In our business, I have seen many instances of damaging targets. I once worked in an organisation where it was decreed that we all should be given a target of contributing a certain number of pieces of 'intellectual capital' to the knowledge bases every year. Unsurprisingly the knowledge bases soon filled up with utter cr*p, but hey, we met the targets so the bonuses were paid. To deal with the quality issue, someone came up with a cunning plan - all contributions had to be rated by one's peers, and only those which had a high enough score would count. So a system of 'you rate mine and I'll rate yours' arose, the knowledge bases continued to fill with rubbish - and we continued to get our bonuses, effectively being paid for wasting the companies' money and time. Other instances are paying people a bonus based on bugs fixed, which leads to the deliberate introduction of easy-to-fix bugs and a nice little earner, or rating people by lines of code written, which leads to an orgy of cut-and-pasting.
Caulkin points out
if enough pressure is applied, people will meet targets - even if they destroy the organisation in doing so. As quality guru W Edwards Deming put it: 'What do "targets" accomplish? Nothing. Wrong: their accomplishment is negative.'
At a high enough level of abstraction, targets are fairly benign. It is when they are chosen just because they are measurable the damage really starts. They also usually encourage short term thinking, erode trust and therefore the feeling of personal responsibility for outcomes. Targets are set top-down, in advance - the management equivalent of waterfall development. Human nature being what it is, targets trump thinking - and if we've learned one thing about software development in the last 40 years it is that thinking generally helps the process.
The agile movement tends to avoid numerical targets in favour of more woolly but also more useful aims such as delivering value to your customer. My experience is that when we trust people to do a good job, and remove the obstacles to them doing so, the results are invariably superior than when trying to coerce that behaviour by the use of targets.
