Originally published in Exame

June 4, 1997
English translation below
from the original article in Portuguese

Reportagem De Capa

Delete Yourself


Fact: billions of dollars are wasted in useless technology investments. Question: can one defend oneself?

By Helio Gurovitz@email.abril.com.br

Horses have forty teeth. Such statement, written by Aristotle in 350 B.C., was accepted as absolute truth until the end of the Middle Age. The Greek philosopher probably opened the mouth of a horse and was satisfied with the number of teeth he found there. Not until the first experimental scientists counted teeth of several horses, nearly 20 centuries afterwards, did one discover that truth was not absolute. Mares have 36 teeth. The horses may have between 36 and 40. Consider now another statement: computers increase the productivity and generate more profits. In the business world, few would dare to deny it. Only in 1996, companies and governments invested 1,076 trillion dollars in information technologies (think that the 13 049 larger companies in the world profited mere 750 billion). Well then. There is not an official estimate, nobody wants to talk on that subject nor likes to admit, but a nice chunk of that money will go, non stop, to the waste basket.

"I put the blame on the management. IT has a high cost, but executives seem to ignore it", told EXAME Paul Strassmann, former boss of the technological section of the Pentagon and of companies as Xerox and Kraft Foods. In the Pentagon, Strassmann controlled between 1991 and 1993 an annual budget of 10 billion dollars in technology. Alone, he says to be responsible for 0,5% of what the mankind has already spent in that area. In his book "The Squandered Computer", recently released in the United States, he does with computers what the first cientists did with horses: he analyzes experimental data of about 500 companies, picked up since 1980.

The main conclusion of Strassmann is demolishing: there is not the least relationship between what a company invests in IT and its economic success, be it measured as profit, growth or productivity. To spend a lot can mean innovation, but also waste. To spend a little can mean prudence, but also backwardness. The important is not how much is invested, but how. Does it seem obvious? And it is. Remind, even so, that it took to mankind nearly 2000 years only to discover how many teeth the horses have.

WASTE - The conclusions of Strassmann are not a surprise. In 1994, a study from Standish Group concluded that, among the IT projects of 365 American companies, 31% were canceled before ready (and that costed them 81 billion dollars), 53% surpassed more than 90% in the foreseen cost (to a price of 59 billion), 13% surpassed more than 200% and only 16% were concluded within time and budget. In Brazil, companies try to hide by all means the flagrant cases of waste, but some already begin to admit that, without facing that problem, the executives of IT sector won't survive. "At that time of high inflation, to increase 1% in the price gave more profit than to save 1% in the costs", says Augusto Marques of Cruz Filho, administrative and financial director of the group Pão de Açúcar. "Today it has changed".

He mentions a classic example of waste inside PDA, that intends to invest in IT 35 million of reais until 1998 (last year the retail store revenues were more than 3 billion). The company bought in 1989 a mainframe computer IBM 3090, able to process 21 million instructions per second. It was an investment of 16 million dollars in two phases, besides 2 million of expenses in the facilities. "It was a mistake. With the Plano Collor crisis, the computer could not make anything", says Cruz. It was a cannon to kill flies. The equipment was left aside and later was put off. In 1995, PDA decided to lease an IBM 9121, that processed 40 million instructions per second. The price? Less than 1 million reais per year. Today, the leasing of an IBM 390, with the double of the processing capacity, costs 1,2 million.

That case is far away from being unique. On the contrary, it is typical of the mentality among executives that work with IT. It is considered that computers have magic powers to solve administration problems, to rationalize processes and to increase the productivity. Nobody discuss how the machines will be received by the corporate culture, nor it is mathematically proven how they will increase the profits. Instead, the fear of competition reigns. It is common to justify investments claiming something like: if I don't do what the others do, I am fried. The same PDA set up three years ago a PC network with software from Novell, le dernier cri of that moment.

"Today we are changing everything for Windows NT, from Microsoft. We knew that in three years everything would turn into scrap, but we didn't have a choice. One cannot be left behind. It is sometimes difficult not to follow the herd", says Cruz. Paul Strassmann compares that mentality to the one in the arms race: "If your enemy has machine guns, you must also have machine guns. But it is not the machine guns that win the war. It is people. You win only by serving the consumers and generating value. Not by imitating what others do."

PIONEERS - Being original also has a price. The retail store Cândia, from São Paulo, decided in 1991 to be the Brazilian pioneer in computerized point-of-sales, or POSs, the substitutes of the traditional cashier registers. At that time, it paid a lot for a product that didn't even exist in the market. "It was a madness. I don't even know how much we spent", says Álvaro Farana Filho, manager of IT in Cândia. Did they have any returns? "We became well-known in the whole world. It was an exceptional marketing happening", says Farana. How much did they profit from that? Mystery. Today the retail net, that earned 320 million reais in 1996, invests in 80 new POSs to its new store in Morumbi. Each one costs about 5 000 real. The moment's fashion: the attendant can communicate through videoconference to make doubts clear when the bar-code of some product fails. That can even speed the attendance. But the amount of products with deffective code, today inferior to 10%, tends only to decrease. And the small camera in the equipment may just become an useless decoration, without generating any profits. "One cannot take the risk of being a pioneer. I won't be a piranha ox. Better to go slowly in the right direction that to run a lot in the wrong one", says Cruz , from PDA. (Note: 'piranha ox' is a brazilian idiom. Shepherds here usually choose an ox to be the first one to enter a river full of piranhas, a flesh-eating fish. This is the only one to die, because all the piranhas swarm around him. It is called 'piranha ox'.)

All in all, three sorts of people benefit from the competition fear and the herd spirit that govern the cycles of technological investment. First, the vendors of the technology, be them software, hardware or services companies. Second, the IT department inside the companies, that are politically strengthened by having a larger budget and by being responsible for fostering changes. Third, the consultancy companies, that have today at least one third of its revenues linked directly or indirectly to the diffusion of technological products. "Not long ago, the consultant was the independent guy that advised the best practice for the business. But more and more he is just called as a systems implementator", says Silvio Genesini, partner-director of Andersen Consulting. Andersen played a decisive part in one of the most famous cases of technological waste in the country, the one of the late Banco Bamerindus.

The bank was encouraged by the success of the first totally digital business table in Brazil, inaugurated in 1992. Andersen was hired then to implement a reengeneering plan, based on IT. One of the goals was to centralize the whole processing in Curitiba and, in a two-year period, turn off the mainframes, substituting them for networks controlled by smaller machines, called servers. Indian companies were contracted to make the software, using advanced techniques for that time. In 1994, as very little had been done, the bank decided to break the contract with Andersen and to make another with HP. But HP couldn' t conclude the task either, because, as it was proved, up to now the bank processing couldn't get rid of the mainframe. The result: they worn-out almost 300 million dollars in agencies repairs, purchase of machines, ATMs and software development. "HP superservers were never turned on. During more than one year, the NCR ATMs were left in the deposit waiting for new software. The whole software ended being thrown away. Done all the calculations, some 100 million dollars turned into garbage", says a high IT executive that took part in the process. The bank could never turn off all the mainframes. Instead, it went bankrupt.

The mistake of Bamerindus was daring to pull the herd. It ended up being just a piranha ox. When everybody was saying that the mainframe was condemned, got it right who doubted and waited. It was the case of financial institutions such as Bradesco or Real. "The new technologies just place the mainframe as a large server. We cannot survive without an equipment of high power", says Adilson Herrero, director of Real Planejamento and Consultoria, the company responsible for the technological infrastructure of Real.

FASHION - But how to know ahead of time when a technology will have success? When to innovate and when to wait? When to be pioneer and when to be conservative? There is not a single answer. The ability of who guides the technological area is exactly in knowing the time to step in the brakes and in the accelerator. There are companies that do well with a heady style, adopting and spreading the fashion. Others are successful by avoiding to succumb to it.

"A lot of people are fascinated with investments in computers. We think the opposite", says Sérgio Rubens Loeb, industrial director of Semp Toshiba. The company didn't have a data center up to 1989. When it went into computers, it adopted a PC network, instead of a mainframe. It is only now thinking about changing the database programs that run in the Clipper system, that, if not Jurassic as the mainframes, could very well be called Cretaceous. Among about 400 microcomputers, few run Windows 95. One and a half year ago, Semp began to evaluate new database and administration softwares, made by companies like SAP and Baan. "They don't sell software, they sell a package with consultancy, courses, manuals and a bunch of paper", says Fábio Tancredi Tonelli, IT manager at the company. "They don't even offer nor a module for imports that talks with the System of External Trade of the Treasury Office, an essential feature for one who imports components."

According to ordinary wisdom, they are late, losing the ship and the opportunity to rationalizing processes and being more competitive. Are they? For four years in succession Semp was considered the best company in the electroelectronics sector. Last year, it received a prize as the company of the year, chosen by BEST & LARGEST, by EXAME, due to its business competence. It had revenues of 800 million reais and profits of 120 million. In software, hardware and IT services, it spent mere 490 000. Detail: in 1996, Semp also bought Lince, a manufacturer - guess what? - computers. "We are not the owners of the truth. The internal unrest with technology is constant, so that we don't come to eat dust in the future", says Loeb. In a recent visit to the company, Japanese executives from Toshiba didn't make a single remark about they being technological laggards.

Conservatism is not an exclusiveness of Semp, not even a warranty of success. The wholesaler Ceval, from Gaspar, in Santa Catarina, invests in technology 1,5% of its 2,8 billion of reais revenues. It has a dispersed network of equipments in 45 towns, and cost is one of its largest concerns. "It is not just because there is a new Microsoft system or software that we change everything in our machines", says Sérgio da Costa, the IT director. The idea of Ceval is to buy software and hardware after the herd has passed. "The advantage is that it is not expensive", says IT manager Ildemar Volles.

That reasoning works very well for modernizing machines and software that, on whole, will end up carrying out the same functions. But it becomes debatable if extended to bursting technologies like the Internet. "In spite of the pressures, we are only now studying how to implant an intranet in 1998. It is still out of cogitation to process information and to sell goods through the Internet", says Volles. Instead, Ceval is investing about 15 000 reais to implant the system of electronic transactions knows as EDI, that began to be disseminated through the country in the end of last decade. The company has already connected 15 customers to the system. Today EDI is cheaper, but there is a big risk that it is swept of the map or absorbed by electronic transactions through the Internet. "We may be surpassed within three years", admits Costa.

Which technology is better: the new (and expensive) or the old (and cheaper)? "It is always possible to wait too much and lose the ship. The secret is to buying the best possible in the moment of the purchase, and wait larger periods among the upgrades", said to EXAME John Dvorak, columnist of specialized magazines as INFORMATICA-EXAME. Very often, to change the equipment is cheaper than to make a new maintenance agreement. "When the three years of warranty finish, to buy a new machine is equal to pay for a warranty of more three years", says Rudimar Dall'Onder, industrial director of Grendene. A maintenance contract for a personal computer can cost up to 500 reais per year. To change the equipment, including the maintenance in the three-year warranty, leaves it for 1 600.

SINGAPORE - Other frequent doubt is up to when to use a personal computer. It is common, in the balance sheets, to depreciate the equipment in two or three years. But there are different views. "A personal computer has to last at least five years. Full stop", says Carlos Roberto Boschetti, director of IT at Volkswagen. "Executives want new technology only because of their huge egos. I don't give notebooks for none of them any more." The last move of Boschetti was known at Volks as Projeto Cingapura (Note: this is actually the name of a project by the former mayor of São Paulo to refurbish shanty towns.) A lot of people was asking for new personal computers. Boschetti then invited 60 executives to come to the company headquarters on a Saturday morning, accompanied of a son, a screwdriver and tongs. He raffled adults and children couples, distributed to each couple a plastic bag with electronic components, a set of instructions and said: "You are now alone. Do yourselves the upgrade of the PCs". The result: with an investment of 48 000 dollars, he transformed what was considered scrap into 60 new Pentiums 133 and 166 MHz.

But with hardware prices falling 45% a year, for mainframes, and 6% a month, for PCs, the cost of the machines seems to be the smallest concern for who invests in IT. The main problem is how to justify ever growing expenses, that increase about 20% a year in Brazil, without a concrete proof that the computers increase the productivity within companies. The so-called productivity paradox, or computer paradox, has puzzled economists since the last decade. Between 1985 and 1995, everything that was spent in IT gave a meagre return of 1% a year, according to a recent study by Gartner Group. "We see computers everywhere, except in the productivity statistics", Nobel of Economy Robert Solow, of Massachusetts Institute of Technology, is said to have said, inaugurating the debate on that subject.

Student of Solow at MIT, Paul Strassmann claims that is impossible to avoid the paradox: "I calculated the statistics in all possible forms. They always show a paradox with computers". To measure the effect of computers in productivity is not trivial. Asked about how to evaluate the profits generated by investments made in technology, all IT executives get confused . "To invest in technology is like a bet. If you discover how to calculate the exact return, then tell me", says Rudimar Dall'Onder, of Grendene.

RETURN - Feeling is the most used word when one asks for a method that simply allows to know in reais the return of an investment made in reais. "Our largest return is the customer's satisfaction. It would be feasible to do all the exact calculations only if there was a better precision of information", says Aluízio Borges, executive director of Bradesco.

Yes. To satisfy the customer is the minimum required from any business. Bradesco is the kind of institution that can give itself the luxury of exploration and of spending 190 million reais in IT. (That expense didn't arrive at 25% of the profits of the bank last year, a modest value if compared to banks such as Real, that spent 30% of the profits, or Citibank, that spent 40%.) Bradesco can also give itself the luxury of being the first bank to offer transactions along the Internet, or of making experiences with smart cards. "Those are things that I would use as customer - and that is important when deciding. The analysis of costs is only good to say no to a project. To say yes, it is not enough", says Borges. OK. After all, based on that sort of reasoning, the bank foresaw the success of ATMs and set up the largest network of self-service banking in the country. Today, the network is responsible for 61% of the bank´s transactions. But still is it surprising that the bank that makes more profits in Brazil cannot demonstrate why computers bring profits.

Can anybody?

There are some attempts. In a deposit of PDA, for instance, each man moved 40 boxes in one hour. In the USA, that number was about 150. After travelling to Canada and bringing a more rational working system, that included IT, the deposit's boss claims he can already move 120 boxes and he wants to arrive at 180. But was technology the sole responsible for the change? No. The working method in the deposit was wrong. For working better, it was essential to adopt two docks and to rearrange the boxes. That could have been made without any computer. It would be a crude mistake to attribute only to technology the credit on something that could be improved without it. The only number that makes direct relationship between technology and profits is the so-called break index (goods with due period or stolen). It dropped from 5% to 2,9% since the stock controlling systems were installed. The difference of 2,1% is equal to 50% of the supermarket profits, or 60 million dollars. But it is doubtful that the whole credit of that result comes from technology.

Paul Strassmann believes that, well managed, the computer can be a crucial weapon in business: "It is the most wonderful invention since the discovery of the fire". Strassmann also has his methods to evaluate productivity. He made a study with 183 companies in several sectors, analyzing indexes of what he calls information productivity, a concept that he intends to explore in another book. His conclusions, although not definitive, give a track of where may be the waste with computers:

* Companies with the largest budgets of IT (average of 253 million dollars) have more than double chances of being among the least productive. Those with smaller budgets have double chances of being among the most productive.

* The less productive ones spent 30% in outsourcing; the most productive, only 15%.

* The less productive ones spent 18% of the technology budget in reengeneering; the most productive, only 5%.

* The most productive spent 4,1% of the expenses with personnel in training; the less productive ones, only 2,7%.

* Among the ones that cut employments, the less productive ones cut 40% of the work force; the most productive, only 12%.

* Among the most productive, 6% of the CIOs reported directly to the CEO; among the less productive ones, 19%.

* Among the most productive, 59% adopted centralized processing; among the less productive ones, only 22% of the operations were centralized.

* In the most productive, 45% of the employees had PCs; in the less productive ones, 64%.


You shouldn't. The cost of the personal computers became one of the largest concerns among executives. If you still think that, installing a personal computer, you are providing your employees with a productivity tool, wake up. You may only be creating another center of costs. In 1993, a study of Peter Sassone, of the Georgia Institute of Technology, showed that, among 1 700 users in several companies, the computer rather reduced than increased the productivity. The Gartner Group estimates that 70% of the projects of ITdidn't bring thje expected return exactly for their flaws in the integration with employees.

LOST TIME - The most curious reason is known in English as futzing. It is the time lost in activities that would not exist if there weren't computers: waiting for programs to run, for reports to be printed, formatting documents, learning how to use new softwares, discussing computer problems with colleagues and even deleting and organizing old files. The company Minnesota Mining & Manufacturing discovered that 30% of its PC users lost data annually and that, on average, they take one week a year to recover them. Each employee loses at least five weekly hours in computer problems with no relationship to work, according to a research with 6 000 people accomplished by SBT Accounting Systems. The futzing would cost between 737 and 5 206 dollars per computer. The Gartner Group estimates that there are annually 26 million working hours lost, or more than 750 million dollars, only with videogames. Not to mention the Internet.

There are today in the world about 200 million PCs. In the year 2000, it is foreseen that 117 million will be sold. The largest unknown is how much it costs to keep each one of them. The property cost involves not only the hardware, but everything that is associated to the machine, from the maintenance to the futzing. Strassmann claims to have listed about 30 possible sources of expenses for each PC. But not even with that does he have a concrete answer. He writes: "In 1996, Fortune put the average cost of ownership for a PC at more than 9 000 dollars, The Economist at 6 400, The New York Times at 13 000, Business Week at 8 000 and Datamation at 10 543. The Gartner Group evaluates between 7 138 and 13 000, depending on who asks the question. It never costed more than 3 000 to maintain a terminal of a central computer".

If you estimate the cost of ownership for a PC at, let us say, 9 500 dollars, nothing more reasonable than demanding an equivalent increase in the employee's productivity. Instead of doing that kind of calculation before purchase, the American companies have been spending every year 4 970 dollars per employee in average in IT, an expense only surpassed by the costs of health, pension and benefits. And that number grew 63,6% since 1994, without the shadow of an equivalent growth in productivity. Some other calculations can be made. Considering all its useful life, a PC costs about 38 cents for every hour of use. But the person that sits in front of it costs at least 40 dollars for the same hour. That is to say: every minute that the machine is not working costs more to the company than one hour that it is tied up. Maintenance and efficient training are, therefore, indispensable.

NETWORKS OF WASTE - The diffusion of PC networks seems to be one of the great responsible for the increasing costs. "The world of the distributed computing is the world of waste", says Antônio Candal, country-manager of Platinum of Brazil. A detail: up to 1990, Candal was one of the largest deffendants of the use of such networks. In 1991, he noticed that, with decentralization, they brought more problems than solutions. A research by InformationWeek revealed that 53% of the PC network projects in 125 companies were out of schedule, out of budget and with less resources than specified. A third of them were canceled. Would you fly in an aircraft company in which 53% of the flights don't arrive to the destiny or arrive late?

In Brazil, it is estimated that the failures in migration for networks controlled by the system Windows NT is twice as larger as the number of successes. The Gartner Group estimates that, during a five-year period, a network of PCs for 5 000 users is 70% more expensive than the same network controlled by a mainframe (241 million dollars, against 141 million). The worst, Gartner says, is that 57% of the cost of a networked PC don't appear in the technology budget. It is part of the so-called invisible costs.

An example: Volkswagen, that spends annually about 80 million dollars in IT, decided one and a half year ago to evaluate how much it wasted in paper. That it wasted, it was evident. All the calculations made, the conclusion was reached that the company spent 228 tons a year in the Data Processing Center. It was about 75 million pages, or 1,66 million dollars. The price of 100 automobiles Gol zero-kilometer, of a medium model. About 80% of that expense was pure waste. It was as if, every month, more than half a dozen Gols came out of the assembly line directly to the waste basket. The company made na outsourcing contract with Xerox, changed all the machines, rationalized the paper use and cut the printing expenses by 74%. The paper waste dropped to 1 ton.

Another invisible cost that few remember to attribute to technology is communication. In a study with 183 companies of several sectors, Strassmann discovered that half of them spent more than 26% of the IT budget in telecommunications. At Volks, the telephone bill in February was 708 000 reais. From that total, there no control for more than 340 000, because the calls are made from old switches. To monitor the remaining part, a digital tarifador was implanted at the PABX. Each employee has a password and is forced to identify, by means of the last digit dialed, if the call is personal or professional. With that, the telephone costs were cut from more than 17 reais for produced car, in December, to around 13 reais, in February, what is equivalent to the price of the duct that links the carburator to the motor. "We could absorb nearly totally the increase in tariffs", says director of Volks Boschetti.

LESSONS - Paul Strassmann affirms that the only thing that has a clear relationship with investments in computers is the overload in administrative expenses. The general expenses caused by computers grew 48% grew between 1988 and 1994. (In the same period, revenues and profits grew around 30%.) Each dollar spent in administration consumes between 15 and 25 cents in expenses with computers. Strassmann says that it is very difficult to estimate how much the invisible costs represent of the IT expenses, but he says that, for half the 142 companies in which he studied that aspect, they are more than 15%, inside and outside the computer science area.

One of the largest causes of waste ends up being, according to the researcher, the software written and re-written several times. The average age of the software running in a large company is from 6 to 8 years. For Strassmann, all software, if well written, can last forever. Who plans it should think in the long term, at least 20 years ahead. But, as most of the responsible persons for the ITarea stay little time in that position (in the United States, a three year-old average), there is no concern about producing a durable software. Result: everything has to be done and re-done. The problem that will affect the computers with the arrival of the year 2000 is just an example of that. Another source of expenses is the constant upgrading of software packages and operating systems. Only the upgrading for Windows 95 would cost for the American economy about 20 billion dollars, according to an estimate published in 1994 by Forbes magazine.

From the more than 400 pages of Strassmann's book, one can get some lessons. First, that one should not just buy technology because the competitors are buying. Second, that one should doubt faulty methods used by consultants and service companies to calculate return on investments. "It is always possible to demonstrate that the computer accelerates some process, but not always that reverts in benefits for the consumer or for the shareholder", says Strassmann. Nobody needs a Ferrari or a Porsche to move in São Paulo traffic jams. A less powerful car may be less exuberant, but has the same effect. With computers, says the researcher, the same logic applies. The best is what does well, not what does best. As the machine is not a given horse, it is good to have a look at the teeth.



And other clues to avoid the herd spirit

There are just two good reasons to invest in IT. The first is to increase sales and profits. The second is to render a better service to the final consumer. Below, recommendations that help you getting close to those two aims:

  • Distrust methods used in the market to calculate the return on investment. Many of them attribute to the technology results that could be obtained only with management changes.

  • Don't trust the vendors and consultants blindly. Make your own calculations. The result should be given in currency, not in time for return or unbelievable percentages.

  • When comparing your expenses with your competitors', never use the revenues as a parameter. Measure expenses per employee.

  • Try to have an exact idea of how each technological investment is added to the value of the products. Costs in reais should be translated by benefits in reais.

  • Never outsource essential systems of the company. In outsourcing contracts, guarantee that, if there is dissatisfaction, to give up is painless and cheap.

  • To copy what competitors do can be disastrous. Distrust miraculous medicines and best practices. Don't follow fashions, nor be the first to adopt a fashion. The cost of the mistake can be your job.

  • Think in medium long terms. Investment in training can give more return than buying machines or software.

  • Software and data of the company should be considered assets and should only be thrown away when absolutely necessary. All the programs and stored data should be produced having in view that they will last the longest possible period.

  • Don't depend on only one software and hardware vendor. Uniformity warranties, spread by the salespersons, may be just artifices to keep you captive. Adopt independent standards and the so-called open systems.

  • When upgrading equipment or software, the burden of proof lies with the change advocates. It is them that should justify the profitability of the investment. When in doubt, don't upgrade.

  • Go back up to the Strassmann, Inc. home page.