Tuesday 8 December 2009

Who knows?

Sometimes you're going to find that no-one in the organisation knows (really knows) what the rules are for some things. Of course it may be better to keep quiet about this rather than make a big fuss, but you'll need a strategy for dealing with it.

How you do this will depend on the client - you have to find a way that suits them, their politics and so on.

Wednesday 25 November 2009

Old Payroll Systems

In older payroll systems (and indeed many other systems) storage was at a premium so they couldn't do some things we can do easily now. One of these is manage transactions historically.

All the older systems tend to treat data items as having only a single dimension - only the current value is stored and anything that went before is lost (or hard to retrieve). This makes it easier to deal with, but ultimately less capable.

These days we can use the history of a range of data items to do more complex payroll calculations.

Monday 23 November 2009

Payroll system Parallel Runs

When I started doing this kind of work in the early 1990s it was customary to run any new payroll system alongside the old one for a month or two to check the results were the same. It seemed a pretty sensible way to evaluate the system configuration.

Just have a think about how everything's changed since the early 1990s - Televisions, Radios, Cars, Computers - you can probably see where this is going.

Payroll systems are much more capable these days and they interact with a huge range of other systems too.

Now have a think about the world of work during that period (if you're old enough !) Consolidations, buyouts, reorganisations, redundancies and management pressures to do more with less have meant that people are often firefighting rather than thinking strategically.

Enter the parallel run - and imagine that the organisation has decided to spend a little on a new payroll system (in order to realise cost savings of course). Everything's been set up and along the way we've figured out that a few things aren't being calculated properly in the current system anyway. We clearly aren't going to design the same errors into the new system - which means they aren't going to agree.

Which is all a long way around to say that parallel runs are a little outdated these days - aside from the fact that for a "true" parallel requires people to do most of their work twice for the duration of the parallel - not an easy sell in a modern workplace (and I'm not sure it ever was).