly in the administration to provide the correct time info. (So it's clearly a bad design, but it does work with herculean effort).
Not everyone came out badly though. HP sold a couple of extra Superdomes. I'm sure the salesman was happy.
No need to find a conspiracy here, when mere incompetence will suffice.