The Customer Service Survey
Self-Service Complexity
Thu - December 8, 2005 04:04 PM in
I learned a lot at EIG's IVR seminar earlier this week. One insight crystalized a point I had never quite been able to articulate before, but on reflection it's obvious.
Most of the interaction customers have with a self-service system is about task identification, rather than task completion.
In other words, a typical IVR or self-service customer service system might have twenty different things it can do: account balance, payment information, get schedules or tracking information, and so forth. In addition, if the call needs to go to a human, the system needs to choose the right group of agents to handle the call. There might be three or four (or forty) different agent skills, such as sales, technical support, billing and collections, to name a few.
Once the IVR has identified what the caller wants to do, the rest is easy. Send the call to the right agent queue, or ask for a couple pieces of information and complete the task.
But the customer doesn't know ahead of time how his task (for example, "I need to fix an error on my bill") translates to the company's internal structure (he needs to talk to a "billing resolution specialist").
So the hard part of designing a useful IVR is to figure out how to ask the right questions of the caller in order to efficiently and correctly decide why the caller is calling. This is made more challenging because most companies' internal structure doesn't line up with the way customers think about their tasks, and because humans can only process a handful of spoken choices at a time.
And the reason why so many IVR designs are so bad is because they don't respect the caller's perspective (for example, by using jargon or marketing-speak), and they don't respect the caller's limits (by offering too many choices at once, giving too much or too little information at a time, or asking questions which don't make sense to the caller).
Posted by Peter Leppik
In other words, a typical IVR or self-service customer service system might have twenty different things it can do: account balance, payment information, get schedules or tracking information, and so forth. In addition, if the call needs to go to a human, the system needs to choose the right group of agents to handle the call. There might be three or four (or forty) different agent skills, such as sales, technical support, billing and collections, to name a few.
Once the IVR has identified what the caller wants to do, the rest is easy. Send the call to the right agent queue, or ask for a couple pieces of information and complete the task.
But the customer doesn't know ahead of time how his task (for example, "I need to fix an error on my bill") translates to the company's internal structure (he needs to talk to a "billing resolution specialist").
So the hard part of designing a useful IVR is to figure out how to ask the right questions of the caller in order to efficiently and correctly decide why the caller is calling. This is made more challenging because most companies' internal structure doesn't line up with the way customers think about their tasks, and because humans can only process a handful of spoken choices at a time.
And the reason why so many IVR designs are so bad is because they don't respect the caller's perspective (for example, by using jargon or marketing-speak), and they don't respect the caller's limits (by offering too many choices at once, giving too much or too little information at a time, or asking questions which don't make sense to the caller).
Posted by Peter Leppik
Posted at 04:04 PM | | | | |