Shaping The Future:

When a company budgets $1 million to develop a new software system it is, in fact, committing to spend more than $4 million over the next five years. Each dollar spent on systems development generates, on average, 20 cents for operations and 40 cents for maintenance.¹ Thus, the $1million expenditure automatically generates a follow-on cost of $600,000 a year to support the initial investment. Development is in many ways the loss leader for maintenance.

Software development is the main discretionary expenditure in IT planning. It is also obviously the key element in using IT for business innovation. To cut back on software investment is to risk cutting back on business development. But software is extremely expensive when measured in terms of full lifecycle costs. To ensure that the true costs of software are factored into business justification and that the capital provides real returns requires major shifts in the management process for IT.

For each of the past three decades, IT budgets-for hardware, software, and telecommunications - have grown at about 15 percent per year, which is far greater than the rate of business growth. It will be impossible to sustain this rate in the 1990s. IT expenditures of a billion dollars a year by leading firms have brought this area to the forefront of capital planning. In financial services, for example, IT operations costs alone are the third or fourth largest expenditure, behind employee costs, real estate, and interest. A growing number of senior executives are worried that IT costs may be out of control.

IT now amounts to about half the incremental investment for large firms. It has become a contender for scarce business capital and not just for tomorrow's expense budget. There is little evidence that the investment is producing adequate benefits. What is more, there are no reliable methods for measuring the business value of IT.

Senior executives are caught in a worrisome double bind: ever greater commitments to IT investment are being driven by competitive necessity and discouraged by escalating costs and uncertain benefits. Put another way: economically, companies cannot afford to increase capital spending on IT; competitively, they cannot afford not to do so. The economics of information capital is firmly on the top management agenda, and corporate managers are clamoring for help.

There are three issues in managing the economics of information capital: managing costs, managing benefits, and managing risk exposure. That these have generally been handled poorly is attributable almost entirely to the historical emphasis on treating IT as overhead, which is managed through budgets, cost allocations, and cost justification. It is a very naive way of dealing with a complex economic good.

In managing the benefits, making the business case, and assessing payoff in relation to anchor measures that relate to operational business indicators, management must accept that there exists no set of accounting ratios or simple formulas that show the business value of IT. In this respect, IT is like R&D; it must be justified as a longer-term investment in the future.

Finally, managing the risks associated with IT expenditures involves market concept risk, technology risk, implementation risk, economic risk, and organizational risk.

Managing IT Costs

Managing costs should begin with the creation of an IT asset balance sheet and commitment of time, attention, and management resources appropriate to the amount of the capital asset (most managers will be extremely surprised by its size). Management must then count all the IT costs, many of which will be hidden. Development compounds future operations and maintenance costs, and organizational, support, and infrastructure costs are frequently overlooked or obscured by an accounting system that expenses IT as overhead.

Creating the IT Balance Sheet

Creating an IT asset balance sheet that capitalizes all IT equipment, software, and data resources currently in use is one way to jolt senior management. The balance sheet is not a legal financial statement-accountants point out that software cannot be capitalized for tax purposes-but instead a management report. Management should know what the annual IT expense is, but it very likely does not know how much capital is tied up in IT resources. It is probably far more than management realizes. And it also often comes as a surprise that there is no way to tell from the accounting system; because software development is expensed, it is often impossible to accurately calculate the value of software in use.

A balance sheet for a large bank that spends about $200 million a year on IT is shown in Figure 6-1. The bank has long been aware of the most obvious capital item, hardware and equipment, but has overlooked the capital tied up in personal computers, departmental telecommunications, and small unit-cost items that together added up to several million dollars.

Top management had no idea that the bank had spent close to $500 million to create the software currently in use, nor that it was an information factory sitting on data that had cost well over $1 billion to create (this figure includes relevant fractions of the salaries of staff in the bank's operations and back-office units). IT assets added up to over $2 billion.

The exercise drew top management's attention to several points that should have been obvious, but that are almost always overlooked when IT is expensed. The main point was that IT assets were under-managed. In this firm, corporate finance has five times the number of senior managers as Information Services, which makes do with a senior vice president and three subordinate VPs. The amount of time spent on IT at top management meetings was measured in minutes per year. Partly as a result of understanding the level of capital IT represents, the top management now commits days per year to IT.

This is not a call to treat IT as "special," but merely to provide a management complement appropriate to the level and business importance of the asset base. Many executives view the trend of creating chief information officers (CIOs) as a ploy by IS to boost salaries. If IT were treated like any other business unit that manages $2 billion of assets, top management would surely elevate the level, number, and quality of senior managers and planners assigned to it.

The $490 million in software and $1.2 billion in data currently in use in the bank are already paid for. Can it be more effectively used and reused? Can new products be derived from it? Can the software be repackaged?

Of course they can and should. The organization now sees many opportunities to reuse its data and to realize direct cost savings by, for example, merging decentralized facilities into the corporate IT utility. Additionally, powerful new mechanisms have been established to ensure that corporate staff, senior management, and line managers spend more time and attention not on IT as such, but on the policies and priorities most relevant to the biggest single element of capital in its fixed-asset base.

Before the firm built its IT asset balance sheet, no one had noticed any of this. Even IS was surprised by the size of the software and data capital asset.

Counting All the Costs

When IS was a central corporate staff function, its costs, if they could not be controlled, at least were relatively easy to identify. Today, with almost half of firms' IT expenditures going for personal computers, office technology, and end-user computing, even identification has become much harder. Many companies do not know what they are spending on IT. End-user computing and telecommunications expenditures are scattered across business units' budgets and many organizational costs are not even tracked. Datamation estimated in 1987 that 40 percent of the costs of IT were not part of the Information Services function's budget.

How can firms make rational decisions about IT when they do not know its true costs? The starting point for managing the economics of information capital is to understand that IT costs are of two types: supplier costs (i.e., the costs incurred by the Information Services function) and user costs (the growing portion of firms' total IT expenditures that is being shifted out of central corporate IS units into the business).

SUPPLIER COSTS.

Corporate IS groups are a form of internal supplier, analogous to a utility, that operate corporate data centers and telecommunications facilities and provide services such as systems development and information management. Their costs are usually charged back to the business users they serve through an allocation mechanism equivalent to transfer pricing in manufacturing. Computer hardware and operating systems and telecommunications equipment comprise the power plant for information services. Frequent changes in the components of this power plant, driven by rapid technological change, require continual retraining of the people who operate and maintain it.

An increasing proportion of the operations function is committed to ensuring the reliability, security, and availability of key business services that depend on the IT base. When an airline's reservation system is down, so too is its business. When a bank's cash management system is not secure, it opens its vaults to thieves, eavesdroppers, and passersby. When a dealer order-entry system is slowed by overloaded computing or telecommunications facilities, service, cash flow, and customer image are degraded. Operations skills and resources are essential to the on-line business enterprise and a vital element in the technical infrastructure. They are also not cheap.

The business services delivered via IT hardware are realized in applications systems. Development, and subsequent operation and maintenance, of these systems is also people-dependent. Whereas systems development costs are discretionary, subject to increase or decrease by management at short notice, operations and maintenance costs are not. Inasmuch as maintenance can amount to one to three times development cost, today's development budget sets the IT budget for the next several years. The head of Information Services in a major bank calculates that at any given time his group has close to one hundred ongoing maintenance/enhancement/upgrade projects for the bank's checking account services, which drastically limits his unit's ability to deliver new systems that business groups urgently need. "They see us as unresponsive and not meeting their needs," he explains, "but it's old systems that dominate our budget and schedules, and I have no way of changing that for at least the next five years".

For every dollar of initial development expenditure for a large system, operations costs will average 20 cents per year and maintenance 40 cents per year. More staff will generally be working on maintaining and enhancing old systems than on building new ones. At any given time, only about 10 percent of a corporate Information Services unit's staff is developing new systems; maintenance occupies 50 percent or more of its scarce human resources.²

The cost of support is growing rapidly as IS groups shift from building systems to supporting business units. The support facilities for a $5,000 personal computer, for example, typically amount to at least $8,000 per year according to the Gartner Group, a leading research organization that tracks IT costs and trends.³ Much of that will be paid for by the business unit, but it is a new burden on IS units, and one that they must take on if they are to be a responsive service unit. New expenses such as these are added to the specialized technical staff needed to manage the IS group's telecommunications, data management, and computing infrastructures. All these resources provide education, training, and internal consulting throughout the organization.

On the supplier side, only new development is truly a variable cost. Data centers, telecommunications networks, and operations and maintenance include variable components, but it is hard to cut overall costs except at the margin.

USER COSTS.

User costs vary widely, depending on company policy regarding IS cost allocation and recovery. Some of these costs, being hidden, are unbudgeted.

Allocations and charge-back of central IS unit-supplier costs obviously become user costs. To these are added direct acquisition and usage costs. Personal computers are a frequent element of the former. Direct usage costs, which largely depend on transaction volumes, can grow rapidly and be hard to control. For example, a single personal computer accessing an outside information service may not need special justification or budget. But with IT, supply often creates demand. The success of the initial system may stimulate rapid expansion. Two hundred personal computers accessing the outside information service will shift the cost dramatically from overhead to capital, especially if it is decided to link the computers in order to share information, messages, and software.

There is an obvious and immense difference between initial small-scale, ad hoc computer use and wider departmental capability. To manage such investment case by case (piecemeal) is to likely overlook how a series of $5,000 personal computer purchases can become a million-dollar capital plan just for hardware acquisition. And the hidden costs of support and telecommunications can dwarf this expenditure.

Central IS units have long struggled to manage compounded costs driven by software development. Business units now face similar stresses.

In the days of central mainframes and IS monopolies, business units had no need to develop technical and operations staff. But with the advent of personal computers, distributed development, office technology, departmental operations, and end-user computing, these units began to incur many of the costs of distributed information management previously borne by the central Information Services group.

The apparently low costs of distributed hardware, such as personal computers, and of software packages, such as spreadsheets and word processing applications, has belied the often substantial costs of support. Organizational costs-including management time, the learning curve for staff, education, and other costs of making the transition from old work systems to new-are rarely budgeted and can make official development costs look like pennies.

IS/Business-Unit Perspectives on IT Costs

Business units often do not see the scale or value of central IS infrastructures. They do see the money they are charged for these resources, whether directly for usage or through an allocation for their part of the shared base. Similarly, many business-unit managers view as an expense, rather than as the essential service asset that they are, the legions of systems programmers, application developers, analysts, operators, and project managers who develop the software and information management asset that is so critical to competitive positioning and a basic element of efficient business operations.

For their part, IS managers have no ready formula for allocating infrastructure costs fairly. Consider this representative sample of the questions they face. Should the first users of a telecommunications service carry the full burden, even though the marginal cost is low? Who should pay for the development of a customer master data base, an infinitely reusable resource that will eventually be of value to many business units? Where outside services are cheaper, should business units be allowed to use them and thereby reduce the customer base for the shared corporate resource? With supplier costs being increasingly fixed investments that may not pay off for many years and development only a fraction of the full lifecycle cost, how can the true costs of a service be determined?

Supplier costs, which tend to be fixed and long term, include (1) the obvious and visible elements of hardware and equipment and the buildings to house them in, and (2) the people needed to operate, maintain, and support that physical plant. The former typically amounts to about 40 percent of corporate IS budgets, the latter 60 percent.

Senior management, if it takes any notice at all, is likely to be unhelpful. Top management response to the fixed-cost nature of supplier costs is usually to demand that IS budgets be cut. In late 1989, for example, top management in one of the world's twenty largest banks, viewing IS costs as too high, decided to cut the firm's IT budget from $520 million to $400 million. IS expenditures on maintenance, operations, and ongoing development represented $460 million of the $520 million, and as support for core business services such as credit card processing, automated teller machines, corporate electronic banking services, and basic processing systems could not easily be cut, IS's only practical choice was to cut corners on testing, operations, and security, which meant cutting quality, service, and reliability. The bank subsequently fired the hapless head of IS; his successor is likely gone by now as well.

It is easy for managers to demand that IT costs be brought under control, but it is reasonable for them to do so only if they acquaint themselves with the origins of those costs. They should understand, for example, that real expenditures grow much faster than software development budgets. We have seen that today's systems development costs compound tomorrow's committed budget for operations and maintenance.

Senior management should also recognize that infrastructure costs dominate applications costs. A firm that moves toward basing most or even much of its operations, product delivery, and customer service on its IT base has to make greater and greater investments in the infrastructures that support the practical and cost-efficient creation of individual applications. There is no cheap and easy way to build the global telecommunications networks and large-scale data and network management systems that constitute the corporate IT infrastructure.

Most senior business managers, and many IS managers as well, are unaware of these basic realities. Tradition has led them to view IT as an annual expenditure. In good times, the IS manager makes the case for an increase of X percent, or adds up all the approved new software development requests from the business units and factors in aggregate operations costs. In bad times, senior management demands either a decrease in the rate of growth or even an absolute cut in the budget (a very new phenomenon for IS).

Both approaches ignore the cost dynamics of IT. Figure 6-2 shows how IT costs are compounded, assuming that every dollar of development generates 20 cents of operations and 40 cents of maintenance.

Following strategy 1, which calls for maintaining a level total budget given the reality of compounding IT costs, would all but eliminate development within five years, virtually assuring that the firm would suffer competitively either from skimping on maintenance and operations and thus on customer service, quality, and reliability, or on development and thus on business innovation. Maintaining or growing development, as called for by strategies 2-4, is not possible without budget growth. A mere 10 percent growth in development, as in strategy 3, would increase the overall IT budget by a factor of four in just five years.

Figure 6-2 oversimplifies the details of IT budgeting but not the reality of compounding costs and of systems development as the agent of compounding. In light of these figures, the 15 percent rate of growth in IT expenditures across the U.S. economy in the 1970s and 1980s no longer seems so large. It is clearly not enough even to keep systems development level to ensure adequate operations and maintenance. The natural rate of growth is closer to 20 percent. For firms that are aggressively spending on IT for effective competitive positioning, the rate is often closer to 30 percent. This latter rate obviously can be sustained for only a few years unless the benefit flow keeps pace with the costs. But unless a firm commits to a growth of at least 15-20 percent per year, it is effectively cutting back its IT capability unless it can dramatically improve development productivity, a challenge that the IT field has not yet made more than token progress in meeting. Software development remains a complex and slow craft.

Infrastructure investments compound the problem of managing IT costs by adding to the fixed-cost base without providing direct benefits. The development expenditures shown in Figure 6-2 can be assumed to have been justified by some business case showing economic returns. If the case is realistic and the costs fully stated-two extremely big "ifs"-then the firm will benefit and there is a rational basis for funding the development. Far more often, the critical investments are not for specific business applications but for infrastructures such as an international telecommunications system, customer information data base, or network management facility. Infrastructures are all cost; benefits come indirectly from the business applications they make practical.

Infrastructures can rarely be cost justified. As more and more development, and to a lesser extent operations, is distributed to business units, the central IS unit's main responsibility is for central systems and shared infrastructure resources, especially the corporate "backbone" network.

There are many opportunities to reduce infrastructure costs by consolidating separate facilities, especially in the area of international telecommunications. One firm, for example, cut its total telecommunications costs by 30 percent by linking separate services to a shared high-speed, low-unit-cost fiber optic transmission facility, centralizing network management, and routing international traffic so as to avoid high-cost countries. It should, of course, have done this years ago, but no one in the firm-literally-knew either how large the costs or telecommunications had become or where they were incurred; expenditures on telex, fax, international phone calls, outside data communications services, and internal facilities were scattered across several hundred budgets. It took several months of study to identify them.

The commonsense development of IT infrastructures is analogous to the commonsense upgrading of the infrastructures of congested, inefficient, and disruption-prone airports such as Heathrow, Kennedy, Frankfurt, and San Francisco, which is being totally blocked by politics, costs, and problems of location and transition. Many people complain that something should be done but are unwilling to pay any personal costs of money, effort, and inconvenience.

With costs growing far faster than business profits, evidence of payoff limited or absent, and small sanctioned increases in annual budgets generating long-term expenses, it is no wonder that corporate IS units are under siege. Frustrated and worried, senior management asks, "How on earth, with the costs of personal computers dropping by the week and the price/performance of the microelectronics industry continuing to improve by 30 percent per year, can our IS spending be growing by 10-20 percent per year with no increase in development or business productivity?"

The answer - "That's the way it is" - is not a justification for uncontrolled IS budget growth but a challenge to business managers to face up to the reality of IT costs and work to ensure that the returns from IT investments exceed those costs.

What Is Management to Do?

Not to allocate central IT costs is to provide a free resource, eliminating any sensible basis for rationing, avoiding waste, or controlling costs. Corporate IS clearly must charge its costs out, but finding a rational and acceptable way of doing so is difficult. Allocating costs on the basis of full cost recovery affords users little opportunity to manage their own costs. They simply incur an annual charge to their budget and may pay extra for increases in infrastructure investments from which they receive no immediate benefit. Additionally, the cost of costing-of tracking usage, billing, auditing, and so forth-can be enormous.⁴

Allocations give business units ample reason to be displeased with corporate IS. They see large charges at the same time they hear about rapidly falling costs of technology. They naturally wish to exploit the latter and avoid the former. So the cry goes up: "Break up corporate IS!" Or outsource it.

Someone needs to voice the counter cry: "No! Rethink the management process for IT." There are no magic, plug-in solutions to the problem of managing IT costs, only commonsense directions for improving, if not resolving, the situation. Management must first ensure that the full lifecycle costs of IT, including all relevant hidden costs, are identified and included in any analysis of IT options.

Further, management must develop an asset, instead of a budget, view of these costs. This means funding the corporate infrastructure separately from business applications. The infrastructure is a shared corporate resource that enables a range of both foreseeable and unpredictable uses. Telecommunications networks, shared data base management resources, and security and network management utilities are all part of the infrastructure and should be funded by top corporate or divisional management as a long-range capital investment justified by corporate policy requirements.

Supplier costs can be recovered through a "quasi-profit center," eliminating cost-based allocations. This follows from the principle of funding IT infrastructures separately and allowing them to recover costs in a manner analogous to a power utility, whose managers spend capital and petition state regulators to allow them to set rates that recover the investment over a period of years at a "fair" rate of return. The utility sets prices; it does not allocate costs. The difference is key. Imagine going into McDonald's and being charged $227 for a Big Mac, the manager explaining that this is the fully allocated costs of the hamburger plus $1; he has just had the kitchen remodeled and must pass on the cost. IS has traditionally been required to allocate costs in this way. The quasi-profit center approach allows IS to set prices sensibly, for example, to price anew electronic mail service below cost to stimulate demand and make sure that early users do not have to pay the heavy initial installation and operations costs.

IS should not, except in special circumstances, become a real profit center. A true profit center, such as a separate IT services company, easily creates a business contradiction. If its services are outstanding it should be totally devoted to the needs of the business that owns it instead of being diverted to serve outside firms. Yet if it is expected to maximize its profits, it has little reason to support the firm's IT needs except where it can command a premium price. Only if it is mediocre should it be allowed to focus on external markets, in which case it makes no business sense for the firm to own it.

Many firms that have adopted the quasi-profit center approach base the pricing policy on a market basket of outside providers' prices. The manager of the telecommunications network, for instance, may be expected to offer services at 80 percent of the average of five specific vendors. Some will allow users to buy certain services outside, limiting them only in cases that affect overall corporate costs, security, and resource sharing. The internal IS organization is then part of a regulated free market. One major additional advantage of quasi-profit center pricing is that it reduces the cost of costing, tracking the details of costs, and establishing often complex allocation formulas.

To protect systems development, management must ensure that development proposals are screened by business managers, costs are fully addressed, benefits are fully and systematically assessed, and that the manager making the claim on the capital is held accountable for delivering the benefits. No IS manager can be expected to set the business priorities for a system and identify system benefits and guarantee their delivery.

Finally, management should encourage the reuse of existing resources. Much of the opportunity to use IT competitively derives from repackaging existing information, and many of the most promising long-term approaches to easing the software development bottleneck relate to reusing program code.

For management to take these initiatives, responsibility for IT application planning must be embedded in the business units. Information Services' responsibilities should lie in infrastructure planning at the corporate level, with senior business management setting policy and Information Services creating the needed architecture.

Pushing responsibility for IT application planning into the business end represents a major organizational shift for many firms. It directly links the issue of managing costs with managing benefits and reveals a question that remains unasked, never mind unanswered, in most firms: "Who is accountable for IT benefits?"

Summing Up: Making Sure to Get the Hidden Costs

Figure 6-3 lists apparent versus hidden costs for two components of IT expenditure, software development and purchased software packages. It shows how large a fraction of the full cost is incurred outside the activities of the programmers who write code or outside the purchase price of a package.

Software development costs vary by size and type of project, type of technology, quality of staff and project management, and number of development tools and techniques. Figure 6-3 presents the cost components of software for typical types of business applications that would constitute the bulk of an Information Services unit's portfolio. It shows how many of them are largely hidden and ignored in the business justification and budget. Few projects, for instance, budget twice as much for education as for programming the computer code.

Improving software development costs and productivity is a major problem for every large organization. It has been so since the 1960s and is likely to remain so through the 1990s despite efforts by vendors, computer scientists, researchers, and consultants. NASA, for example, carried out a project to develop "perfect" software, software that is close to error-free. It succeeded, at a cost of $1,000 per line of program code. The cost for a commercial organization is typically between $20 and $50 per line. NASA reported that it found no high-tech tools or new techniques that ensured the level of quality it sought; its approach was the tried and true "inspect, test, retest".

Computer-assisted software engineering (CASE) is a promising approach to improving development productivity. It uses computer tools to help computer staff, through a mix of design disciplines, analytic tools, and design and document management aids. CASE is part of the emerging organizational and technical infrastructure of IS. It is, however, roughly at the same stage as personal computers were a decade ago. It also demands major shifts in IS professionals' work, skills, attitudes, and effort; they often show as much "resistance to change" as their "users" did when IS brought them new systems.

Packages and related "fourth-generation" development languages are often seen as one solution to the problem of software development productivity. Packages, which are general-purpose software that can be adapted to meet specific needs, substitute for the assignment of a team of corporate IS analysts and programmers to custom-build a system. Packaged software may not satisfy all the requirements of its users. IT people refer to this as the 80-20 rule; users get 80 percent of what they want for 20 percent of the cost. To get the other 20 percent they will likely have to spend an additional 80 percent.

The same is true for fourth-generation languages (4GLs), special-purpose development tools that can dramatically reduce the cost of developing small-scale and ad hoc applications. Personal computer spreadsheet software and many data base management, graphics, electronic publishing, and accounting systems are variants on packages and "4GLs".

The price of packages alone should not be the basis for making the decision to build or buy. A package may be a bargain at twice the cost of building a customized system if it reduces the maintenance burden. Conversely, it may be far more expensive in real terms at half the cost of building a system in-house when support and operations are included.

Almost every new increase in software productivity has been wildly over-hyped when it first appeared. Fourth-generation languages were promised as the path to applications development without programmers, just as office automation was supposed to create a paperless office of the future in just a year. CASE was expected to generate automatic code. All these tools offer value but demand time, change, learning, and effort. There is no instant bargain.

That is also true for personal computers, which are widely seen as the best bargain in IT. But the term "personal computer" is in many ways misleading. The telephone is "personal," too, but it depends on a complex infrastructure, and telephone charges include an allocation of the costs for this infrastructure. To the extent that personal computer users want to connect their machines to others and access information stored on larger computers, they will need an equivalent infrastructure. In calculating the real costs of personal computer use in organizations several years ago, the Gartner Group likened the obvious cost to that for the telephone handset, and the hidden costs to those of the telephone company.

Gartner calculated that the hidden additional cost for a personal computer with software and printer, bought for just over $4,000, is close to $20,000.⁵

The organizational IT infrastructure requires an entirely different kind of business justification than do specific business applications and incidental hardware. The business case for support infrastructures must consider trade-offs between longer-term corporate needs and current application-specific needs. It involves paying a premium to invest in infrastructures that ensure continued flexibility and quality of service and that help the firm avoid being pushed into a position of competitive disadvantage. The decision to pay that premium is, like any other decision to make a strategic capital investment, part of the senior business-management policy agenda. The cost is likely to amount to as much as 10 percent of the corporate IT budget; in many large firms, these infrastructures amount to as much as 30 percent.

It generally takes about seven years to bring to completion any major business innovation that depends on building a comprehensive IT infrastructure such as a global telecommunications network or new generation of data base management systems.⁶ This means that the components of the IT expenditure for infrastructures cannot be allocated as a direct cost but must be treated as a corporate capital investment. Though the cost of application development and operations can be charged directly to a business unit, to make the first users of an infrastructure bear the cost is to block innovation.