The Story of Symbol Technologies
by Theo Pavlidis
Draft of October 29, 2005 with notes added on
See below for additional notes, incl. the Motorola aquisition
Three year ago Symbol Technologies was the leading company in barcode scanning when it was shaken by accounting scandals and a big drop in its stock price. In the summer of 2002, Bill Nuti joined Symbol as president and within a year and a half he had become the CEO and had replaced all the old management plus the board of directors. These changes resulted in improvements in the outlook of the company but in the summer of 2005 Nuti also left the company and the company's stock price returned to the 2002 levels.
The purpose of this document is to show that the downfall of the company was due mainly to ignoring technological issues and all other factors were secondary. This version is a personal and rather informal account and some dates or numbers may not be accurate. I have been revising the document in response to comments of readers and I plan to continue to do so. I did not intend and I did not try to cover the full history of the company - a good source for the latter are the memoirs of Paul Berge, a former senior associate of Symbol Technologies. Also I do not discuss the accounting scandal; a lot has been written in the press about that, see for example, the SEC Litigation News Release. However it is best to "google" for results on the litigation since news releases may not have a long life span. (Noted added on August 10, 2009: Link to Paul Berge's memoirs updated. Text about the accounting scandal revised.)
Author's Background: I was a consultant for Symbol Technologies from late 1986 to the end of 2000. From Jan. 1, 2001 to June 2002 I served as Chief Computer Scientist of the company. An indication of the extent of my involvement with the company is that my name appears in at least 12 Symbol patents. I left in the summer of 2002 (by coincidence, about a month before Nuti joined the company) after several acrimonious discussions with the senior executives of the company. My letter of resignation stated as the reason of my departure "strong disagreements with the management."
Note added on December 18, 2005
Recently I was shown a copy of the August issue of the WCCN (Wireless Communications, Computing & Networking) Letter that features an article titled Symbol's Last Chance (see summary). The article is quite critical of the Nuti period but also covers several errors of the "Pre-Nuti Symbol." For the latter period I found the article quite credible (I have been familiar with several of the errors they discuss) and, I believe, all these problems are a consequence of the failure of the company to acquire technical expertise besides what they had already in the scanner business. For those interested in corporate governance my posting provides the underlying reasons for the errors detailed in the WCCN Letter.
Note added on September 28, 2006
Note added on January 10, 2007
Today Symbol Technology ceased to exist as a separate company. The ticker symbol SBL was removed from the NYSE and typing "www.symbol.com" leads to a Motorola web site with the announcement that "Symbol Is Now Motorola" and that its ticker symbol has been removed from the NYSE. (Apparently the merger was finalized because of yesterday's European anti-trust approval.) I wish the best for the Symbol people under the new management but for me (as one who spent over 15 years associated with Symbol in one capacity or another) it is a sad day.
The "Good" Years
Symbol Technologies was founded in 1975 and in 1980 it introduced into the market its first product, the first hand held laser scanner LS1000. I started consulting with the company soon after I arrived in Stony Brook in the Fall of 1986. At that time the company had about 300 employees and all engineering work was on hand held laser scanners where the company held an enormous lead over the competition. A fateful event occurred in 1989 when Symbol acquired MSI, a company manufacturing hand held terminals. That company was based in Costa Mesa, California and it was the same size as Symbol, so suddenly Symbol doubled its size. For a while the company even changed its names and it was known as Symbol-MSI. Interestingly, this major acquisition is not listed in the corporate history posted on the company's web site.
The acquisition made a lot of sense: the combination of hand held wireless terminals and bar code readers was a powerful one. However the company did not appreciate the extent to which it had to broaden its area of technical expertise. The design and manufacturing of laser bar code scanners requires expertise in optics and electro-mechanics and the company had some truly outstanding engineers and scientists in those fields. Bar codes scanners require also a computer program to decode the signal, but as programs go that is quite simple. My students could write in C a basic bar code decoder within a single day and one of "industrial quality" within a week. However the wireless terminals required far more sophisticated programming skills. It turned out that several of the skilled software engineers of MSI had left the company before the acquisition and Symbol itself did not have any strong software engineers, since their skills were not needed in their laser scanner business.
This situation led to a crisis in the summer of 1990 when a new portable terminal was coming into the market and while the hardware was ready, the software was behind schedule. I was asked by the company CEO (Jerry Swartz) to take a semester's leave from the University and devote my full time to dealing with the problem. What was needed was a C language library to support the writing of applications for the new terminals. I found the old MSI staff highly uncooperative but, fortunately, the manager of a Symbol VAR in Belgium agreed to work with me and provided the required API specifications. I hired two C programmers and, in cooperation with the Belgium VAR, they developed the needed C library routines within a few a months. While that particular crisis was taken care of, it was clear that the long-term problems remained.
I recommended that the company should make it a top priority to hire an experienced software manager and I even recommended a particular individual for the position. There was little follow up. They thought that the person I had recommended was asking for too high compensation and they did not pursue any other leads. Clearly the senior company management did not appreciate the importance of software but there are other factors involved. There were several people in the company claiming software expertise and the company felt that could get away cheap by relying on them with the help of consultants. Hiring a senior software executive would have cut in the turf of the existing senior executives and they did not want to give up anything. The downside of this non-decision was that while the portable terminal division had a higher sales volume than the scanner division, it did not contribute to the company's profits. But the worst was yet to come.
While the company was struggling to absorb this acquisition, many exiting things were happening in the scanner division. I was closely involved in one of them. One of my students (Y.P. Wang) had joined the company full time and together (and the help of several other people) we developed the two-dimensional bar code PDF417. A special purpose scanner for PDF417 was also built and the product was launched in 1991. Wang had assembled a new software team because the programming for the two-dimensional bar code was far more complex than that for the one-dimensional bar codes. It soon became obvious to me that the company did not know what to do with the new product. Wang and I proposed that the company should develop some demonstration applications in order to prove to potential customers the advantages of the new technology but that suggestion was ignored. Frustrated, Wang left Symbol and in 1993 started his own company, Metanetics.
Early Signs of Trouble
Sooner or later all the people Wang had recruited also left Symbol. It became a pattern. Whoever I considered to be the best software person at Symbol would leave the company. The lack of a senior executive with any competence in software (to say nothing about the lack of a senior experienced software manager) made software engineers feel unappreciated or worse. This continuing outflow did not seem to perturb the senior management. "We can always replace people" was their answer. They did not seem to notice that replacements were either of lower caliber or if they were of comparable quality they would also leave after a couple of years. While turnover amongst highly skilled software engineers tends to be higher than amongst other specialties in all companies, the turnover at Symbol was well above what I had observed elsewhere. In 1992 I had also changed my own involvement with the company by insisting on a specific consulting contract that limited my services to bar coding decoding issues as well as reducing the overall amount of consulting I would be doing. Eventually I convinced Symbol to increase the funding of my laboratory at Stony Brook University because it was more likely to achieve results there than in the rather haphazard R&D environment at the company. In this way I distanced myself from the management problems of the company.
Because the laser scanner business was still going strong, the continuous loss of highly competent software engineers did not seem to have any effect on the bottom line of the company. However, by the late 1990s there was no room for growth in the scanner business and this is when the real trouble started. The obvious path to further growth went through the portable systems division but, thanks to the neglect of software issues, that division was in no shape to contribute. Some engineers also made efforts for developing an RFID product but they were told that RFID was never going to be competitive with barcodes and they should not bother with that area. (Apparently there were suggestions to make a serious efforts in RFID at least as early as 1995 but the company took no action.)
The company had put its hopes for future growth in the "personal shopper," a project that had started in 1993. The idea was to provide individual scanners to store customers so they could scan the items as they loaded their shopping cars. Then at the check out they would only hand their scanner to the cashier, pay the amount due and leave the store without unloading their shopping cars. Great idea because the potential market was enormous. There was only one catch. There was no way to make sure that customers did not put items in their shopping cars without scanning them. While spots checks and surveillance may cut on such theft, the problem remained. Because merchandise shrinkage affects seriously the bonus of store managers there was no way for the product to be adopted until the theft problem was solved. Because each time a customer scanned a product, a record of the time was made and this time series of scans was available for analysis, Swartz believed that such data would provide solution and put his hopes on data-mining. Unfortunately, there is no social or scientific reason to expect differences between the time series of an honest customer and the time series of a dishonest customer. In addition such data mining was likely to raise privacy issues that might discourage retailers from adopting the personal shopper.
The Really Troubled Years
I took early retirement from my University position in January 2001 and I joined Symbol as Chief Computer Scientist. I should point out that this was not a management position (it was within R&D) and I had no staff. Also my technical expertise is in Graphics and Image Analysis and while I have a fair amount of experience in software development it is limited in applications related to imaging. In my view the best contribution I could make to Symbol's software development problems was to help recruit a senior software manager to oversee the technology in their wireless and portable terminal division. Subsequent events convinced me that this is not what the company had in mind and I was recruited to that position because Swartz believed I could solve the data mining problem associated with the personal shopper.
Being a regular associate allowed me to take a closer look at the operations of the company and pretty soon I realized that the problems of the company were far worst than I had thought. The company seemed to be dominated by sales or marketing people and technical advice was ignored, and not just on software issues. Three products that have been introduced recently to the market all had performances problems. After I looked at each case carefully I realized that there was a common factor. In all three cases the original design had been modified to meet price goals with the people in charge seemingly oblivious to the effect such changes would have had on performance. While price-performance trade-offs are common in engineering design what was happening at Symbol was not a trade-off but a unilateral change in design. (Ironically one of the products was a laser scanner.) I wrote a memo describing the problems and met (separately) with both Razmilovic who was now the CEO and Swartz who had retired as CEO but was still chairman of the board and chief scientist of the company. It was clear that the top two people in the company were barely in speaking terms and my conversations with them were dominated by criticism of the other person while ignoring the specific issues I was bringing up. The company had become in effect leaderless and was allowed to drift without any direction.
Razmilovic cared only for immediate results and, as it came to the light in less than year later, he found it easier to manipulate the sales figures than worrying about actually increasing the sales. The proper development of new products was quite far from his mind.
For his part Swartz was pre-occupied with the personal shopper because he thought that its success would solve all the company problems. The system had been installed in certain stores in Europe and there were sample time series available so I was asked to investigate how these data could be used to detect dishonest shoppers.( These stores were performing exit audits of the customers so each time series was marked whether it was from a customer that had scan all items in the shopping cart or from a customer that had not.) As I mentioned a few paragraphs earlier there was no reason to expect that this could be done but I went ahead with a thorough statistical analysis of the data that confirmed my prior judgment. There were no statistical significant differences between the series of shoppers who had not scanned all items in the cart and those of customers who had not. Swartz would not accept this conclusion and this was the start of our falling out.
There was another issue where I had a disagreement with Swartz and he asked another long term Symbol consultant for a "second opinion." He and I and two other Symbol engineers met in late June 2001 to discuss his conclusions. It turned they were the same as mine and we quickly agreed that the issue at hand was not important (which is why I do not describe it here) but that the company had other serious problems. With the exception of Swartz himself, there were no strong technologists at senior positions. Because Swartz continued to be pre-occupied with laser scanners, no one was looking seriously at other technologies that might lead to new products. We agreed that the company had at most three years to turn itself around. In particular the portable systems divisions was still not profitable and if it continued to be so the best solution would have been to sell it and Symbol become again a relatively small scanner company. We also agreed to try to convince Swartz about the problems the company faced.
Soon after, the company's stock took a beating dropping to about $12/shr as a result of poor quarterly results. As the plot below shows the big drop in the stock price was not due to the general tendency of the market. This did not seem to cause as much as concern within the company as I might have expected and the aimless drifting continued.
Some people who saw the first draft of this document suggested that a comparison between Symbol and Intermec would have been more interesting than between Symbol and NCR. However, Intermec is a subsidiary of UNOVA, so there is no Intermec stock. It turns out that a comparison between Symbol and UNOVA is even less flattering for Symbol than a comparison with NCR.
This may be the right place to talk about the company's R&D division that was also my home organization. The division had close to 30 people and it was headed by a senior VP who reported directly to Swartz. It was a poorly run organization. Razmilovic wanted to abolish it because he thought it wasted the company resources; he called it "Jerry's sandbox." But Jerry Swartz himself thought that the only problem of R&D was lack of resources. In my opinion, the big problem with Symbol R&D was inability to decide which projects to pursue and which to drop. Everybody was involved in more than one project and many projects had less than one full time person devoted them. I believe it is a good idea to allow people to devote a fraction of their time to new projects chosen on their own initiative. But is also a good idea that after six months or at most year each new project be evaluated and either dropped or more resources devoted to it. But this was rarely, if ever, done and a lot of projects were limping along without adequate resources and without any idea on how important were for the company. In my opinion Katz, the head of R&D, was afraid to make any decisions without the approval of Swartz and Swartz was too pre-occupied with other things to take a close look at the various R&D projects. Even if he did, his view was colored by his commitment to laser scanners. There was an engineer in R&D who was pushing for work in RFID. He was allowed to spend most of his time on this project but was given little additional support. Senior company executives made frequent statements about why "RFID is never going to fly" and Katz toed that line. If I may run ahead for a moment, when the new management came on board in late 2002 they tried to catch up on RFID by acquiring another company, but a lot of valuable time had been lost in the meantime.
Swartz was a brilliant electro-optical scientist and he liked to run a tight ship on things that interested him. This was an excellent quality when the company was small (a few hundred people only) and its only products were hand held laser scanners. Besides outstanding technology, focus, and hands-on management were the reasons behind Symbols initial stellar success. But the last two qualities became liabilities as the company grew and diversified.
In March of 2002 Razmilovic was forced to resign the president and CEO positions that he held. Apparently the board of directors had realized that something was amiss with the reported financial results. By that time the stock price had dropped to around $8/shr and all kinds of ideas were proposed for turning the company around. The leading contender was for Symbol to become a "solutions company," providing integrated solutions to retailers. Given the weakness of the company in software issues I felt that this was a wrong direction. I tried to explain that the way to provide solutions in an economical way was to create software tools that would allow quick customization of generic applications. Unfortunately, no one seemed to be listening and I resigned from the company in June 2002. Because my major financial incentive for joining the company in 2001 was a block of stock options, once these became worthless I had little reason to continue working for Symbol.
False Hopes and Ultimate Reasons
About a month after I left Bill Nuti joined Symbol as president, a position that had been left vacant after Razmilovic had resigned. Rich Bravman took the position of CEO and Swartz remained as chairman of the board. Eventually Nuti instituted a complete change in management. By the end of December 2003 Nuti was the CEO and only two of the top 16 people of the company (senior VPs and up) were left in their positions. The rest had either left or were demoted. This story was unusual enough that it made the New York Times. Of the two remaining senior VPs, one left a month later and only the senior VP for human resources lasted till the Spring of 2005 when she was also replaced. (The R&D organization had also been dissolved and its staff was re-assigned within Engineering.) Then, in the summer of 2005 the Chief Financial Officer resigned followed by Bill Nuti and the company stock fell back to the level it was in the summer of 2002. The company had earlier announced layoffs of over 10% of its work force. The departures of the senior executives must have been sudden because their duties were taken by one person who at the time of this writing held the titles of Interim CEO, Interim President, Chief Financial Officer, and Chief Admin. and Control Officer.
It seems that while the new management fixed the most egregious faults of the previous management, they could not do much about the deep-seated problems of the company. They did hire a senior software manager but he did not last long. The failure to hire a strong software manager around 1990 (or a little later) had doomed the company. It takes time to turn an organization around and that time was not available in 2002. Thus, while a lot of attention was paid to the accounting problems while Razmilovic was president and later CEO, these were only part of the long-term problems of the company.
We may ask why the company never took the steps needed to fix the wireless and portable terminal part of their business. I believe that it was basically an issue of greed and ignorance. The existing management did not want to "cut in" a new person in the pie. That the pie might become much larger after he or she joined the company did not occur to them. This in turn was due to their lack of knowledge of computing and what software development entails. If Symbol had stayed a scanner company, it would still be quite successful, although of smaller size.
It has been pointed out by some readers familiar with the company that not all was "roses" with the scanner division and there were several failed projects there too. My personal opinion is that these projects did not cause any major damage to the company and one might argue that a certain numbers of failures are inevitable for a company that is in the forefront of technology.
Other readers have been pointed out that the accounting fraud was truly gigantic and there was other corruption as well, plus atrocious management in several areas. These observations might be true but I still maintain that they would not have been so damaging if the technological issues had been taken care of. (Text added on 10/29/05.) In an earlier draft I had included in this space praise for the management actions of the Nuti team. Since then I have also heard some negative opinions about them. Because I have not been close to the company after June 2002 it is best not to make any statements about how the company was managed after that time. The one position where I hold firm is that the neglect of the problems of the portable systems division "doomed" the company in a way and unless these problems are taken care of, the company cannot be turned around.
Some readers also asked me why I ignored the role of various individuals. I have tried to stay away from naming specific people except when they were so well known and they would have been recognized even if I had referred to them indirectly. I wanted to focus on the main reason for Symbol's problems and I did not want to digress by trying to assign individual blame outside the top level. Also all my work was in R&D or in new products and while I would hear stories about the follies of some executives, I had no first hand knowledge of their actions.