Press # for More Options:  The Value of IVR Usability Testing

from John Wooden, Director of Usability Services

on October 04, 2010

2 comments

Many of us have had at least one frustrating experience with an interactive voice response (IVR) system – getting lost in a maze of menu options, never hearing an appropriate option, never being offered an option to speak to a customer service representative, arriving at a dead end, getting cut off during a transfer to a representative, and so on. My father ended up shouting at an IVR system with voice recognition because it kept saying, “I’m sorry, I did not hear you. Please choose from the following options …” Eventually, he just hung up.

IVR system usability has not received nearly as much attention as Web usability, and perhaps it’s no surprise that over the years IVR systems have collectively developed a bad reputation. This doesn’t mean there are no good ones, but if you ask people, most will tell you they don’t have a favorable impression of them. Instead of being perceived as useful tools for self-service, they are commonly thought to be obstacles deliberately placed between customers and a live human in an organization’s customer service department. In 2005, Paul English was frustrated enough with his IVR system experience to publish “The IVR Cheat Sheet,” which listed the codes that would allow a caller to speak directly with a representative in dozens of companies.

If an organization’s primary objective in having an IVR system is truly not to block customers from speaking to an agent or representative, but rather to try to provide a good automated self-service experience, then it needs to take IVR usability seriously, just as seriously as it takes the usability of its websites and applications.

In most cases, this means conducting usability tests. The methodology for usability testing IVR systems and websites is essentially the same in most respects. You need representative users from your main user groups, a list of task scenarios and key questions to ask, a quiet place to test, and for IVR testing, a phone with a speaker. (You can test an IVR script before it is recorded simply by having the facilitator read prompts and asking the tester to describe which options they would select.) We also use Techsmith’s usability testing software to record the calls (with tester permission) and to capture tester actions and feedback. By observing testers and listening to their questions and comments, usability analysts can learn what is working well in an IVR system and what needs improvement.

For example, in a recent test of a government IVR system, the overall feedback we heard was positive. This was already a relatively straightforward system to begin with – most of the prompts involved a simple binary choice: 1 for yes, 2 for no. Still, the test revealed dead ends in the menu, some common misunderstandings of prompts and transitions, and issues with the password process. And so now after testing, this organization is able to make their system even better and thereby reduce the number of callers who want or need to speak with a representative.

Of course, the major design constraint of any IVR system is that it is primarily an auditory medium. IVR systems require users to listen – often closely – and each option must be presented sequentially, which places a load on the user’s working memory. In contrast, the web is primarily a visual medium that can use layout, color, font size, text, and images to organize and present information. And links can offer a user tremendous navigational control and flexibility. The ability to see information and control pacing and progress are the key reasons why more users prefer doing self-service online than through an IVR system.

Still, even within the constraints of an IVR system, it’s possible to provide a good experience by following some important guidelines:

1. Keep prompts and menu options short – no more than 4 to 5 items.


2. Avoid deep menus.


3. List the most requested or important options first.


4. Ensure that prompts are clearly distinct from each other.


5. Provide an option to repeat a prompt.


6. Ensure that prompts and instructions are clearly enunciated and read at a moderate pace.


7. Describe the function, and then the key to support it (e.g., to do X, press 1; to do Y, press 2).


8. Use “press” for single key inputs and “enter” for field inputs.


9. Use terms that are familiar to your users, and use the same terms consistently in all prompts and instructions. Avoid jargon and technical terms.


10. Ensure that the sequence of prompts is logical (e.g., from the general to the more specific).


11. Provide iterative coaching – use examples for error and timeout re-prompts. Avoid using “Invalid Entry” as the only response to missing or incorrect entries. Provide more information on what is required by the system – do not leave the caller at a dead end.


12. Where applicable, allow the user to recover from errors by means of a series of yes/no prompts to confirm an intended selection.


13. When a user has completed a process, provide confirmation. Give the user the choice to return to a main menu or to end the call, as opposed to automatically ending the call.


14. If an IVR process requires the user to speak to a customer service representative (CSR), indicate this near the beginning of the call. Always inform the user when the call is being transferred. If the user has a choice about whether to speak to a CSR, let them know.


15. Provide the option to use the touch-tone keypad as an alternative in any IVR system that allows voice responses.


16. Choose an appropriate persona for the system voice and use professional voice talent. Generally, users consider male voices more authoritative and female voices warmer and friendlier.


17. Ensure the system goes through usability testing.




Note: Many lists of IVR system heuristics provide similar guidelines. One especially useful source that I consulted for this entry was Bernhard Suhm's article, "IVR Usability Engineering Using Guidelines and Analyses of End-to-End Calls," in Gardner-Bonneau and Blanchard's Human Factors and Voice Interactive Systems, 2008.

Comments

Excellent blog topic. I try to adhere to the guidelines, but there is always room for improvement on an IVR.

by Bruce Mielke
on November 11, 2010 - 2:58

---

HHIS I slohud have thought of that!

by Kaydi
on April 30, 2011 - 1:54

---

Add Comment

Name:

Your e-mail address:

Message:

  Remember my personal information

  Notify me of follow-up comments?