What Is A Smarter IVR?

Nicholas Luthy is Services Consultant for Interactive Intelligence’s PureCloud platform.
Nicholas Luthy is Services
Consultant for Interactive
Intelligence’s PureCloud

Interactive voice response (IVR) systems are only as intelligent as the information that they’re provided. The deployment of a smart system is a blend of art and science. The primary trait of a smarter IVR is that it contains a vision versus a sequential confederation of features. Smart IVRs are never complete: Just as your agents require ongoing investment, training and development, so does your IVR.

So how do you create a smart IVR? While the solution is elegantly simple in concept, the execution takes progressive restraint, time and corporate investment. You create a smart IVR the same way you create smarter employees. A new employee is an ongoing investment of energy upon which an iterative platform of knowledge and skills is built. Who do you want your IVR to be when it grows up?

An Argument for Humanity

It may help to humanize your IVR system in a way that separates it from other technologies deployed. Your IVR has (or should have) the capability to present itself as a semi-sentient human replacement. While some may view developing a vision and persona for an IVR to be a trite or inauthentic approach, you’ll see the value as you begin implementing various features and functionality. Thinking of your system as a human entity will draw stronger correlations to processes already palatable to the organization—specifically around agent education. As an industry, we train our agents, capture samplings of performance, and score their overall quality. The output of that evaluation is actionable whether a refinement of script, or additional training. So it should be with our IVRs, we should continually review performance, finetune behaviors and “train” new features. As your IVR’s persona develops, it could even expand beyond the immediate organization into marketing and published materials.

The IVR persona could be a great way to connect to and acknowledge your customers by building a profile that reflects their own characteristics. If your average customer or consumer tends to be a female millennial, you may want to develop the persona of “Ashley,” which, according to Wolfram Research’s Alpha computation engine, saw a tremendous peak in naming frequency in the late 1980s and early 1990s. Ashley may only speak English, but offers SMS interaction as a replacement for completing the current inbound voice call. Whereas, if you are focused on mid-level business men, you could select the persona of “Robert,” as an example. Robert may not be as inclined to use SMS, but he definitely speaks multiple languages. Additionally, you may find that speed and brevity are key when presenting to a specific audience, in contrast with more in-depth prompting and informational scripts.

Investing in a persona allows you to humanize the process of deploying, updating or enhancing your IVR. The persona can guide you through the decision processes while maintaining the focus on your intended audience.

Defining the intended role for your IVR and expectations of your persona is critical. Oftentimes, IVRs are viewed as unwieldy, challenging to navigate and outright unhelpful. You can avoid these types of problems by developing and communicating the correct vision—and maintaining that vision throughout your IVR’s lifecycle. To be effective, your vision should:

  • Be measurable and allow for incremental feature availability

  • Be concise

  • Align with the corporate vision

I’ve been involved with more than just a handful of IVR implementations, and one of the major challenges is that—for reasons primarily related to cost—the IVR is designed upfront in one fell swoop. This is often a problem because it delays the immediacy of feedback from users, business stakeholders and others. In some cases, the feedback can be a painful realization that the implementation needs to start over.

Earlier feedback could alleviate such pain. If an IVR is built out in smaller chunks and sprints, then it is easier to schedule development, provide status, present progress and receive feedback, and apply fixes. If an IVR is approached in this manner, it also ties in with the theme that we are building the skill sets of our IVR persona. For instance, a sprint may be adding an additional language to the entire application.

Practical Pointer: I find that IVR reporting is not well understood, and it is seldom acted upon. In reality, everything an IVR does should be monitored; more specifically, everything actionable that an IVR does should be monitored. Knowing how your customers are interacting with you is critical to the future development and maintenance of the IVR. Analysis of that information

can provide trend data that drives new features, corrections or enhancement to current features, rescripting prompts to add clarity, and so much more.

Data, Who Has My Data?

A critical component for improving the overall intelligence—or at least the perception of intelligence—of your IVR is to provide it with the most accurate and comprehensive data possible.

There are companies that specialize in delivering user information for a price—which not every company is willing to pay. Forgoing sizeable purchases of information from big data, there are a number of avenues for any company to extend its data reach and they may make more sense than purchasing large quantities of expensive information.


The first step in obtaining a more complete customer picture is understanding what information you currently possess about the customer, and what would be additionally beneficial to know. You may be surprised to find that the list of unknown information is known by someone else in your company. Perhaps marketing or accounting has a separate database or CRM. Connecting to these data sources is essential, not only to know more about the customer, but also to ensure that the right hand is speaking to the left.

Keep in mind that, while an IVR application or a web process may be a linear step-by-step affair, the collation and mining of data about your customers doesn’t need to be so asynchronous. If you are constantly monitoring your users and populating a summarized dataset, the IVR can connect to that data to make any of a number of smarter decisions.


One of the challenges that we have in the realm of IVR functionality is that initially the only piece of data known about a caller is the number they are calling from—or their ANI. Some IVRs solve this by asking the caller for information that is cumbersome and rarely a value-add to the caller themselves. How then, with only one data point, can a wealth of information be unlocked prior to the caller even hearing their first prompt? Luckily, there has been an emergence of online resources that allow the querying of data with relative ease, and with which we can start to build a surprising profile of information. Would it be helpful to know someone’s home address, spouse’s name, property value or even an approximate household income within a margin of error? This is all possible with public and semi-public information if you know where to look and how to use it.

smarter_ivrAs an example, an IVR could communicate with a data-as-a-service vendor (for a small fee) or use any of a number of free data sources, to query the home address based upon a phone number. Then utilizing the street number and zip code, the IVR could query public tax records posted online to obtain an assessed property value. From the assessed property value a ballpark income estimate can be made—which admittedly could easily be skewed by the emergence of additional data—with the assumption that the mortgage payment front-end ratio averages just shy of 23% of the gross monthly income; typically the maximum is 28%. This method may not be accurate enough to file your taxes over the phone, but it can help build an initial profile to categorize your callers.

Ballpark Income = (Assessed Property Value/12) x ( 1/.228)

In addition to the possibility of delving into profiling a caller’s personal finances, the same input information—the caller’s phone number—could be used to determine whether they are calling from a landline or mobile phone. If the caller is on a mobile phone, various lower-cost deflection techniques could be utilized in order attempt to self-service the caller. Should the caller reach an ACD queue, the customer’s engagement could be transitioned to a series of SMS messages rather than be assigned an agent. Various messages could be passed back and forth prompting a scripted interaction with much more predictable projection of cost per interaction than speaking to an agent. Alternatively, the caller could be messaged a URL to a web or mobile portal to complete their interaction in a self-service manner.

In my two previous examples, data is retrieved from public and DaaS sources based upon the phone number. Social media acts as an additional information source and allows organizations the ability to query information ranging from a limited set of data to a complete profile of information often by providing only a user’s name. If your customer is a known entity within your dataset with a full name, a quick query of public posts and comments could be performed to see if your business organization was mentioned recently. The IVR experience could be greatly tailored to the caller individual needs in the event information was or was not found.

Tailoring the Experience

I mentioned earlier that an IVR is a mixture of art and science. There is indeed a logical approach to obtaining information about a caller based on minimal input and educated extrapolation. However, knowing just what to do with the information once you have it is where the art takes over. This is going to drastically vary businessby- business, but it should be used to offer a dynamic IVR experience that heavily leverages all available information to deflect or reduce the volume of calls actually delivered to agents, while still servicing the needs of the consumer.

The user’s informational profile can be leveraged to customize not only the options presented, but the voice and persona of the IVR, the communication channels available to a user, risk-based assessment for caller priority, targeted scripting, and other applications.

Please feel free to contact me with questions, comments, or concerns with the ideas that I’ve presented here. What techniques have you employed to make your IVR smarter? I enjoy the discussion immensely.

Nicholas Luthy is Services Consultant for Interactive Intelligence’s PureCloud platform.

– Reprinted with permission from Contact Center Pipeline, www.contactcenterpipeline.com

Related Articles

Back to top button