IVR applications can solve many problems for contact centers and enterprises alike. By automating a company’s most repetitive calls, users (whether they are customers or employees) can access information in a quick and easy fashion without being held in a call queue. Also, the average call length of an automated call is much shorter than an agent call. However, all of the benefits of IVR are lost if menu options, survey questions or requests for information are too complex. As we discussed in a past post about VUI (voice user interface) design, the call flow or caller user interface must be designed in a user-friendly fashion in order to realize the benefits of IVR technology.
Keeping the call flow simple is the best way to minimize caller frustration, improve containment rates and decrease abandoned calls. Keep the menu options short. If there are a number of options offered to the caller, the first set of options should be limited to five selections. Additionally, the most frequent or routine caller requests should be presented as the first menu options.
Wording choice is also an important part of designing a caller friendly IVR application. Wording of instructional prompts and requests for input should be simple. Stay away from using complex or specialized language. Consistency is also important. IVR applications should prompt callers to enter data or navigate menu options using similar phrasing as not to confuse callers. Using consistent phrasing also lets the caller know what to expect and how to interact with the IVR, which makes for a more user-friendly caller experience.
If shortening the call length of a typical agent call is an objective, only use automatic speech recognition to collect data that can’t be entered by a caller through a touchtone key pad. Examples of the type of data that is difficult to collect through touch tone would be a caller’s name, address, email address or product names. Using speech recognition to allow callers to navigate through simple menu options can add an element of sophistication to a call flow. However, it is not necessary and can sometimes lead to caller frustration. If the speech recognition engine does not understand the user’s utterance due to a loud calling environment, or other limiting factors, callers are likely to opt out of the IVR to speak with an agent. Speech recognition is not a perfect technology and does not always understand a caller’s spoken commands on the first try.
There is a number of best practices that one can apply to the design of a VUI. Many are dependent on the actual call flow and the purpose of implementing an IVR solution.