Best Practices In IVR
An Interview with Behavioral Science Associate Walter Rolandi, Ph.D


How have best practices in IVR design changed in recent years?

I think there is good news on two fronts. First, the use of DTMF-based IVR best practices is proliferating. Keep in mind that these best practices are far from new: many have been well understood for decades. The persistent problem has however fallen between education and adoption.

What do you mean when you say education and adoption?

Today, best practice educational materials are available through a number of seminars, publications and Internet resources. At the same time, market effects are now driving adoption. What we are seeing is that the emergence of “better” DTMF based IVR systems in the market is stimulating corporate concern over the old “press one for this, press two for that” legacy systems that we have all come to abhor. As more and more callers experience the advantages of well-designed systems, increasing pressure is being brought on those responsible for maintaining the status quo. All this has contributed to more progressive scripting practices.

Second, scripting practices have been revolutionized by the introduction of extremely powerful speech recognition engines. The state of the practice in speech-based voice user interface design indicates that its best practices are still evolving yet this is a high stakes and rapidly evolving field. All in the space of just 6-7 years, speech VUI design community has tacitly adopted some underlying design standards.

Could you give us an example?

Sure. Many disparate speech applications, all developed by different vendors, now exhibit similar error recovery and timeout strategies.

Overall, scripting practices are rapidly improving and applications are accordingly becoming more effective. On the other hand, there is much to be done. If you were to call a random sample of twenty nationally deployed IVRs, you would rightly get the impression that scripting practices still often ignore even the most basic and recognized best practices. I have performed such studies for clients and have repeatedly found this to be true. . Most of us can confirm the fact based solely on our personal experiences with IVRs.

What's the ideal number of menu options? I've heard it's 5 but a lot of companies are still violating this rule. Why do you think this is?

Well, first let me address the “magic number” question, then on to the question of “why”. I’m not sure where you heard “five” but I suspect that it could stem all the way back to George Miller’s classic 1956 study, The Magical Number Seven, Plus or Minus Two: Some Limits on our Capacity for Processing Information. In this study, Miller documented the fact that most people can “process” only about seven items of information at once. Depending on the type of information and the individual being tested, sometimes the ability is limited to five but can be as high as nine. In any event, most research in this tradition focused on the ability to recall or respond to lists of visual stimuli. As it turns out, the limitations of short-term memory are even greater when the stimuli are auditory and sequentially presented. While five item menus may be common in the industry, I try to limit menus to just three possibilities.

What are your recommendations on setting up a well-designed menu system?

A well-designed menu system should lead the user into the two or three most frequently requested tasks and then branch accordingly as needed. Note that limiting the number of options that are presented becomes increasingly important for older users and non-native speakers of the language.

What are the top 5 design mistakes you see people make?

In the DTMF world:
  • Menus too long
  • Illogical or confusing presentation of information
  • Jargon
  • Ineffective failure recovery
  • Ineffective policy for accessing human help

In the speech world:
  • A tendency to focus on entertaining the user at the expense of serving him
  • Crude attempts at conversational dialog and setting unrealistic user expectations
  • Persona “bologna” or excessively animated, human-like interfaces
  • Ineffective failure recovery
  • Ineffective policy for accessing human help

What are the costs of getting it wrong?

I gave a speech a few years back, which I began by rhetorically asking that same question. I answered the question by just playing a recording of someone slamming down a telephone handset!

But more specifically, some of the consequences of faulty IVR design are:
  • Frustrated, angry customers
  • Failed transactions
  • Increased hang-ups
  • Greater user errors
  • Increased call durations required to complete transactions
  • Increased cost due to greater transfers to Customer Service Reps

Overall, these types of customer experiences are frequently shown to lead to reductions in customer satisfaction. This has implications for the company brand name, its products and services.

What are your recommendations for getting it right? Any tips for
do-it-yourselfers?

First, avail yourself of the many fine and easily available resources on industry best practices. Second, I always encourage my clients to experiment and to create their own designs. Third, do some sort of usability testing prior to coding or minimally have your design reviewed by an expert. Designs that are already in production can benefit from an expert review. Such a review can accurately pinpoint user problem areas and identify where to make design changes, and what changes or enhancements are necessary to improve the customer and system interaction. I find myself reviewing other peoples’ designs more and more frequently. This is by far the least expensive way to ensure high quality design and IVR effectiveness while at the same time, bringing the know-how into the enterprise development group. Clients don’t want to be indefinitely reliant on external resources and expert reviews provide an effective means to transfer VUI expertise in-house.

Any tips for scripts that will be translated into other languages?

Yes. You can always associate with a company that specializes in product internationalization but I have found that using more than one native speaker, preferably including one who has a writing or publishing background in the target language while still in the scripting phase can ensure high quality implementations. For instance, the first pass at scripting can be attempted with the help of one native speaker who is not particularly skilled or trained in IVR design. However, subsequent design iterations should include another native speaker who can bring a cultural or literary sense to the table. I’ve found this combination of “novice” and expert to be very effective.

How can people get in touch with you?

I am currently collaborating with Behavioral Science Associates, Inc. (BSA) in offering these kinds of services. BSA is the first VUI specific consulting company directed by experienced Ph.D. psychologists with direct hands-on technical experience. BSA specializes in the dialog design, application development and evaluation of voice user interface systems. If you are interested in having your IVR system evaluated please contact me for more details. I can be reached through BSA at 800-722-8127, or send an email to services@bscience.com. For more information, please visit the BSA website at www.bscience.com.

To contact The Great Voice Company, please call 201-541-8595 or email at sales@greatvoice.com

Get Great Voice Overs

Voice Over Recordings & Voice Prompts in any language for IVR Phone Systems, Auto Attendants, Music On Hold, Corporate Narrations, Commericals, e-Learning, Audio Books, Web Audio and much more..

Follow Us

Follow our activity on the web and get to know us.

Contact Us

The Great Voice Company

110 Charlotte Place
Englewood Cliffs, NJ 07632

Phone: 201-541-8595
              800-333-8108

info@greatvoice.com