IVR Design – Best Practice

My nightmare

I am in a hospital lobby. It is dark. No-one else is around. I need to book an appointment but the lighting is terrible and the signs are full of words I don’t understand, like echocardiology and gastroenterology. I follow what I think is the correct arrow.

I get to the next signpost, but this is no better. I must have taken a wrong turn. I go back and try again.

And again.

Then I see the big red button on the wall marked ‘Press for help’. I suppress my own feelings of inadequacy for not understanding the signs and, relieved, I press it. And wait.

And wait…

You can see where I’m going here.

The aim of IVR is either

• to enable users to help themselves without agent involvement, or
• to route users to the right agent who can satisfy them in a single call

But many people’s experience of IVR systems has been rather like my (fictitious) nightmare – more Kafka than customer service.

This is primarily because of bad IVR design. There is no justifiable reason for it (although designing complex IVR logic is notoriously difficult without a visual design tool), and it has put many people off IVR for life.

IVR design

IVR systems often frustrate and annoy by not following these simple pointers to best practice.

1. Design should be customer-centric. IVR is about what the customer needs, not what you can offer.
2. Put the most popular choices early, e.g. ‘To pay a bill, press 1. For anything else, press 2.’ Analysis of customer call patterns will give pointers as to what is high on the list.
3. Pain #1 in using IVR is often ‘Can’t find the right option’. So…
   • use as few options as possible. Research suggests a maximum of 3. This should remove the need for the ‘To hear these options again…’ option.
   • be quick about it; the time to get from ‘Welcome’ to the end of the last main menu prompt should be less than 30 secs.
   • use the everyday language your callers use.
   • choose your categories carefully.
4. Use as few levels as possible. Research suggests a maximum of 3.
5. Offer every chance to ‘zero-out’ (i.e. ‘Press zero to speak to a representative.’) Most of the time, you don’t want users to do this, but a recent survey showed that one third fo users ‘always’ zero-out, and one third zero-out ‘most of the time’. Still more will get lost and/ or frustrated and need help. It’s better to have them speak to an agent, than to hang up and perhaps never call again. A well designed IVR system will keep abandons to a bare minimum. And by the way, under no circumstances should users have to request to speak to a representative twice.
6. Keep marketing out of IVR. Users have called you with a goal in mind. They will only be receptive to marketing once that goal has been satisfied.
7. Always offer a way back to the previous menu or to start again. Also, once in a hold queue, offer a way back into IVR if they don’t want to wait.
8. Test with real users before going live. Refine, and try again. And again.

Customer antipathy toward IVR may be increased by the fact that, whereas most service channels (e.g. Facebook, SMS, voice) are chosen by the customer, IVR is forced upon them whether they like it or not. If the customer has called to speak to a human, IVR is just an obstacle – unwanted and therefore annoying.

A note on Speech Recognition: recent studies show that 57% of users prefer to use a touch-tone keypad over SR to enter data, citing apprehension about SR problems as the main reason. (24% reported no preference, and only 19% preferred speech input.) The challenge of handling background noise and a variety of accents makes accuracy in speech recognition tough, but it is improving rapidly. Unfortunately, these improvements may take a while to filter through to the public consciousness. In the meantime, early negative experiences still dictate behaviour.

Bottom line: whatever your goals in utilising IVR – minimising costs, routing to the best agent, taking payments or simply providing information – make sure you are designing a dream, not writing a nightmare.

P.S. Here is a real-life example for your amusement:
from an Australian recorded information line, set up to answer questions about the new Goods and Services tax plan –

If you understand English, press 1.
If you do not understand English, press 2.