The Customer Service Survey
I'm Not a Consultant, But...
Fri - July 28, 2006 02:13 PM in
I received a call this morning from a VocaLabs client, asking if I could meet with a prospect of theirs about what we've discovered regarding consumer acceptance of Speech Recognition technology. The customer is contemplating a major system installation that includes a speech platform, and has asked for some outside reassurance that customers will be better served.
We've done hundreds of usability tests and received feedback from many thousands, but I again explained that VocaLabs isn't a consultancy. We do not advise based on our findings surrounding system performance. Rather, we present the results, and it is up to others to choose what to do with the data.
I went on to say that the flip side is that it is impossible to have the above experience and not form some strong opinions; that I would be more than happy to meet with the client, but that I would not set VocaLabs up as experts in customer service automation design, best practices or anything that gave the appearance of favoring one vendor.
They agreed, and stressed that the independent third party view was exactly what was needed, and no issue, even if negative, was taboo. So I've accepted, and a call came in a few minutes ago saying that a meeting is being arranged for next week.
Here is what I will tell the prospect. It is advice applicable to anyone considering the installation of a large Speech Recognition customer service application:
Posted by Rick Rappe
I went on to say that the flip side is that it is impossible to have the above experience and not form some strong opinions; that I would be more than happy to meet with the client, but that I would not set VocaLabs up as experts in customer service automation design, best practices or anything that gave the appearance of favoring one vendor.
They agreed, and stressed that the independent third party view was exactly what was needed, and no issue, even if negative, was taboo. So I've accepted, and a call came in a few minutes ago saying that a meeting is being arranged for next week.
Here is what I will tell the prospect. It is advice applicable to anyone considering the installation of a large Speech Recognition customer service application:
- If you build it with a focus on saving overhead versus serving the caller, it will do neither. But if you build it with a customer service focus over cost control, it will do both.
- Humans actually prefer self service over speaking with an agent if the application serves their needs well. It is not "talking to a computer†that people object to; but badly designed automation that we all love to hate.
- Have an intra company meeting to get all departments to buy in and have input on the design. But then have one leader to get it done so as to avoid the "too many chiefs" problem that leads to mediocrity.
- Trust the designers. They have a good handle on best practices, how to phrase instructions, the impact of the voice persona the computer will use, etc.
- But, getting a second opinion on design issues is never a bad idea. Voice User Interface designers (VUIDs) at both Nortel and Nuance are top shelf, but they are also engineers at heart and this can lead to problems. (We've seen too many instances of clever designs, that didn't focus first on caller needs.) If pressed, I will recommend a particular consulting firm I consider the leading experts in self service automation best practices and usability design.
- Test, test, test. During the design phases, at pre-launch, and regularly once the system is operational. Things change. Over time user expectations rise, new products are sold, new marketing strategies unfold; the system is adjusted, and eventually service and caller satisfaction drops, and no one is quite sure why. We call it "application drift" and the best solution is a constant test-revise-test again plan that keeps the system performing.
Posted by Rick Rappe
Posted at 02:13 PM | | | | |