1.800.323.3639 | Support | Training

Finding Your Voice in IVR

By: Avtex

Interactive Voice Response, or IVR, has become the preferred menu system for many contact center administrators due to its simplicity, effectiveness and value. When executed properly, IVR can streamline a customer’s interaction and intelligently route them to the support or resolutions they seek. This results not only in a more effective experience for the customer, but also reduced call volumes and agent workload. Unfortunately, when done improperly, IVR can have exactly the opposite result – leading to customer confusion and frustration, increased call volumes and stressed agents.

How can your organization get IVR right? While there are many factors to consider, IVR success all starts with the big three – professionalism, consistency, and clarity. Robert Wakefield-Carl, Avtex Sr. Director of Contact Center Innovation, explains.

Robert Wakefield-Carl:

It amazes me whenever I call into IVR systems the sheer level of variance of voices and types of playback offered to callers. You get male voices mixed with female, text-to-speech mixed with prompts and different quality between prompts and languages. The variation between systems can make it difficult for an organization to tailor the IVR directly to their brand personality, as they are often beholden to the system’s settings or architecture.

Since the IVR system is one of the first points of contact between the customer and the brand, it sets the stage for that crucial first impression. Failing to convey your organization’s personality and attention to detail through your IVR may result in lost or unsatisfied customers. This makes it crucial to make every effort to ensure that your IVR and prompts convey your personality and are professional, consistent, and understandable.

Focusing on those last three factors, professionalism, consistency, and understandability, will help you create a successful IVR. Let’s explore each of these in detail.


There is nothing worse than calling in to a company and hearing a mixture of different voices. I remember calling into one number and the introductory “hello” was a female voice, the menu was a different female voice, and the please wait message sounded like the technician that installed it.

It is easy to understand how this happens. First, the installers place temporary prompts in place for testing. Then, they ask someone like Beth in Marketing to be the “voice of our company”.  That is all well and good, provided that Beth continues to be available to record future prompts – if she leaves or gets promoted above these mundane tasks, you may have issues. Let’s say Whitney takes over for Beth. Rather than recording the entire voice prompt system in Whitney’s voice, the organization just asks her to record any prompts that someone asked to be changed.  Soon, you have prompts recorded in multiple voices with different qualities and no idea of what was captured when – this isn’t the best way to convey professionalism.

To convey professionalism, it is vital to maintain clear, consistent voice prompts – in the same voice and quality, throughout the menu system.

That is where companies like GM Voices come in.  They record prompts by professional talent that never fluctuate in their delivery, are perfect every time and consistently the same voice. They also keep extensive records of changes and a list of all the prompts for the system.   The only issue with most of these services is that you may not have access to record the system prompts like “one moment please” or “You Entered” or numbers like “First and Third”.  I PureConnect these are embedded into prompt handlers, but with PureCloud, you have easy uploading of recordings that make using professional services for system prompts even easier. Also, PureCloud gives you access to all the system prompts as well, so if you want the entire system in one voice, that is totally possible.


One of my pet-peeves is inconsistency with voices and menu selections.  I never know to press * or 9 to return to the previous menu.  I am listening to a list of options and they skip from 1 to 2 to 4 to 7 – now I have to actually remember what number to press instead of which selection I need.  Data entry in one place asks for a # and another does not – meaning that when I do press # the second time, the system does not understand my response.  Verification on one caller input is 1 and on another #.  When you add speech recognition it gets even worse – do I say ‘operator’ or do I say “representative.”

Probably the most frustrating part of these inconsistencies is how easy they are to avoid.

A visual mapping tool, like PureCloud’s Export Call Flow, gives users a comprehensive view of the menu system.  When you have it on paper, you can better see the inconsistencies in menus, selections, and options.  One overlooked feature of PureCloud is the ability to print call flows, giving you a nice review method to pass around and documentation for future revisions. Few take advantage of this feature, but it is something that every company should use once they have a working call flow that everyone has agreed to.


Recording quality and delivery is important when it comes to understanding a prompt – especially on a cell phone. But, recording quality isn’t the only factor impacting your customers’ ability to clearly understand the IVR menu system – diction, technology and other factors also play a role in the clarity of the system’s message.

For example, using Text-to-speech (TTS) is always hit or miss because acronyms, proper names, alternate pronunciations, and other strange aspects of common speech cannot be properly spoken. Tuning TTS is imperative if you want callers to actually understand what is spoken. For instance, typing IBM would be spoken as ‘ibem’, but if you simply type in I B M with spaces, then it will be read properly. Proper names are always hard, especially last names. When asking a computer to read back a name, always remember that TTS engines don’t have the 12 years of schooling necessary to know when a ‘g’ is silent or when a vowel is long or short. When implementing TTS, try to reduce proper names down to the least number of letters that make the same sound of the name. If you are in doubt, find a music major who took diction – they learned the hard way how to pronounce properly in multiple languages.

PureCloud has a great interface for using TTS in multiple languages. With prompts language-agnostic, you simply keep track of a single prompt in any language you want. TTS is becoming more and more human-like and that means that in the near future, you could create an entire IVR with TTS and have translation text for instant multi-lingual IVR.  PureCloud gives you the opportunity to Bring Your Own Voice, or BYOV to the platform, meaning you will be able to use voices from AWS Polly, Acapela Voice, and others. With this and machine translation like AWS Translate or Google Translate, you have all the tools necessary to create a clear, consistent, and professional sounding IVR that is easily updated and maintained.

The following two tabs change content below.


Latest posts by Avtex (see all)


Let’s Talk About CX