Budget slashed? Creating a cost-efficient IVR your customers will love
Can cost-effective automation really provide value over human agents?
Common wisdom in the contact center space is that human customer service is by its very nature better than automation. But is that always the case? What if you could create an Interactive Voice Response (IVR) that your customers love while providing a better experience for both the caller and the agent?
Here’s the dirty little secret in the human vs automation debate: what customers care about is getting their issue resolved quickly and accurately the first time. Most don’t care about the method, as long as the brand is one they can rate as easy to do business with.
So how does a brand create an IVR that will leave a customer pleasantly surprised at the ease of doing business while leaving the call center manager happy at the cost savings of doing business? There are two key functions of the IVR that can be technologically addressed to accomplish those goals: knowing who the customer is and what she is trying to do, also known as caller identification and caller intent.
1. Caller identification
The first key to an easy customer experience is identifying the caller. The challenge is that every business has a different way of positively identifying the caller, a system we refer to as “matching,” since the caller is being matched with a record in the organization’s Customer Relationship Management (CRM) system. Elements such as the ANI (the number the customer is dialing from), caller name, email address or member number are common identifiers for positively identifying a customer in the organization’s CRM system.
- ANI lookup While it’s easy to automatically look up the ANI, match rates for ANI-only lookups tend to be quite low for a number of reasons. The caller could be calling from a different phone, such as a VoIP phone or an office phone that doesn’t happen to be the phone number listed in her customer record. So a second piece of information is typically required to positively match the caller with her record in the CRM database. But each additional piece of information has its difficulties in terms of IVR collection.
- Name First and last names can be difficult to hear and spell correctly via most speech recognition systems, leading to repeated questions, caller frustration and opt-outs
- Email address Email addresses are notoriously difficult to accurately collect via speech recognition systems, once again leading to repeated questions, caller frustration and opt-outs
- Member ID Numeric member numbers are ideal for IVR data collection. However, most customers do not have their number memorized, so they tend to skip this step or opt out.
Addressing name and email address matching How can the challenges of name and email address matching be addressed? Our answer is through technological innovation. The Spoken Smart IVR adds a human in the background to make pinpointed corrections to traditionally difficult-to-understand caller utterances (such as name and email address). Through a patented software interface, corrections can be made on up to 10 simultaneous calls to caller utterances where needed. The human sits at a dashboard like the one below and only listens briefly when a tab flags red or yellow:
The result is a dramatic increase in the caller identification match rate, which is the cornerstone of a personalized caller experience and accurate call routing. And here’s the proof for one customer:
- Average Handle Time was reduced by 12%, since the IVR was doing the heavy lifting of caller identification
- Self-service rates increased to 73%
2. Call routing
The second half of the formula for an easy and elegant IVR interaction is accurate call routing. This is typically based on a combination of the caller identification and of the caller intent. The better the caller identification match rates, the higher the likelihood that the call will be routed to the right agent accurately, the first time.
Let’s look at an example. A customer call in to an organization and asks for technical support. However, from a combination of the ANI match and the email address, the CRM record indicates that the caller isn’t entitled to technical support over the phone. So instead of routing the caller to the technical support queue, forcing the agent to do a manual look up and having to deliver the unpleasant news that the caller isn’t entitled to live technical support, a better solution would be to route the customer to either a Frequently-Asked Questions (FAQ) self-service queue to deal with a common question or to a sales queue so he can purchase the support he needs.
And voilà! Misrouted calls decrease dramatically. In the case of Neat:
- Misrouted calls were decreased by 30%
It isn’t machine automation that is the enemy; it’s poor IVR construction and poor speech recognition. And we believe that we’ve solved the issue of poor speech recognition with the Spoken Smart IVR: by adding the human in the background working on the software interface to make pinpointed corrections to caller utterances, the accuracy of caller identification match rate and caller intent will increase, resulting in cost savings all around.
Want more detail? Download the full case study here: