In the description of my CIO App Development group (on LinkedIn) I make a rather extreme-sounding claim, which is that I specialize in saving companies hundreds of thousands to millions of dollars on each of their app developments. I imagine that you might view this with some well-justified skepticism. In my upcoming articles I’ll be gradually disclosing my track record and my methods of actually doing this, but I don’t expect you to be patient enough to wait for those. So in this, my first article of this CIO series, I will give you some idea of my capabilities by giving you an immediate way to start saving thousands of dollars every month on your current app development. The method is very simple, but I’ve never seen it used in any corporate environment, not even once.

All of my clients, to the best of my recollection, have had a policy of distributing essentially the same computer to all of their employees and consultants. Whenever new people arrive or equipment is replaced, someone on the capital-equipment side of the house decides upon some particular desktop or laptop model, and a batch of that model is purchased and distributed. Everyone, from the administrative assistants to the programmers, gets the same machine. I call this machine the “secretary computer” because it is well suited to running typical applications like Word and Excel.

Unfortunately for the company and for the programmers, in order to operate efficiently the programmer’s tools require many times more hardware than the secretary’s computer contains. You won’t get much helpful advice from Microsoft in this matter – Microsoft recommends about the same “minimum” system requirements for its popular Visual Studio programming tool as for the secretary’s Microsoft Office. But in practice I’ve never had a problem running Microsoft Office on a “secretary computer”, whereas I’ve had plenty of problems running Visual Studio and its various add-ons on such machines.

My estimate is that I lose five minutes out of every hour by using a $500 machine instead of a more suitable $1,500 developer-appropriate machine. I’ll estimate the loaded salary of your typical programmer (whether salaried employee or hourly consultant) at about $100 per hour, and your typical programmer’s workday conservatively at eight hours per day. Based on my time-loss estimate, 8.3% of all the money you’re paying to your entire development staff is being lost due to the poor performance of their computers. Since you’re spending $4,000 per week per programmer, that’s $333 per week you’re tossing out the window for every programmer you employ.

This means that every three weeks you lose the entire amount of money that you saved by buying the programmer a cheaper computer (i.e. $1,000) by paying that programmer to wait for the machine. So you're spending about $17,000 per year in order to save the initial $1,000 on the machine’s cost. Multiply that by the number of programmers you employ, of course. But that’s just the loss due to hourly salary. Since you’re losing 8.3% of your entire programming staff’s entire productivity all day every day, you can be fairly certain that your projects are running as much as 11% later than possible just due to this problem alone. That’s almost six week lost per year.

What is the opportunity cost to you of being unnecessarily six weeks later on all your software projects at the end of every year than you need to be? What if your competitors aren’t?

Your people can tell you what sort of machines they’ll need in order to get the job done. At this writing it would be something like a quad-core 64-bit Windows-7 desktop with an I7 processor, preferably equipped with a Raptor disk. But don’t take this advice a year from now, since the requirements change constantly. If your capital equipment people don’t like the price of this machine, remind them that you’ll be saving the entire cost difference every three weeks after the first month you install it.

Perhaps you’re more up-to-date than most companies and you’re already running your company on virtual machines. What then? The same principle applies. Each one of your programmers needs about four times the computing resources of the typical employee. If you’re smart, you’ll give it to them.