61% of customers are frustrated with your IVR — here’s how they can help fix it

Most customers have felt the immediate sense of disappointment when a voice bot opens with a friendly, "How can I help you?" only to reveal it can only understand a few pre-programmed keywords.  

This sort of experience is common when traditional Interactive Voice Response (IVR) solutions fail to direct a customer in the right conversational direction — leading to confusing and unnatural interactions as customers struggle to figure out how to respond. 

There’s nothing inherently wrong with IVR call routing. In fact, there are still specific use cases where IVR can facilitate customer journeys more effectively than open-ended, AI-powered voice bots. But when the experience doesn’t align with the expectation, it breeds frustration. This immediate CX broken promise is a key reason why a staggering 61% of customers agree that IVR systems contribute to a poor customer experience. 

As a result, customers have been conditioned to speak to automated systems in a stilted, unnatural way. For years, this was a functional, if clumsy, workaround. But here’s the problem: it leaves immense opportunity on the table, especially in an AI era where natural, fluid conversation is more possible than ever. 

As conversation data and user intents become increasingly valuable currency for maximizing and streamlining the customer experience — powering everything from real-time recommendations to conversation intelligence strategies — a new challenge arises. How can organizations build modern Intelligent Virtual Assistant (IVA) experiences that retrain customers to give these smarter bots the information they need to facilitate higher-value interactions? 

Why IVRs are a problem worth fixing 

Encouraging customers to respond more conversationally to IVA isn't just about improving sentiment or satisfaction; it's a powerful business imperative. Getting the right information in as few back-and-forths — or turns — as possible allows the IVA to either resolve the issue quickly or route the customer to the best-fit resource the first time, keeping customer effort and operating costs to a minimum. 

Consider the data: 

Across industries, the average containment (or success) rate in a traditional IVR system hovers around a mere 30%. Every transfer to a live agent drives up costs and injects friction into the experience. But each back-and-forth turn of an automated conversation drives up the risk of the customer dropping off by 7%. So being able to minimize those conversational turns within the bot means fewer people "zeroing out" to an expensive human agent out of sheer frustration.  

Imagine if you could drive your containment rate up by 10% or 20%? The bottom-line impact of an improved IVR experience can be massive. 

Using conversational AI to drive more than CX cost savings 

Beyond the cost savings that can come from boosting containment, conversational AI can drive revenue generation too.  By combining conversational insights with customer data, organizations can predict and recommend personalized next best actions, enabling new use cases previously out of reach for conversational AI. 

For example, interactive shopping guidance is a relatively untapped retail strategy where automation can truly shine. Whether it's assisting with product research, or making targeted recommendations, modern conversational AI solutions can leverage 360-degree customer data to deliver hyper-personalized experiences without ever involving a live sales agent. And with access to multi-dimensional catalog data, bots can sift through thousands of product options to outperform even the best salesperson.  

When this heightened processing power is wrapped in a flexible persona that can dynamically adapt to the conversation, this gives you a scalable army of sales agents who are constantly learning from each conversation and sharing their learnings with each other.  

Leveraging conversational AI to efficiently help customers find the perfect product in real-time can be the difference between converting them on the spot or losing them to a competitor.  

Conversation design strategies to maximize IVA value 

So, how do you guide customers from exchanging just a few keywords with an IVA to engaging in more natural, comprehensive conversations? 

1. Set conversational expectations early 

• Why it matters: Users often have no idea what the bot can or can’t do. They carry baggage from past negative experiences and assume this new interaction will be just as limited.

  When met with a traditional greeting, they'll tend to respond using the same stilted language they've used in the past. These simple back-and-forths fail to unlock the full potential of the conversation for both the customer and the business. 

• How to apply: Use a brief, friendly onboarding message to explicitly state the bot’s capabilities. A simple change can make a world of difference. For example: “In your own words, let me know what you need. You can say things like, book an appointment, pay my bill, update my account, and so forth.” This style of prompt provides helpful guidance while encouraging a more natural response than simple keywords. 

2. Prioritize accessibility for all 

• Why it matters: Your IVA needs to be able to adapt to different abilities, backgrounds, and attention spans. Failing to train your conversational AI to use simple, clear language can alienate users — and the threshold for losing their engagement is much lower when it comes to automated conversations. 

  Also beware of cognitive overload. Working memory for audio processing is generally very limited, and customers are rarely able to dedicate their full attention to a service interaction, especially over the phone. It’s surprisingly easy to get overwhelmed by long, complex prompts, especially if they’re played in quick succession without enough pauses in between.  

• How to apply: It's critical to take the time to research your caller population, including their speech patterns, how they engage with your agents, and their conversation preferences. Make sure the pacing and intonation are clear, especially when relying on automated text-to-speech prompts. 

  For starters, look for any customer profiles from branding or marketing teams that you can use as a starting place for your IVA persona. From there, review agent transcripts or customer panels and surveys to get additional context into the ways your customers refer to products and services, as well as what might not be working for them. Ruthlessly simplify the language in your automation, using online reading-level checking tools to align with your customer base. And invest the time to review, test, and optimize the experience with a variety of representative users before exposing it to your customers. 

3. Design your IVA for recovery 

• Why it matters: We often don’t notice it when talking with other people, but there are lots of “disfluencies” that come up constantly in conversation: we talk over each other, misunderstand something, or need a word or phrase to be repeated. In general, we’re able to naturally smooth these over quickly, thanks to our instinctive recovery mannerisms.

  Even today, automated conversation partners are not quite so sophisticated, and they’re more prone to disfluencies since they’re slower, have a more limited vocabulary, and can’t (yet) recognize important non-linguistic info such as cues about the user’s emotional state or confidence level. Even though we do our best to avoid disfluencies during the design process, they’ll always be inevitable, so we need to make sure we’re handling them in ways that tap into people’s natural mechanisms for recovery. Otherwise, we risk escalating the problem and driving our caller to a human agent. 

• How to apply: Accept that errors are likely to happen and build graceful error handling and intelligent guidance into the conversational AI's persona. Don’t blame the user OR take too much responsibility. Over-apologizing erodes confidence, and messages like “It looks like you’re having trouble” can be perceived as condescending.  

  When errors do happen, it’s important to give the user a quick audio cue, coupled with helpful info to drive towards a successful repair, such as restating the last message in more detailed language. For example: “Hmm, I didn’t catch that. In just a few words, tell me why you’re calling today – for example, check my bill, make a payment, or book an appointment.” This refocuses the interaction without making the user feel like they’ve somehow failed, or that they’re wasting their time talking to a dumb, disrespectful robot. 

Bonus consideration: Is conversational AI even the right solution?  

Of course, at a more foundational level, you’ll also need to decide whether your proposed IVA use cases are a good scenario for conversational AI at all. Deploying new technology for new technology’s sake is often a recipe for misaligned experiences customers didn’t ask for and won’t want to engage with. As well, some intents are simply better handled by a live agent — especially when empathy, sensitive data, or complexity are involved. In those cases, routing straight to a human may deliver a faster and more satisfying experience.  

In today’s AI landscape, offering a sub-par IVR doesn’t cut it. By designing AI-enabled experiences to feel intuitive and fit seamlessly alongside other important touchpoints, brands can restore trust, reduce friction, and unlock new value from every interaction.