Online Customer Service Tough To Implement

The message from consumers this holiday season is very clear. They expect e-tailers to beef up their customer service operations.

E-tailers are now scrambling to come up with the right solutions to satisfy customer service needs. While the most comprehensive solution is to provide live representatives who are available by chat or even Internet voice, it is also by far the most expensive solution.

As a result, many cost-conscious online merchants are looking at so-called natural language service agents that allow customers to type in questions in English. The big question is whether it will work.

Nike’s Ask Jeeves System Just Doesn’t Do It

Giant athletic shoe manufacturer Nike, for example, recently entered a deal with natural language search engine Ask Jeeves to implement “Ask Nike.” The service supposedly lets customers ask questions in English about Nike products and services.

In order to see how well Ask Nike worked, an E-Commerce Times reporter went on the site to give it a test. The reporter chose the Nike iD program that allows customized shoes to be made with the buyer’s name, initials or short message on the back of the shoe.

After logging onto Nike iD, the reporter discovered that you could customize the shoe’s base color, accent color and personal ID. Not knowing what a base color was, the reporter queried Ask Nike, “What is the base color?”

Nike iD responded “We’re sorry, but Ask Nike was unable to answer your question.”

So much for the value of Ask Nike. What good is it if it cannot answer the simplest of questions?

Must Be Done Correctly To Avoid Going Backwards

The goal of this exercise is not to embarrass Nike. They did that themselves. The purpose here is to point out that natural language systems are very complicated to program properly, so that they know in advance the answer to all the questions they might be asked.

It is also to point out that customers are not going to be very forgiving when Web sites tout a supposed technological advance that does not work well.

Can It Be Done Right?

Are natural language customer service systems ready for prime time? The risks are very high unless the company doing the implementation has done it very carefully.

While Ask Nike left a lot to be desired by not even being able to answer the simplest question imaginable, for example, the Ask Red service from Neuromedia.com shows substantial potential for a natural language agent to provide acceptable customer service.

Neuromedia designs and develops natural language customer service systems. Ask Red is the name of its own system, which bills itself as a Virtual Representative, or vRep.

Ask Red carries on a complete online chat as if you were communicating with a human. The E-Commerce Times reporter who tested out Ask Red was quite surprised by its ability to communicate. When Ask Red told you it was a machine, for example, it came across like you were conversing with a person.

Are Live People or Robots the Wave of the Future?

The bottom line to this tale is that a good online customer service operation is tough to implement. The companies that do it well will win business. Those that do it poorly will lose business.

While implementing natural language systems may seem to be the least expensive alternative, in the long run they may turn out to be the most expensive unless they are implemented very carefully. If implemented poorly, furthermore, they will prove to be an embarrassment.

From my viewpoint, having a live human available for online chatting is by far the best alternative. Better yet, have both available. If customers cannot get the right answers from a well-implemented natural language system, they will be a lot happier if they can press one button to reach a live person.

What do you think? Let’s talk about it.