How oil firms can control runaway enterprise-wide software systems

Stuart Diamond
Vice-President
CSC Index Inc.
Dallas

• Management Perspective: Enterprise software [84,586 bytes]

Few topics can be as controversial within oil and gas companies as enterprise-wide computer software systems.

For all their promise and potential, the installation of these systems has been marked by severe cost and time overruns, internal strife and disillusionment, and acrimony between energy companies and their software vendors (OGJ, Oct. 7, 1996, p. 31).

For example, one major oil company budgeted $250 million for installation of its enterprise-wide system. By last fall, it had spent $350 million.

Today, the company has spent about $450 million and, 3 years into the project, is still struggling to get the technology up and running.

Another large energy firm that had planned to spend $20-30 million has already spent more than $100 million and is behind schedule.

What is truly amazing is that, in both cases, with all this money being spent, there is no clear line of sight to any business benefits. Many senior executives in the oil and gas industry have a similar story, filled with complaints about the scarcity of business results and the armies of onsite consultants whose bills keep mounting.

Short of expectations

The consensus is that despite the allure of seamless, integrated, enterprise-wide computing, these systems cost more, take longer, and deliver less than expected. There are hidden costs, too. Companies have been unable to pursue other strategic, growth opportunities because valuable, limited resources are immersed in systems installations.

What is especially distressing about companies being bogged down in large-scale systems implementation projects is that the industry is experiencing a boom. If there was ever a time to pursue an externally focused growth agenda, it is now.

Data matrix

So it's timely to ask: What is this beast called an enterprise-wide system? How can a company gain control of its installation and implementation and not get eaten by the beast?

Simply put, enterprise software is an integrated system that links together information across an organization, regardless of department, location, or function.

The goal is that the company uses one software package with a number of integrated modules, rather than multiple, conflicting systems, each with its own language and data formats.

With this integrated model, a user in materials management could see the same data on his or her screen as a colleague in finance or sales.

Data is entered only once for the entire enterprise, instead of continuously recaptured and reentered.

An executive who wants to see inventory levels throughout the company could double-click on a single command and view real-time, accurate data.

The leading player in the enterprise application industry is Germany-based SAP. Their product, R/3, was introduced in 1992 and dominates with market share of 40-50%. SAP faces increasing competition from such companies as Oracle, Baan, PeopleSoft, and Computer Associates.

This article is not intended to evaluate SAP or any other vendor but to discuss how an oil and gas firm can get business results and complete the installation in a timely and cost-efficient manner.

Five issues

From the experience of CSC Index, there are five major issues that management must emphasize in order to gain control of a runaway enterprise-wide system project.

The first is that an enterprise-wide system project is, in fact, not a systems project. If business benefits are to be achieved, much more than new systems will be required.

Processes need to be changed. Jobs, skills, and organizations must be aligned with the new processes and underlying technologies.

Rewards and incentives must be re-thought to induce new behaviors.

Simultaneously, the culture will have to be examined to determine how best to achieve and sustain the anticipated changes and benefits within the organization.

This type of broad, wide-reaching change demands superb project management skills and strong, decisive leadership. Most importantly, this leader should come from the operations side of the business.

Typically, most enterprise-wide projects wander off course due to a lack of effective business leadership and accountability.

Strain seen

The second issue is the lack of appreciation for the strain on an organization that occurs with a change of this magnitude.

Many companies seriously underestimate how tough it is to overcome pockets of internal resistance when people see themselves losing accountability and control to an enterprise-wide system.

These pockets of resistance arise because enterprise-wide systems represent a return to a more centralized, "standard" management of systems, in which a single integrated package is the information backbone for the entire enterprise.

Not since IBM was in its mainframe heyday in the 1970s has the centralized model been in corporate practice. In the intervening 25 years, different functions and departments have installed a plethora of standards and systems, depending on personal preferences, vendor relationships, or just plain serendipity.

Chevron Corp., for example, found prior to beginning one of the world's largest SAP installations that its finance department alone had 200 or so financial systems and processes.

That's why it is critical that leadership must come from a strong individual, supported by a small, cross-functional core team (8-10 people) that can impose standardization across the company.

The project manager should be a senior business person with clout and technological credibility-as well as the authority and ability to cut off debate when necessary.

On this last point, CSC Index recently saw an energy company in which leadership of an SAP installation was ineffective and vacillating. Even after an issue seemed to be nailed down, it would arise in a fresh controversy a few months later.

For 18 months, the project team debated the issue of information hierarchy-the basic flow and sequence of information into and around the company-without resolving what should have been a basic question.

The challenge of strong leadership is especially relevant to the oil and gas industry, which traditionally grants much autonomy to people in the field.

For instance, a refinery manager is used to running the show and will have many plausible reasons why his or her information systems, formats, and protocols must be different than, say, the chemical division's.

However, the value of an enterprise-wide system depends on reducing customization and eliminating different configurations.

Making this level of standardization happen is not a technological issue-it is a leadership issue.

If your company's installation is running over time and cost, the first place to make changes is in the composition of the core leadership group.

Timing concerns

A third issue is the pace and timing of the installation.

Many companies make the mistake of moving too cautiously and slowly, in the mistaken belief that an incremental roll-out strategy is the best methodology.

Unfortunately, this traditional "waterfall" implementation (so-called because of the way it looks when drawn on a flow chart) doesn't work well for enterprise-wide system projects.

We often tell executives during an enterprise-wide installation, "The less you put in, the more it will cost you."

That's because enterprise-wide systems were built for a "big bang, wall-to-wall" installation. It is extremely difficult and complex to create good interfaces between enterprise-wide and existing "legacy" systems-and that is precisely the reason that so many projects veer out of control.

The more legacy interfaces you need to construct, the more frustration, confusion, and delay you'll get in return.

Rather than moving cautiously, CSC Index recommends installing enterprise-wide systems aggressively in several modules. This approach is a break from the past, when systems were built and added incrementally.

You should limit interfaces with legacy systems to only those that are absolutely essential-and then obliterate the rest of the legacy system and its encrusted years of "spaghetti code." If your project is behind schedule, it is very likely that this is where you are losing valuable time.

Not only is the aggressive pace of the installation important, but so is the sequence of modules installation. SAP likes to implement them "inside-out" (see chart, p. 25). Inventory and financials should be installed first, followed by sales and distribution, material management, and so on to the outer rings.

While this sequence optimizes the organic design of the software, it often is in conflict with the demands for business results. Some managers may want production planning installed first in order to get a quick hit in manufacturing and justify the software's expense.

If a symptom of your installation is that the modules aren't working together, we recommend that you make a mid-course correction and alter the sequence in which modules are installed, in order to complement the natural architecture of the software.

Training approaches

The fourth point has to do with training. In case after case, executives say that they have dramatically underestimated the training costs associated with an enterprise-wide system installation.

A rule of thumb is that only 20% of a project's costs come from buying hardware and software; internal development and deployment consume the rest.

A common trap in training is to chase false economics of scale. For example, one U.S. West Coast energy firm purchased a training program for all 6,000 employees across five business units, on the premise that spreading training costs over the maximum number of users was the most economical approach.

However, what users needed to know was so different from group to group that the single "overview" program had limited relevance.

It was full of generic examples and was inadequate and irrelevant when logistics staffers or finance staffers tried to figure out entry codes or protocols particular to their functions.

It is better to customize training for each user group. While this may increase the initial training outlays, it will produce far superior returns on investment.

One electrical utility has gone even further and broken its thousands of users into market segments, much as a consumer marketer would do for a product roll-out.

By designing a matrix that matches user needs against skill sets, the utility has been able to pinpoint what kind of training is needed for, say, a field engineering group as compared with an employee benefits group.

The company has identified 30 distinct segments and is working with its vendor to co-develop these training modules.

In another instance, Chevron began its training with a traditional stand-up classroom, with a cost per user of $4,000. If it had continued with this approach, it would have spent at least $40 million by the time it rolled out the program to all 10,000 users.

But Chevron chose an alternative. It developed, internally, a computer-based training program that has completely replaced the classroom curriculum and will cut per-user costs in half.

Chevron has created specific curricula for 80 different functions and plans to add 40 more. We think more large companies will adopt this innovative approach.

Limitations

Finally, users should recognize what the software can do and what it cannot-and deal decisively with the limitations.

In the case of R/3, it is excellent for core manufacturing and production applications in the oil and gas industry. This is not surprising, given that SAP's origins were selling to huge German chemical and pharmaceutical corporations.

However, there are aspects of the energy business that R/3 is ill equipped to handle, particularly in arcane financial applications.

One of these weak areas is leasehold management. This can be an enormously complicated topic for oil companies, such as when fractional interests need to be paid royalties from a well site.

Users need to develop their own customized software for applications such as this, or work with the vendor to cope with the product's limitations. Otherwise, delays and overruns will arise as frustrated users try their own makeshift solutions.

Thinking about the installation of enterprise-wide systems as a technology project is the quickest road to ruin.

These are large-scale, institutional change efforts involving hundreds of people and hundreds of millions of dollars.

Only the best business manager, who truly understands all the dimensions of change required (processes, jobs, skills, organization, management systems, technology, and culture) and who has the strong, demonstrable support of the CEO, can pull this off successfully.

All the change dimensions are critical to keeping the reins taut on an enterprise-wide software system.