Years ago we simply walked into a store and met face-to-face with the person who sold us the product in question. We resolved our issue on the spot, and that was it. Then came the toll-free help line, which provided little more than another distribution channel for elevator music. The norm today is a recorded voice that instructs us to press an endless sequence of buttons on the telephone handset, followed of course by an encore of elevator music. 

The Internet is bringing a frightening new twist to the customer support saga: self service. Also known as surf service, we are now made to answer our own questions by wading through dozens, sometimes hundreds of help pages on a corporate Website. When we send an e-mail in the hope of reaching a real live person somewhere in the bowels of corporate America, we often get a robotic guess at the answer. After all, there’s a reason for the oxymoron "artificial intelligence." Occasionally, we receive a 10 page list of frequently asked questions in our in-box, which is tantamount to the previously discovered surf service. 

Know what you’re trying to accomplish. While all companies are, in principle, committed to providing great support, most ultimately focus on financial benefits. World class Internet support is defined as providing the right answer the first time in 24 hours. Decide if you want world class support, or if a simple autoresponse will suffice. Be honest about your dedication to quality, availability of budget, and timeline for results.

Clarify your mission by defining a simple service level. A service level is a quantitative measurement of performance. The most current framework is a certain high percent completed within a reasonable time period, for example: "95% completion in 24 hours." Avoid extremes such as 100% completion or average completion, since these can be easily biased and they don’t provide a complete picture of performance. Mission critical products and services will probably need a higher service level, for example 99% completion within several hours or minutes. 

Internet support has extreme cost advantages over conventional phone support—compare $10 for a typical phone-based incident to $2 for a typical Internet-based incident. Automated systems can be even cheaper if they reliably provide the correct answer the first time. Unfortunately, most auto agents are only accurate 30% of the time. The downside of automation is that if it doesn’t work well, customers tend to call in to the help line. That scenario not only misses the savings opportunity, but it incurs an extra expense in implementing the new, unsuccessful automated technology. 

Be aware of auxiliary expenses. Systems, training, IT support, ongoing management, foreign language support, and quality control can have serious budgetary impact.

This is the death trap for many. There are so many systems out there that you can easily spend a full year testing them, only to discover that new versions are out and hence the need for another round of tests. This phenomenon is known as "analysis paralysis." It sounds silly, but I’ve seen it happen many times. Give yourself a timeline for system selection—2 months should be enough—and stick to it. If you can’t finish it by then, your best option is early retirement. 

When selecting systems, choose a workflow engine to keep track of all those e-mails whizzing around. A knowledgebase engine is needed to keep track of the 200 or so most frequently asked questions, along with their updates. Both the workflow engine and the knowledgebase engine should interface with the Website to create a single support architecture. 

There are too many vendors to list here. My own advice is to keep it simple. Since most systems in this category are in version 1.0, be prepared to upgrade to newer and better systems every few months. Be sure to test systems against your particular needs. Whatever systems you choose, check that their reports adequately support your service level reporting needs. Ask about any capabilities that you will want in the near future, such as foreign languages, or allowing CyberReps to work remotely from home. If you are considering outsourcing, be sure that your systems support that.

A strong knowledgebase greatly improves the quality and efficiency of support. Invest the time to build a thorough list of common answers, and be sure to check the answers for internal consistency, spelling, and grammar. It’s wise to elicit input from other groups such as product development, sales, and even PR. 

Customer support is inherently labor intensive. Hire the best people you can afford. Simple mistakes, like accidentally canceling a valued customer’s account, can be costly and embarrassing. People remember the worst support mistakes, so hire people that are reliable and consistent. The best approach to hiring is to target responsible individuals who have an interest in the product or service at hand. Test all applicants for their knowledge of the product, the Internet, writing style, typing skills, and genuine interest in the role.

Beware Pandora’s box. Once you open the floodgates of Internet-based support, it’s easy to get inundated with thousands of
e-mails. I’m aware of one company that unexpectedly received over 100,000 e-mails in a single day! Be sure that policies, systems, and people are ready for life on Internet time.

You now have systems, staff, a knowledgebase, and pilot experience. Ideally, the only remaining challenge is to expand the program to accommodate the entire volume of incoming messages within the confines of the service level. If all previous work was done correctly, scaling up is easy. Focus on daily performance levels, and routine improvements to the knowledgebase. 

As luck would have it, the moment the first iteration of Internet support is in place and stable, the world will change. Someone in the organization will want to include foreign languages, additional products and functions, and the ability to allow CyberReps to work from home. New products will need new support, competitors will exert new pressures, customer expectations will rise, systems will crash, and the Internet itself will evolve. Be prepared to reinvent the Internet support program every 18 months. 

But some things won’t change. New products will still go to market before they are truly stable, and customers will still want personalized service. Let’s just hope elevator music never comes back in style.