Zarlink: Chipping AwayBy Mel Duvall | Posted 2004-08-01 Email Print
The mandate from the semiconductor maker's CEO: slash technology spending by 40%. Here's how the company beat that figureand increased productivity to boot.
The market didn't know it yet, but Zarlink would report a first-quarter loss of $116 million (U.S.) on revenue of $105 million for the quarter ending June 29, 2001. That was a stark about-face from the profit of $28 million on revenue of $184 million the company had earned in the quarter a year earlier. Worse, reports from the field indicated that fiscal 2002 would be a bloodbath for the semiconductor industry.
The message that newly hired CEO Pat Brockett delivered to O'Connor, his chief information officer, was blunt: Spending in the company's information-technology department had to be slashed, fast and deep. "It wasn't just a matter of cutting spending," says O'Connor. "I was also being asked to increase the level of productivity."
Brockett gave O'Connor his marching orders: He had one month to come up with a plan.
The company had other challenges. Zarlink had just been spun off from telecommunications company Mitel Networks as a pure-play semiconductor fabricator. During the technology boom in the late 1990s, Zarlink had acquired several competitors but never fully integrated their systems. That meant there was no single view of the newly independent company's financial reporting systems. Reports and records had to be pulled together by analysts to get a complete picture of the company's far-flung operations in North America, Europe and Asia.
Brockett, who had headed up National Semiconductor's $1.5 billion analog and wireless group, was hired to begin the new chapter at Zarlink. O'Connor was also fresh onto the scene from Singapore, where he had led Mitel's Asian technology operations. Zarlink's previous CIO had left after the split from Mitel. O'Connor had a background in finance, and Brockett wanted someone who knew how to crunch numbers to plot the company's technology strategy.
Zarlink's primary problem, O'Connor recalls, is that as a result of the Mitel spin-off, the company inherited a technology infrastructure built for a much larger enterprise. Mitel had annual revenue of $1.6 billion just before the split, and its semiconductor division was responsible for about $600 million of that figure. But with the economic downturn, Zarlink was on target to record only $222 million in revenue in fiscal 2002.
Chief among the inherited systems: an SAP enterprise resource planning system installed in 1999 and 2000 at a cost of $24 million. Of Zarlink's 112-person information-technology staff, 36 were dedicated to the SAP system. The company's annual technology operating budget was $20 million, of which about $6 million was directly attributed to SAP.
Not only was SAP eating up a huge chunk of the company's technology budget, it was not delivering an appropriate return on investment. "We essentially had a system that was built for a company twice our size, and we weren't getting the benefits out of it that we should have been," O'Connor says.
For $24 million, Zarlink's management expected to have on-demand access to the company's key performance metrics, such as daily sales, receipts, sales backlog and inventory. But because SAP had never been fully integrated with the rest of the company's core applications, such as budgeting and forecasting, inventory management and human resources, reports had to be cobbled together by analysts each week. All told, the company had 70 to 80 different core applications running its operations.
"The I.S. organization was struggling when Bill came in," says information-systems manager Macrae Morse, who lived through the spin-off from Mitel. "We were working with I.S. fiefdoms and as a result, there was no single sense of financial truth."
Hashing Out a Plan
O'Connor huddled with two senior team members, Morse and senior business analyst Paul Donnelly, to plot strategy after the meeting with Brockett. Hashing out their plan on a whiteboard, the team came up with a strategy they believed could dramatically cut costs and improve reporting capabilities. They also knew it would be controversial. Key to the strategy:
The business case projected reducing technology spending by $3 million in the first year to $17 million, followed by a further reduction of $3 million in year two and $2 million in year three. By 2004, O'Connor believed he could shrink annual technology spending to $12 million, a savings of about $17 million over the three years.
O'Connor presented the plan at a Zarlink managers' meeting, and braced for battle. And he got it, especially from senior managers responsible for purchasing the sap system. "Doesn't it make more sense to leverage its capabilities rather than freeze development?" they asked.
O'Connor's proposal to build custom business intelligence applications to pull information out of the existing applications raised red flags. Some executives were worried about development costs and project failure.
"There was a lot of angst around that boardroom table, and most of it was directed at me," O'Connor recalls. Only one person in the room fully supported the plan, but it was the voice that counted.
"I love it," Brockett said. O'Connor was told to get moving.
The I.T. team met to go over the plans. There was a tremendous backlash. Critics said that freezing further development of the sap system was a colossal waste of money and good technology. sap, they argued, should be the primary tool to gain intelligence on the corporation, not just an engine to input orders or track receipts.
For many in the 36-person SAP development and support team, O'Connor's pronouncement was tantamount to getting a layoff notice. O'Connor knew he would have to watch for friendly fire. If the project ran into even the slightest problem, a small army of threatened workers would be waiting to pounce.
The first application had to be rock solid and important enough to quickly win converts. The initial phase involved installing the data warehouse and selecting the business intelligence tools. The team picked Oracle's 8i platform for the database and Cognos for the reporting software. SAP's Business Information Warehouse was considered, but the cost was more than $1 million for the software, hardware and development, compared to just over $100,000 for the Cognos implementation, according to O'Connor. He also liked the fact that Cognos was based in Ottawa and its experts could be called upon quickly.
For the first reporting tool, O'Connor decided to build a daily report that could show the company's bookings, billings and order backlog. O'Connor knew the reporting tool would not be accepted unless it was validated. His first project hire was Karyn Houle, an analyst from the company's finance department whose job was to verify that the daily Cognos reports were accurate.
The reporting tool was completed within the 90-day time frame, and it was an immediate hit with senior management. Rather than waiting until the end of the week to see an Excel spreadsheet on sales, billings and order backlog, they were now getting graphical reports every morning with the previous day's numbers. The tool also allowed managers to easily drill down to see such items as sales by product, region or customer. The finance department continued to produce its spreadsheet reports in parallel with the new daily reports, but after two months of validating the results, the management team decided to rely solely on the interactive reports.
"That was a watershed moment," O'Connor says. "Until that point, there was zero interest in the finance community in what we were doing. Now we had their attention."
Over the next two years, the development team methodically produced a stream of other reports and scorecards for areas such as cash flow, accounts receivable, revenue budget and forecast, inventory, material budget and forecast, head count, and compensation planning and analysis. The reports were all designed to be self-serviceeasy enough for managers to use without the need of technical assistance. All development was kept to the 90-day cycle, increasing credibility as each new report was rolled out. In addition, making changes to the Cognos reports was relatively simple. Houle says a change could typically be made in four to six hours, compared to the four to six weeks required to change sap.
The impact was soon felt on the bottom line. Since the sap system had been relegated to a transaction engine, O'Connor was able to reduce the number of licensed users from 1,090 to 260. In addition, the number of technology staff dedicated to sap was cut from 36 to just one. No changes were allowed to the sap system, and it continues to function reliably. Where possible, redundant report applications for areas such as inventory and personnel were eliminated along with their support staff. By the beginning of 2004, O'Connor's I.T. head count had been reduced from 112 to just 34.
The $12 million annual-budget target was achieved and taken further. Fiscal 2004 technology expenditures came in at around $7.6 million. The most recent run-rate is $1.5 million per quarter, or $6 million per year, a 70% reduction from 2001 levels.
In total, O'Connor says the company invested about $500,000 in the business intelligence project over three years and has generated about $13 million in direct annual savings.
Ray Homan, vice president of high tech industries for sap, maintains Zarlink might have gained greater business benefits if it had expanded the use of sap in areas such as business intelligence, but he admits that business pressures may have forced the company's hand. "At the time the decision was made, it was probably prudent," he says. "It may not turn out to be the best decision over the long term, though."
In addition to the information-technology department savings, there have been other significant benefits. Since analysts were no longer required to pull together weekly reports, the finance department was able to cut its staffing roughly in half. The team also implemented a self-service human-resources application that also allowed staff cuts. In all, O'Connor estimates that about $10 million in annual savings were generated elsewhere in the company as a result of the technology overhaul.
The staff cuts were difficult but necessary, according to O'Connor. "One of the keys to our success is that each time an application was implemented, we went back through the process and made sure all of the savings were harvested," he says.
There is no denying that the past three years have been heart-wrenching for many of the people laid off by Zarlink. But rather than create a dysfunctional technology department, O'Connor maintains the opposite is true.
Houle concurs: "We feel we're playing a key role in the company's drive to profitability and future growth."
Zarlink Base Case
Company: Zarlink Semiconductor Inc.
Headquarters: 400 March Rd., Ottawa, Ontario, Canada K2K 3H4
Phone: (613) 592-0200
Business: Designs and manufactures semiconductors for voice, broadband and wireless telecommunications products.
Chief Information Officer: Bill O'Connor
Financials in 2004 (fiscal year ending March 26, 2004): $199million (U.S.) in revenue; $39 million loss.
Challenge: Drastically reduce the company's technology spending and, at the same time, improve reporting capabilities.