TechWatch: Technology-Enabled Self Service—Where are the Boundaries?

Most of us have been frustrated by calling a business colleague with an urgent question and getting her voice mail. Worse, is trying to reach that person through a series of voice mail prompts that leave you looking for ‘press 6’ when there are only five choices.

This is one of many examples where automation made possible by technology has replaced service by a person. The motivation for this is clear. Cost pressures in an ever more competitive business climate cause businesses to look carefully at any opportunity to save a dollar. In addition to this, new innovations in technology continue to open new opportunities for these savings.

There is a catch, however. Like any other technology, such automation can have a ‘bite back’ effect. (See Why Things Bite Back, by Edward Tenner.) Savings may be lost with frustrated customers. Sometimes savings may not be there at all.

An Early Automation Example

Such automation is not new. I was reminded of an early example when I went to a meeting at the Smith Tower in downtown Seattle last month. When it was built in 1914, the tallest building in the U.S. west of the Mississippi, it had an elevator operator. It still does.

Now, almost all elevators have been automated, and the passenger does the work once done by the operator. It would seem this is the perfect example of no downside through technology enabled self-service. It doesn’t really impose any additional burden on the customer to self operate the elevator, and the business saves the cost of an operator.

However, there are several things to think about. The transition cost is the first. The old elevators were not ready to be operated by just anyone, so there is the cost of the new system. While it may seem like there is no learning curve to stepping on the elevator and pushing the button yourself, I would guess there were some reluctant first timers. Finally, there are the more subtle things that are difficult to measure. What role might the elevator operator have played in security for the store or office? What role might have been played in customer relations?

Other Self-service Examples

There are many other examples which are not so clear cut. In the banking industry, through the use of home computers and ATM machines, the customer does work once done by a bank employee. Phyllis Campbell, President of US Bank of Washington, will give her views on this and other subjects in the banking business in the IBTE Conversation in the October issue of Ethix.

Airline reservations and ticketing offers another example. Earlier this year Delta Airlines announced discounts for those passengers who would purchase their tickets through the Internet. They backed away after customer complaints. Alaska and other airlines do offer additional miles to those passengers who use electronic tickets and check in through kiosks.
Because of the PC, many executives and scientists draft their own papers. Remember the days of writing something out long hand and giving it to a secretary to type? Direct entry into the computer is an obvious benefit for those of us who have keyboarding skills (once known as typing!). It eliminates the errors introduced by poor penmanship and secretarial error and, for many, it is faster to type than write.

For those without keyboarding skills, however, this can be a real hardship. Some say the technological advance of voice recognition will overcome this. Even today with the significant error rate, it is useful for some, particularly those with repetitive stress syndrome. As a universal solution, however, it is not clear. Do you really want to spend the morning talking to your computer?

One common denominator on these and similar examples is gaining comfort with computers, ATMs, kiosks, phone menus, and the like. Often, but not always, once the familiarity is there, the new process is simpler and quicker than the old. Customers don’t mind self-service if it brings value to them (extended hours, faster service, etc.). But getting past the learning curve is an individual thing. For example, some people still have not used a cash machine.

A broader class of self-service involves the elimination of overhead resulting in internal self-service. Support personnel are eliminated and executives answer their own phones (or people calling them use voice mail) and make their own copies. Lobbies are closed and employees let themselves in with a security card, or go to the door to admit their visitors.
I think this trend in many areas of business is more problematical. The salary of the person eliminated is measurable so it is easy to ‘book the savings.’ On the other hand, with highly paid executives, scientists, or mechanics now doing the work, which other lower paid individuals once did, it is not clear this savings is real.

It is not an issue of snobbery. Rather, a reverse trend may just be good business. Treating mechanics as surgeons, where the right tools are always available in the right place to do the job, frees the mechanic to use his or her abilities.

If we go beyond the savings (or lack thereof) of reducing support personnel, there are other considerations. The subtle things like customers getting lost in the phone system, or engineers putting out lower quality documents without the benefit of secretarial proof reading may be the real concern here.

More Help from Technology

There are a couple of hot technologies, in addition to voice recognition, which offer some promise around this self-service issue.

The first is usability engineering. A combination of psychology, design of experiments, and statistical analysis, usability engineering seeks to assure that human interfaces to the phone, computer application, or kiosk are readily understandable by someone other than the designer.

Many may assume that software and other interfaces to automation tools are routinely tested in this way already. This is far from the case. A recent collaboration between industry and the National Institute for Standards Technology (NIST) has begun a major effort to put standard tests in place. Some major businesses have decided to make usability testing a requirement before they purchase software.

The second relevant technology is intelligent agents. Intelligent agents are computer based tools designed to anticipate needs and handle details for the user. This is a very complex area for two quite different reasons. Many of these needs come from the way multiple computer systems work together, so it is not enough to have agents in each application. Building architectures enabling them to work together across different applications is an ongoing research problem.

Another difficulty here is that agents have to anticipate needs, and thus fall into a category once known as artificial intelligence. A characterization of these problems is that they are always harder than they seem. Early attempts have been less than satisfactory except in specific niche areas, though some improvement is being made.

Some of us remember Microsoft’s product ‘Bob.’ This was an intelligent environment for helping users find things on their computer system. Introduced with much fanfare just a few years ago, it was more cute than useful, and it wasn’t
very cute.

More recently, other agent-like features on your computer system have been slightly better, but not much. Perhaps you’ve been annoyed by the little icon that says ‘It looks like you’re writing a letter. Would you like help?’ Or perhaps you have created a list with bullets, and the system insists you indent the next line.

Another simple example is trying to write ‘(c)’ in Microsoft Word. The “agent” is sure you are trying to write the copyright symbol © and automatically converts it for you. Putting extra space around the c and later taking it out is the way I’ve found to override the system, but this adds work rather than helps. It is a useful way to create © however.

A broader application of agents is in the travel services area. The unreached goal is to have a computer agent do many of the things a good travel agent does: know your profile, anticipate needs, and the like, taking this to the level of technology enabled self-service. Such systems are not very good yet.

These negative examples only serve to underscore that the problem is difficult. Good solutions are near for some problems, and a long ways away for others. As in many ‘AI’ examples, very small apparent differences between two problems may render one easy and the other extremely difficult.


In our cost cutting, competitive world, technology enabled self-service looks like a dream come true. At its best, it offers less cost for the business and better service for the customer. At its worst, it frustrates the customer (thus hurting the business), and actually raises costs for the business. The data is not always available to measure these differences, so a careful collaboration between technology, business, and human factors is essential.


Al Erisman is executive editor of Ethix, which he co-founded in 1998.
He spent 32 years at The Boeing Company, the last 11 as director of technology.
He was selected as a senior technical fellow of The Boeing Company in 1990,
and received his Ph.D. in applied mathematics from Iowa State University.