GM’s Matrix Reloads

How an org chart can inspire internal competition, drive process efficiencies and make a business more competitive.

The IT challenges at General Motors have always been huge. The immense scale of the enterprise, the extreme autonomy of the business units (until recently, Buick, Cadillac and Oldsmobile weren’t just brands, they were baronies), the hijacking of the entire IT function by Ross Perot’s EDS in the early 90s — all combined to create one giant hair ball of IT problems.

When Ralph Szygenda became GM’s first corporate CIO in 1996, EDS had just been spun off from GM, but it was still running all of the company’s systems. Consequently, GM had no IT staff of its own. “There weren’t any IT people to speak of; there was no IT leadership,” Szygenda recalls. “How do you transform that?”

His answer, which got quite a bit of publicity at the time, was to build an organisational matrix of IT managers unlike that found in any other company. Szygenda hired five divisional CIOs to correspond roughly to GM’s business divisions: North America; Europe; Asia-Pacific; Latin America, Africa and the Middle East; and finance. At the same time, he hired five process information officers (PIOs) to work horizontally in different specialities across all divisions around the world: product development, supply chain management, production, customer experience and business services (HR, legal and so on). These CIOs and PIOs came on board in 1997 to form the management organisation of GM’s IT, formally known as Information Systems & Services (IS&S).

CIOs and PIOs work from divergent perspectives and have different reporting relationships. Each CIO reports not only to Szygenda but also to business heads; PIOs report to Szygenda alone. IT managers refer to “the matrix” or “the basket weave” to determine their relationship with one another and to explain their occasional clashes (see “How the Matrix Works”, page 40).

This matrix, proven over time, has been a critical part of how IS&S took control of IT spending from EDS. During the past seven years, Szygenda’s team has lowered GM’s IT budget by $US1 billion (25 per cent). Where previously GM used 7000 different information systems, there are now fewer than 3500.

By setting up overlapping, intersecting responsibilities among his direct reports, Szygenda designed the matrix to create internal competition, believing that was how to improve processes. “CIOs are driving efficiency in their world and PIOs are driving efficiency horizontally,” says Cherri Musser, PIO for supply chain operations, who has twice been a CIO within IS&S and who was one of Szygenda’s first hires.

GM’s experience illustrates the impact an organisational structure can have on business. But before you start whipping up internal competition in a GM-style matrix, consider this: The effectiveness of the matrix rests as much on the managers who work within it as on their defined reporting relationships.

The Matrix in Action

The matrix works because it integrates different viewpoints for better, more thorough decisions, says Dan McNicholl, CIO for GM North America. To illustrate his frequent debates with PIOs such as Kirk Gutmann, who heads up global product development and global service delivery, McNicholl gives this example: “I might say: ‘Kirk, what are the next areas I should be focusing on in my product development cycle? What tools should we be standardising on? What architectures should we be using? What do you think are funding priorities?’” McNicholl balances Gutmann’s input with what he hears as a member of the North American Strategy Board, which includes business-side executives from the division. “If Kirk says: ‘I think we ought to focus on A, B and C’, I can say: ‘Well, you know, I’ve heard [the exec committee] talk about A and C a lot, but B doesn’t get a lot of attention, so maybe that isn’t a priority’,” McNicholl explains.

Those who work within the matrix sometimes grumble about it. The matrix “doesn’t come without a price”, says Steve Hanna, PIO for business services and application solutions delivery. “Sometimes it takes longer to get to a decision — we circle around it for a while.” McNicholl, who unlike the PIOs has budgetary responsibility, says: “Every PIO thinks they’re the most important, [but] we can’t afford everything.” CIO-PIO disagreements get aired at the six-and-a-half-hour meetings Szygenda holds with his direct reports every Friday morning in a room high up in Detroit’s Renaissance Centre office towers.

But the price seems worth paying. “It’s important to have creative tension,” says McNicholl. “Most companies don’t have that. They spend too much on IT, or fund he who yells loudest.” Hanna adds: “When you don’t have that cross-pollination [created by CIOs and PIOs], you lose that broader view.”

Indeed, IS&S’s matrix has been so successful that GM’s business side has adopted it — the ultimate compliment. In early 1998, CEO Rick Waggoner set up global process leaders in the business units who parallel IS&S’s PIOs — a case of IT leading the business side to alignment.

GM’s matrixes arguably have led the company to huge gains in manufacturing quality and productivity in recent years, although the public’s perception of GM vehicles remains sceptical. In May, JD Power and Associates released its “2003 Initial Quality Study”, viewed as a benchmark of new car quality. Although GM as a company rated only average, Cadillac, Buick and Mercury were among the top six makes, ahead of Toyota and Honda. (The top 11, in order, were Lexus, Cadillac, Infiniti, Acura, Buick, Mercury, Porsche, BMW, Toyota, Jaguar and Honda.) Szygenda argues that “GM is so close to Toyota and Honda, it’s a nonissue. We had a product problem in 1996. We don’t any more.”

Many factors have contributed, but IT’s role in GM’s turnaround shouldn’t be underrated, says Kevin Prouty, automotive research director for GartnerG2, which analyses vertical industry. “The IT backbone is what drives processes” at auto companies, he says. “The technology and business processes are so intertwined, you have to have someone in charge of processes.” That someone is Szygenda’s invention, the PIO.

The Matrix Evolves

Now Szygenda is adding another layer to the IS&S matrix in an effort to gain even greater efficiencies. One executive calls it “a third dimension”. Last April, Szygenda announced the creation of three virtual “factories” within IS&S:

• An IS factory, pulling together all IT executives and staff who maintain and run hardware and software, including software development and e-business systems.
• An operations factory, tying together IT people who are primarily concerned with manufacturing and supply chain operations.
• An enterprise business management factory, for internal processes such as contract management, purchasing and HR.

Each factory formed its own steering committee to look for ways to re-engineer its processes. Although GM’s operations have improved in the past few years, there’s still a fair amount of fat, says McNicholl. “In one year, I was asked by PIOs to fund four different lessons-learned databases. It’s good to do that once, but three?” Responsibility for reining in eager database developers now falls to the new IS factory.

The factory overlay may sound like a confusing twist to an already complex structure, but it’s what GM needs, says Maryann Goebel, IS&S’s chief strategy officer, who now also heads up the enterprise business management factory. “As crazy as it sounds, [the factory concept] does help simplify things, to think of each activity falling within one of three factories,” she says. “The familiarity we [already] have with the matrix makes us ready for the next step.”

The Matrix Replicated?

Given the impact the matrix has had at one of the world’s largest companies, why haven’t more CIOs adopted a similar organisational arrangement? Szygenda has an answer: “The matrix bothers a lot of people.” It’s too complicated, too redundant. GM’s size and the overgrowth of IT systems prior to 1996 made it right for them but, says Szygenda, “You can’t take the same model across companies.”

Two big auto parts makers, Johnson Controls and Delphi, have adopted a form of GM’s matrix, but Delphi spun off from GM, and Johnson Controls is a big GM supplier. Prouty thinks the real reason matrix-style structures haven’t spread is that only a few executives can tolerate and make creative use of the tension they foster.

“The thing that’s unique at GM is that the people who work there [are able to] balance between collaboration and competitiveness,” Prouty says.

Szygenda says that ability was precisely what he was looking for in 1996 when he interviewed 300 candidates. “A lot of the 300 were great CIOs or consultants, but I didn’t think they could team” with others. Ultimately, he says, the effectiveness of any organisational structure comes down to the personal qualities of the executives, managers and employees who work within it.

STRATEGIC DIRECTIONS

By Susan Cramm

It’s the only way to ensure project success

MIT’S Peter Weill says that “an effective IT governance structure is the single most important predictor of getting value from IT” (from his paper Don’t Just Lead, Govern: Implementing Effective IT Governance, co-authored with Richard Woodham). If that’s true, how do we explain the fact that many organisations have not set up governance in a way that promotes value?

That question was recently posed to me by one of my clients. Even though she is responsible for her company’s current governance structure, she sees it for what it really is: nothing more than a well-run project approval process, with ROI calculations, business sponsor testimonials and meetings attended by all the usual suspects. At no time has the governing body revisited an approved project in light of new information (also known as “change requests”), terminated a project or assessed the company’s track record for value realisation. Sound familiar?

Let’s be honest. At many companies, governance should be called “govern-once”. Although there is broad agreement about what governance should be — “the decision rights and accountability framework to encourage the desirable behaviour in the use of IT”, as Weill puts it — in reality what passes for governance is often a one-dimensional, checklist-based and attendance-based effort focused solely on project prioritisation and approval.

In The Information Paradox, John Thorp writes about full-cycle governance. CIOs must ask themselves four “ares” to reach this ideal state:

1. Are we doing the right things?

2. Are we doing them the right way?

3. Are we getting them done well?

4. Are we getting the benefits?

If you think about these four questions, you can easily see that many companies focus their efforts on the first but never address the other three concerns.

Since most competent IT executives understand what governance should be, it is safe to assume that they have good reasons for not pushing harder for full-cycle governance. In my view, they don’t believe the effort is justified.

Without question, it’s hard work implementing full-cycle governance. Many organisations have immature investment disciplines typified by a “if we build it, the value will come” approach. They will not devote the effort necessary to establish measurement systems that will sustain a focus on projects for years after they have been theoretically implemented. In addition, since the majority of CEOs believes that CIOs should be held accountable for realising value from IT investments, it takes a lot of rewiring to get the accountability shifted to where it needs to be: the business leaders who own the three P’s of value realisation — people, processes and P&L. Finally, a lot of IT types don’t believe that value targeting and monitoring is practical (for example, proponents of the “ROI is dead” position).

What’s more, from a payoff perspective, the link between full-cycle governance and CIO success is hard to see. CIOs understand that as long as they ensure that the right projects are selected and successfully delivered, they will not only survive but thrive. The effort required to achieve full-cycle governance is just too much for most CIOs. As a result, they limit their project management to govern-once. These CIOs believe they can make most any project succeed as long as they have enough business sponsors, subject matter expertise and money.

That belief is wrong. Business sponsorship is a weak substitute for leadership. Many CIOs don’t understand that full-cycle governance is an industrial-strength method of improving the success of IT projects. By driving project success, IT value becomes more than a projection — it becomes reality.

It does so by transforming a business project sponsor into a business project leader. Using a parenting analogy, full-cycle governance forces business leaders to raise the project as one of their own rather than send it to the IT boarding school. This increased accountability motivates business leaders to assign their top talent, establish baseline measurement systems, manage risks and manage project scope, and see changes through. Full-cycle governance induces business leaders to cascade project accountability throughout their organisations rather than leaving accountability to the CIOs.

It’s time for CIOs to reassess full-cycle governance. The govern-once mechanisms in place at most companies are only a first step in realising value from IT-enabled business investments.

Reader Q&A

Susan Cramm answers questions on “A Cry for Full-Cycle Governance”

Q: What is the best way to codify the life cycle of corporate governance?

A: The first step is getting the CEO and COO to make it clear that business executives are accountable for value realisation. The second step is showing that value monitoring can be made practical and consistent with the company culture. Partner with the CFO and find a receptive general manager and a project that can be used as a test case. After a few iterations of this process, the company commitment will have solidified, and the requirements for a codifying tool will be understood.

Q: In your example, the project review process is typically performed once. My experience is that changes in excess of a certain monetary or value threshold require that a project be re-evaluated. The review process is both time-consuming and expensive.

A: You bring up an important point: Successful governance mechanisms depend on ensuring that the initial and subsequent project evaluations and decisions are performed quickly. Review processes can be expedited if the executive investment management board: 1. understands how much it can spend and what return the project should generate; 2. is able to convene virtually to make critical decisions quickly; 3. is supported by an independent group of value analysts who ensure quality submissions; and 4. delegates responsibility for decisions that are below certain risk thresholds.

Q: We have taken the first step in implementing governance councils with relative success. However, we struggle with keeping project updates focused on pertinent issues.

A: Great councils rely on good staff work. Good staff — in the value management world — will focus the council on the decisions it needs to make by presenting the relevant information in a succinct format. The value management analysts should be independent of the IT organisation — most likely staffed from the finance function.

Q: How can we get answers to John Thorp’s questions? For example, how do we benchmark IT change control? We have already checked ITeL [the IEEE Information Technology Library], some consulting companies and our peer organisations.

A: Don’t start from scratch, but do remember to start your education process at the top of your organisation. Don’t benchmark with peer organisations; benchmark with companies that are more mature in their IT practices. Look for organisations in industries or segments that have a higher level of information intensity. For example, if you are in retail, visit distribution companies. If you are in brick-and-mortar manufacturing, visit online manufacturers. If you are in health care, visit insurance companies. If you are in insurance, visit banking.

MANAGEMENT REPORTS

Toolkit Pulling Out All the Stops to Get Restarted

Balanced scorecard, benchmarking, core competencies — companies’ use of these management tools and techniques and many others have boomed since the start of the recession in 2000, according to a survey conducted by consultancy Bain.

On average, companies in the survey — which was mailed to CXOs at companies of all sizes and industries worldwide — used 16 out of 25 tools and techniques in 2002, versus 10 out of the same 25 in 2000. The top three were strategic planning, benchmarking, and mission and vision statements, all used by more than 80 per cent of the respondents.

The turn to management tools is driven by a search for profits, says Darrell Rigby, a Bain director who launched the annual “Management Tools and Trends” survey in 1993. “Now that they’re squeezed, people are using these methods in search of growth,” he says.

The two tools showing the sharpest increases in usage are of particular interest to CIOs: 78 per cent of the 708 senior execs responding to the survey said they used CRM in 2002, up from 35 per cent in 2000; and 62 per cent of respondents applied knowledge management methods in 2002, compared with 32 per cent in 2000.

The popularity of CRM was the survey’s biggest surprise, in Rigby’s view. “Some people thought CRM would follow the route of re-engineering,” he says. But CRM is proving effective at many companies, as indicated by rising satisfaction levels among survey respondents — from well below average in the 2000 survey to just about average in 2002. Rigby attributes this increase to greater sophistication in companies’ use of CRM. Rigby notes that many executives use a low-tech definition for CRM rather than automatically equating it with software. “CRM is almost becoming the new vocabulary for a customer-focused strategy,” he says.

The survey’s signals are mixed on knowledge management. KM registered some of the lowest satisfaction rates. “When people are measuring the net benefits, they’re saying the benefits aren’t as high as they thought they would be, and the costs are 10 times as high,” Rigby comments. Even so, the use of KM increased among survey respondents, in line with the general upswing for all methods.

Another survey finding of interest to CIOs: Only 19 per cent of the executives agreed with the statement that “We have cut back too far on IT spending.” Fully 57 per cent disagreed. Is that a call for CIOs to align all IT projects with business needs, or just to rename them as management tools and techniques?