There’s an old joke about a farmer who loans his mule to a friend. The mule will work hard if treated well, he explains. All you have to do is speak gently and the mule will do what you need it to do.

His friend tries it out, but the mule won’t do anything – it just sits in its stall taking up space and food. When the owner comes back his friend complains that the mule wouldn’t work at all.

Puzzled, the owner picks up a two by four and starts whacking the mule about the head and shoulders.

“I thought all you had to do is to speak gently,” his friend said to him.

“Well sure, but first you have to get its attention,” explained the farmer.

The joke isn’t all that funny, but the punch line can be used in a wide variety of circumstances. It probably explains, for example, the exorbitant estimates bandied about for the Total Cost of Ownership (TCO) of a personal computer. The promoters of these ridiculous estimates are just trying to get your attention.

This is the last of our three-part series on the subject. The column two weeks ago showed that TCO asks the wrong question — TCO adds together part, but not all, of three independent statistics — costs associated with improvements to personal productivity and effectiveness, with improvements in communications, and with the automation of company core processes (that is, development and deployment of production application systems).

Last week’s column focused on the statistic you should care about the most: the fixed overhead costs (Personal Computer Overhead Costs, or PCOC) associated with LAN-attached personal computers. PCOC, it appears, comes to about $3,000 per year.

We’re going to wrap things up this week by looking at the number you may have thought TCO measured: The cost of personal computing. Personal computing is the term we’ll use to cover the use of word processors, electronic spreadsheets, personal information managers and stuff like that — software designed to enhance personal productivity and effectiveness.

It’s tempting to calculate a total cost for this category, but it would be a mistake for two reasons. First, the benefits, while huge, defy quantification. (The proof: PCs have completely transformed the workplace — not one PC-enabled job looks remotely like what equivalent employees did 15 years ago.) Comparing quantitative costs with qualitative benefits can tie your brain into knots.

Just as important, these are variable costs — they go up with usage. Your goal when dealing with variable costs should be to reduce unit cost, not total cost. And the cost we’re interested in isn’t the cost of personal computing itself — it’s the cost of the work supported by personal computing (since that’s where the benefit comes in). Let’s walk through an example.

Figure an average employee gets one day each year of formal training, another full day of support, and loses two full days each year just figuring out how to do stuff. Including the employee’s time and that of support staff at a standard rate of $40/hour (a reasonable fully loaded cost for a $50,000 per year employee), that comes to about $1,600 per year. Allocate a third of the PCOC cost to this category of benefit (the other two thirds goes to the other uses of a PC) and you come to a total annual cost of $2,600 per employee for improved personal productivity and effectiveness.

Let’s figure about half of the employee’s total work — about 1,000 hours per year — is improved through personal computing. That makes the unit cost of personal computing about $2.60 per hour. Expressed as a percentage it comes to an overhead cost of 6.5% of the work affected.

Here’s a wild guess: personal computing leads to improvements in productivity and effectiveness that vastly exceed 6.5%. Reduce your support costs? Sure — so long as it has no deleterious impact on the employees you support.

One of my favorite advertising slogans is “Don’t Blame Desinex”. It’s incredibly useful – no matter what problem someone is whining about, from problems with Windows/95 networking to crop failure, I can respond “Don’t Blame Desinex” and be pretty sure the conversation will change direction.

In poring through the awesome array of expenses toted up in the Gartner Group’s Total Cost of Ownership (TCO) model I have the same reaction: Don’t Blame Desinex. For that matter, don’t hang these costs on the personal computer, either.

Last week we saw that TCO calculations answer the wrong question, and even the right answer to the wrong question will mislead you. TCO answers the wrong question because it addresses cost instead of value, and because the PC has multiple uses with very different value equations.

That means we should be focusing on the costs … and benefits … of the various ways companies use PCs. We identified three last week: increased personal productivity and effectiveness; improved communication; and streamlined core company processes.

It’s certainly worth knowing the cost of increasing personal productivity and effectiveness, so long as we view our goal as being more effective at achieving even more benefits. It’s just as valid to understand what we’re paying to improve communications, looking for opportunities to keep on improving it while reducing the cost of doing so. As for company core processes – that translates to getting better at developing and deploying new, primarily client/server applications. Always a good idea.

But toting up all these costs into a total cost of ownership? You’ll end up like the sales manager who reduced the cost of travel in his sales force. He succeeded — travel costs dropped 25%. Of course, sales dropped 50% at the same time, but he got his travel costs down.

There is a legitimate issue hiding in all the confusion about TCO, and that’s the Personal Computer Overhead Cost (we’ll call it PCOC). It costs real money to provision a desktop with a personal computer, and you may want to know the amount.

Rather than quibble with the Gartner numbers I’ll use them, inflated as they are. Bottom line: PCOC comes to just over $3,000 per year. That includes all PC and LAN capital and administrative costs, plus the cost of supporting network operating systems.

PCOC excludes all costs that index with benefits, like training and support. It also excludes “end-user operations” which GG defines as “non-job-related PC activities that are necessary due to the presence of PCs.” Why? End-user operations include both costs unique to PCs (time wasted rebooting when the PC freezes up) and costs indexed to benefits – for example the cost of managing files, which is much lower on the PC than in a filing cabinet. The benefits probably outweigh the costs by a large multiple, but let’s be conservative and just net it all to zero.

PCOC also ignores all support and applications costs. The cost of buying, installing and supporting applications should be allocated to personal productivity and effectiveness, improving communication, and streamlining company core processes – not PCOC. More on this subject next week.

Now, lest I seem unbalanced … let me rephrase that: Lest my perspective appears to lack balance … I need to say something nice about the Gartner Group. (I hate this part!)

Once you get past the TCO cost analysis, which has received all of the publicity, the Gartner Group makes some reasonable recommendations on how to reduce costs without affecting benefits. Recommendations like centralizing administration. Using remote-control software to reduce the number of trips to the desktop. Installing automated tools to manage your networks better. Designing more effective training and support programs. And so on.

These are good things to do. End-user support is too often an afterthought in IS, and you probably need to pay more attention to it. Institute a continuous improvement program. Re-think it completely. Especially, make it a career opportunity.

PCOC adds about 5% to employee overhead. While knocking a few points off that number won’t change the world, it will certainly improve both your budget and image.