With CIO budgets under pressure, there\u2019s an almost irresistible allure to the notion of chargeback, wherein IT costs are literally charged back to the user departments on whose behalf they were incurred. In this way, the IT budget stops being an ever-expanding money pit, endlessly consuming hardware, software and bandwidth?and instead the tab is picked up by the people who actually use the stuff. At a stroke, budget justifications become a cinch. "Of course we need this level of expenditure," the CIO can assure the board. "If we didn\u2019t, the users wouldn\u2019t be ponying up the money for it." Even better, the shock of actually having to pay for what they consume transforms users from hungry conventioneers at an all-you-can-eat buffet into disciplined dieters.No wonder, then, that chargeback\u2019s proponents are so enthusiastic about its merits. Bob Odenheimer, for example, believes chargeback not only shifts the burden of IT expenditure justification onto the newly cost-conscious users but also injects a note of rigor into the enterprise\u2019s business model. "If you look at a small business, its IT costs are reflected in its profitability," says Odenheimer, senior vice president of IT operations and telecommunications at managed behavioral health services provider Magellan Health Services in Columbia, Md. "It buys the gear?and has to pay for it. Why should it be any different at a large business?"It\u2019s more than just a question of even-handedness. When IT expenses are retained within an overall IT budget, rather than charged back to departmental cost centers, companies can\u2019t get a handle on the real costs of winning new business, complains Odenheimer. "If it helps their unit\u2019s profitability, there\u2019s nothing to stop them from taking as much IT as they can get?even if the actual IT cost [to the business as a whole] is higher than the revenue it produces [on the unit level]," he says. Chargeback, in Odenheimer\u2019s view, prevents that by forcing users to reconcile within their own budget the marginal cost and marginal profitability of new business.Bob Svec, president of TSL, a chargeback consultancy in Parsippany, N.J., sees a chargeback-related slant on the age-old centralization versus decentralization debate. Chargeback, asserts Svec, "offers the best of both worlds. You\u2019ve got the efficiencies of centralization, combined with the ability to let people see what they are getting, and how much they are paying for it."Taken together, such arguments have quickly elevated chargeback into so-called best practices territory. Consultants laud its merits, and vendors have sprung up to handle the dollars and cents of charging. But CIOs have found a dark side to chargeback too. It can become a political hot potato and an administrative nightmare. Many CIOs have come up with workarounds, but others find chargeback so problematic that they\u2019ve sworn off it entirely.The essential difficulty with chargeback is a conceptual one. Weighing the pig, critics charge, doesn\u2019t actually make it any heavier. Shouldn\u2019t IT resources be spent on, well, IT, rather than on figuring out who owes what to whom? And isn\u2019t the allocation of IT resources better carried out by, say, a steering committee than by a clerk with a spreadsheet?The Trouble with Chargeback: "What Forest? All We See Are Trees"Richard Scannell, cofounder and vice president of corporate development and strategy at Framingham, Mass.-based consultancy GlassHouse Technologies, observed the use of chargeback while at Motorola, where he spent 12 years as a senior IT manager, ultimately heading up a group of 160 IT people in Chicago. Chargeback at Motorola was based on headcount. That works in theory, says Scannell, but in practice quickly turns sour."If the electric company turned around and said, There\u2019s 2,000 people in your town, and we\u2019re going to divide the town\u2019s usage by 2,000 and charge each of you a two-hundredth of the total, then you\u2019d likely object," he says. "If it doesn\u2019t work with power, why should it work with IT?"In reality, Scannell contends, people don\u2019t consume resources evenly?nor do IT systems. "A lot of data is held in databases, not files, and is accessed by multiple functions and multiple people?how do you even begin to charge back in that scenario?" he says. Then there\u2019s the problem of people moving about the organization. "If I have a lot of files associated with an individual working in human resources, what happens if that individual moves to marketing? Does marketing get charged for human resources\u2019 files?" Because of unanswered questions like these, Scannell chose to do without chargeback at Motorola.Kevin Vitale, president and CEO of Ejasent, a chargeback software vendor in Mountain View, Calif., sees a link between chargeback and poor server utilization. Data centers typically have server utilizations on the order of 10 percent to 15 percent, he asserts, which in normal circumstances would prompt a server rationalization project. "But then, chargeback gets more complicated," he says. With one server per application or per business unit, chargeback is relatively straightforward. Combine applications on a server, however, and the arguments begin over how big a tab each department should pay.It\u2019s precisely such wrangles?or rather, the prospect of them?that influences John Nordin\u2019s decision to hold back from full-blooded chargeback in favor of a less contentious middle ground. "I\u2019ve worked in IT environments where every single IT expense was allocated back out, leaving IT as basically a zero-cost operation," says Nordin, vice president and CIO of specialty metals manufacturer A.M. Castle, headquartered in Franklin Park, Ill. "But it\u2019s way too much work. It\u2019s a never-ending debate over what constitutes a fair share. In the end, it comes down to this: What am I being paid to do? And I think I\u2019m being paid to do more that just debate MIPS and CPU cycles."That\u2019s not an exaggeration. When working for a multibillion-dollar health-care company in the mid to late 1980s, Nordin recalls, "we had an entire department within IT called IT Accounting. And that was all it did: count MIPS." There\u2019s no way, he insists, that he\u2019s ever going to manufacture that much red tape himself.Making Chargeback Work: Simplify, SimplifySo how, then, does Nordin charge back at A.M. Castle?"We don\u2019t try to charge back everything, nor do we avoid chargeback altogether," he says. His golden rule is that he charges only for things that are inarguable. "I charge for nonshared IT expenses where I can put my hand on my heart and say, This is your expense. So it\u2019s things like printers, phones, fax machines?that sort of stuff," he says. Business applications, on the other hand, reside at the center on a number of servers and are not charged out. "We tend to bring that sort of expense up to the top, and call it IT Budget," he says. "We\u2019re running a lean business in a lean industry, and the idea of having people running around pushing numbers just appalls me."Of course, business applications can account for the majority of enterprise IT spending. Some CIOs have figured out how to charge back those costs without getting swamped. B. Lee Jones, CIO of San Jose, Calif.-based Stratex Networks, a manufacturer of wireless transmission equipment, sits down with the company controller at the start of the budgeting process to review chargeback allocations "to make sure that they are reasonable. After that," says Jones, "the system pretty much runs itself."And runs itself fairly efficiently too. Stratex has no one employed in either the finance or IT functions whose primary responsibility is operating the chargeback scheme within the company, says Jones. He estimates that it probably takes no more than a couple of hours a month to work out the allocations.Which sounds reasonable. On the other hand, the company\u2019s IT systems undeniably lend themselves to such a lean and laid-back approach. Jones?one of 2002\u2019s CIO 100 Award honorees?sits atop an IT infrastructure populated by almost every vendor known to man. "All our projects are user-driven, which is why we\u2019ve wound up as a best-of-breed shop," he says. "We\u2019ll always pick the best application for a particular function."The result is considerable simplification of the chargeback process, since each application is run by a single group of users. The PeopleSoft app is charged to the human resources department. Agile? Charged to document control and quality assurance. Oracle? The manufacturing part gets charged to manufacturing operations, the finance part to the finance function. Clarify? That would be the maintenance and repair function. Siebel? Charged back to sales. PC technical support gets charged back based on the number of PCs each function has, while storage is charged back based on each function\u2019s headcount. There is some central overhead, but most of it is charged out, using such straightforward allocation rules as headcount.The underlying philosophy of chargeback at Stratex is simplicity and practicality. "We don\u2019t put any monitoring timers on equipment in order to see how it\u2019s used," Jones says. "We assume that usage will be a normalized distribution both within individual departments and across the enterprise." In other words, over time, the assumption is that any inconsistencies will average out.But will they? While not every organization\u2019s business apps can be divvied up so tidily as Stratex\u2019s, CIOs can still apply simplicity and practicality to chargeback. For example, after Magellan Health Services\u2019 Odenheimer charges back at actual cost any expenses that can clearly be identified within a particular business unit or site, he adds up other expenses such as LAN and desktop support costs and allocates them on a headcount basis?on the assumption that IT support costs will be evenly spread. "As they will be," he insists, "if you use the right desktop tools" to standardize user PCs around an approved configuration.Furthermore, Odenheimer retains a specialist company to analyze all telephone, fax and data line invoices, screen them for errors, and then break them out by business unit or department. "They always find something?always," he says. "It\u2019s an ROI of 300 percent?they save us three times what we\u2019re paying them."Change With?or Without?ChargebackChargeback is no panacea for getting a handle on an enterprise\u2019s IT spending, and there is no perfect chargeback system. Nordin of A.M. Castle has no quarrel with chargeback\u2019s theoretical advantages?just its practical downside. "Would I ever espouse a full chargeback system?" he muses. "Yes, but only if I could get a 100 percent metered technical process in place?which in our environment is unlikely." The difficulty: a mix of mainframe boxes, Unix and NT servers, packaged software, and homegrown systems. "Vendors are looking at the problem from the data center perspective, but that\u2019s only part of it," he says. "In reality, people have phones and BlackBerrys and PDAs as well?so I don\u2019t believe we\u2019re going to get there any time soon."Ultimately, chargeback is meant to prod users into changing their behavior. If a chargeback system fails to accomplish that, it\u2019s a failure, no matter how good the accounting is. Chargeback is wearily familiar to Keith Kaczanowski, vice president of process improvement at Brady Corp. of Milwaukee, a manufacturer of signs, labels and associated printing equipment. "If all you do is reshuffle costs within the company, then chargeback probably isn\u2019t worth it," he says. "But if you think that people will make better decisions by being forced to confront the cost, then there is value. "We used to charge $X per person per year, but found no one changed any decisions," concludes Kaczanowski. "Instead, they just grumbled about a perception of low value. So now we don\u2019t charge out at all." It was, he adds, "pretty much a consensus decision. A lot of our IT expenses turned out to be for enterprisewide activities?servers, WANs and enterprise applications. Taking it back does a better job of matching where the decision is taken to where the costs are."