Speech IVR With Artificial Intelligence Is The Bees Knees

A short time ago I was approached by a friend of mine who is working with a startup called SmartAction.  He asked me if I was interested in a demo of how they are building  a new way to do Customer Service.

You know me, I am sucker for those things – so of course I said yes.

SmartAction has taken a different approach to solving the problems of Speech IVR: using Artificial Intelligence technology to make the navigation faster and more accurate.  If you ever used or implemented Speech IVR you  know that navigation is not the best feature they have – for some reason most of the providers think that simply replacing key punches with similar words is sufficient (hint: it is not).  Even if you get past speech recognition (which by itself can be troublesome when you are fast-speaking analyst with an Argentinian accent), understanding what you spoke is limited to words and simple phrases (what we call guided speech recognition – if I know the words you will say, I can match them faster).  If you stray from them, the system can probably recognize the words you spoke, but not the meaning. I remember one of the first implementations I used of this technology, I tried to buy an airline ticket and took almost 90 minutes to complete (the main problem came down to the credit information, street address and city – and, yes I completed the purchase in the name of research. You are welcome.).

This is the problem that SmartAction set out to improve.  An early startup, their software is quite impressive in the ability to understand Natural Language Processing (NLP) and guide an IVR script with it.  The recognition rates I noted were far above what I had seen before for natural language.  I was sold, there is something there.

Following the meeting they asked me if I could find research to show that Speech IVR and Artificial Intelligence was superior to standard IVR.  I thought it would be very interesting, agreed to do it, and they sponsored the research project.  I focused on the ability to retain a customer in an IVR session (as opposed to “escaping via zero” to an agent) since that is – to me – the measure of a good IVR solution.

I found some very interesting things in that research: did you know you could save almost 60% in time using Speech IVR with Artificial Intelligence? Did you know you can save almost 20% in costs? Did you know that 88% of people hate IVR (well, you probably knew that – but now you have the citation)?

Did you know you can read more about it in the White Paper I wrote (Our Success Rates Have Changed, Listen Carefully)?

If you don’t like reading, did you know you can sign up for the webinar on October 20th and you can hear me talk about it?

There is good information in there – download the paper, sign-up for the webinar… but whatever you do, don’t press (or say) zero.  No one is here to take your call right now.

2 Replies to “Speech IVR With Artificial Intelligence Is The Bees Knees”

  1. Hm, a meaningful conversation with a robot? The future is knocking at my door! That is amazing, IMHO the numbers in this post should be even more devastating. 100%, not 88% of people hate IVR.
    If there is greater evil than IVR, it should be ‘waiting on hold’. According to Newsweek “Of all the depressing statistics about a lifetime of consumer existence, this may be the most distressing: each of us is destined to spend roughly 1.2 years on hold.” There are several interesting solutions to the waiting ‘on-hold’ problem.


    1. Bob,

      I have not found that statistic, 100% of people hate IVR, anywhere – could you provide me a link? I did find plenty of evidence that when done properly, customers don’t mind IVR over waiting on-hold or not getting an answer. The fact that around 10% of IVR deployments are done well speaks to the other part of the 88% statistic.

      You can indeed have meaningful conversations with a robot — but the question is what do you call meaningful – right? right? Meaningful as in fulfilling at the humanity level, or meaningful as in fulfilled the need at the moment (answer, resolution, etc.).

      Thanks for the read and comment!


Comments are closed.

%d bloggers like this: