The Customer Service Survey
VocaLabs' weblog providing news and commentary on the challenges of providing good customer service.
Don't Make Me Think
Monday - November 21, 2005 12:54 PM in
I recently read an outstanding little book on web site design called Don't Make Me Think, by Steve Krug. There are many things about this book which make it among the best I've seen on the topic: it is short enough to read in a couple hours, it includes cute little illustrations, and it is chock-full of practical advice for people stuck in the real world. In fact, I've been inspired to redesign parts of the VocaLabs web site based on what I learned (look for a new homepage soon).
But the best thing about Krug's marvelous little opus is that you don't even have to buy it, because he distilled the most important point of the book into the four words of the title: Don't Make Me Think. A web site should not make people think about how to use it. It should present obvious choices at every step, and it's OK if people sometimes choose the wrong thing and have to go back, because that's how most people use web sites anyway.
That's a lesson which can just as well be applied to IVR and Speech Recognition systems. Automated customer service--of any kind--should not make the caller think. Every choice should be obvious, and it should be easy to go back if you choose wrong.
For example, take this initial prompt (please!), which I've heard when calling the 800 numbers of more than one big retailer:
"If you are a BigBox customer, press One. If you are a BigBox.com customer, press Two."
But what if I'm both? Or neither? What's the difference between BigBox and BigBox.com anyway? Aren't they the same company? And what happens if I choose wrong? (Answer: because the calls are being routed to two completely different contact centers, if I choose wrong I have to hang up and call back.)
This particular form of corporate weirdness happens because BigBox tries very hard to create a single brand for its retail and online stores, but in fact they are two completely separate entities. Target.com, for example, is actually operated by Amazon.com. But you can hardly be surprised when the customer is confused, then frustrated when she chooses the wrong option.
Then there's this doozy, from a design which (thankfully!) never saw the light of day. It was an account information system for life insurance policies, and at one point it asked the following:
"If you want to know your withdrawal value, say 'withdrawal value.' If you want to know your surrender value, say 'surrender value.'"
WTF? I spent nearly five years working in finance and I'm a Chartered Financial Analyst, and even I don't know the difference between the "withdrawal value" and "surrender value." How can you expect your average customer to have the slightest clue?
A better prompt would be something like "If you want to know how much you can borrow against your policy, say 'Borrow.' If you want to know what your policy is worth if you cash it in, say 'Cash it in.'"
But even better would be to eliminate the prompt entirely: "You can borrow up to XXXX dollars against your policy. If you were to cash in your policy, you would receive XXXX dollars." Just providing both pieces of information takes less time than asking the caller which piece she wants.
Then there's this greeting, which is becoming more popular:
"Hi, I'm Jill, BigBox's automated helper. I can help you check your order status, place a new order, or track a shipment. So, let's get started. What can I help you with?"
Call me a curmudgeon, but what's wrong with "Thank you for calling BigBox. Would you like to check your order status, place a new order, or track a shipment?" I understand that the people who designed "Jill" are proud of their work (and often rightly so: some of these systems perform very well on our service quality benchmarks). But all the automated perkiness doesn't fool anyone and takes twice as long.
Worse, it relegates the most important piece of information in that recording (the name of the company) to second place behind a completely irrelevant piece of information (the name of the fictional automated assistant). The very first thing any customer service system (automated or human) should do is confirm that the customer did, in fact, call the right number.
Of course we all have our pet peeves. But if you follow Krug's advice and don't make the caller think, you're already well ahead of the pack.
Posted by Peter Leppik
That's a lesson which can just as well be applied to IVR and Speech Recognition systems. Automated customer service--of any kind--should not make the caller think. Every choice should be obvious, and it should be easy to go back if you choose wrong.
For example, take this initial prompt (please!), which I've heard when calling the 800 numbers of more than one big retailer:
"If you are a BigBox customer, press One. If you are a BigBox.com customer, press Two."
But what if I'm both? Or neither? What's the difference between BigBox and BigBox.com anyway? Aren't they the same company? And what happens if I choose wrong? (Answer: because the calls are being routed to two completely different contact centers, if I choose wrong I have to hang up and call back.)
This particular form of corporate weirdness happens because BigBox tries very hard to create a single brand for its retail and online stores, but in fact they are two completely separate entities. Target.com, for example, is actually operated by Amazon.com. But you can hardly be surprised when the customer is confused, then frustrated when she chooses the wrong option.
Then there's this doozy, from a design which (thankfully!) never saw the light of day. It was an account information system for life insurance policies, and at one point it asked the following:
"If you want to know your withdrawal value, say 'withdrawal value.' If you want to know your surrender value, say 'surrender value.'"
WTF? I spent nearly five years working in finance and I'm a Chartered Financial Analyst, and even I don't know the difference between the "withdrawal value" and "surrender value." How can you expect your average customer to have the slightest clue?
A better prompt would be something like "If you want to know how much you can borrow against your policy, say 'Borrow.' If you want to know what your policy is worth if you cash it in, say 'Cash it in.'"
But even better would be to eliminate the prompt entirely: "You can borrow up to XXXX dollars against your policy. If you were to cash in your policy, you would receive XXXX dollars." Just providing both pieces of information takes less time than asking the caller which piece she wants.
Then there's this greeting, which is becoming more popular:
"Hi, I'm Jill, BigBox's automated helper. I can help you check your order status, place a new order, or track a shipment. So, let's get started. What can I help you with?"
Call me a curmudgeon, but what's wrong with "Thank you for calling BigBox. Would you like to check your order status, place a new order, or track a shipment?" I understand that the people who designed "Jill" are proud of their work (and often rightly so: some of these systems perform very well on our service quality benchmarks). But all the automated perkiness doesn't fool anyone and takes twice as long.
Worse, it relegates the most important piece of information in that recording (the name of the company) to second place behind a completely irrelevant piece of information (the name of the fictional automated assistant). The very first thing any customer service system (automated or human) should do is confirm that the customer did, in fact, call the right number.
Of course we all have our pet peeves. But if you follow Krug's advice and don't make the caller think, you're already well ahead of the pack.
Posted by Peter Leppik
Posted at 12:54 PM by | | | |