[bunderbunder]

That may be the case if you're just doing in-house software.

If you're writing software that others will be running, the calculation's a bit more complicated. You've got to consider your customers, and how much a performance change impacts their purchasing decision, and use that to estimate your revenue impact, and then compare that to the cost of developer time. And what comes out the other end of that is that developer time is often much cheaper than dissatisfied customers.

The case is similar on the Web, where tiny minuscule barely noticeable changes in application responsiveness can have a huge impact on conversion rate. Or worse yet, make the difference between your servers falling over or ticking along smoothly when that oh-so-important Cyber Monday traffic surge comes along.