Honda's ASIMO, an example of a humanoid robot
Image via Wikipedia

There are puppy robots who sit on command, human shaped robots who can climb stairs, and even robots that paint new art based on random mathematics.  Despite all of the advancements in computer behavior programming, and the fact that my Sims 2 character can be frighteningly realistic at times, 2010 is not the year for automated web chat.

I have had numerous sales emails and even calls from nice people trying to sell me automated sales chat on my company’s website.  The idea is to connect with visitors that are leaving your site without purchasing.  You find out their friction or pain points, identify an ideal offer, and lubricate their way back into your sales funnel with a hefty discount.

One email I received today highlights a 10%-30% increase in sales- and all without requiring you to do, well, anything except pay for the service.

Before you go sign some agreement and insert that little Javascript code on your site, let’s talk about this new experience you’re creating for your customers.

Human Chating is Caring

When taking a few chats for my company’s live chat implementation, one of the more frequent first questions is: “Are you a robot?”  Not surprisingly, no one is ever disappointed to hear that I am, in fact, a real live humanoid.  Skeptical, I’ve had one chatter then ask me what “infinity times infinity” was- just to make sure I breathed oxygen instead of positively charged ions.  My answer was “Why it’s 4, of course.”  And then a successful human-human chat followed.  There is a BIG plus to the customer experience when they realize that you care enough to be available.

Let’s take it a step further, why do they ask chat representatives if they’re human?  The chat window has my name on it, and I say hello.  They ask because of these automated experiences they’ve had in the past- and frankly, because not too many websites put their money & time where their mouth’s are about caring what the customers are thinking.

Wool is for Sheep & Sweaters

This brings us to the first big fatal flaw of most automated chat solutions.  They like androids of old, are on a quest to be human and they are presented as if they were.

In eCommerce chat, the goal of the chat is answering the tipping point questions that may be preventing a customer from completing their purchase.  They can be a simple question, or if your site is lacking- say, a comparison chart, they can be trying to figure out which product is right for them.

For some reason, I get the feeling that automated chat solution companies think their goal is to make their system as human-like as possible.  Trying to fool them with chat sets up the question: “If they misrepresent their chat, what else are they saying that isn’t true?”

A great point was made by @Rich_Kolb on Twitter:

“Does it mean they don’t care? I’m not sure, but if they admitted they were fake I’d be ok with it”

Pretending to offer someone a chat with a real person only sets them up for disappointment when they realize that the cute face wearing a headset in the photo is really just a collection of impersonal sub-routines.

The Turing Test Dooms Automated Chat

The Turing Test was another topic Rich hit on in his replies to my original Tweet ranting about automated chat.  For those of you who haven’t heard of the Turing Test, it’s a staple of survey level Computer Science courses in school these days.  The basic idea is that it’s a test of a computer’s ability to “think.”  What’s the easiest way to see if someone can “think on their toes?”  Why to talk with them of course!

The Turing Test puts a human in front of a computer with a chat window.  He or she begins chatting with TWO people (often 1 at a time) via the chat window- like Aim or Skype.  One of those people on the other end is a computer and the test- or game really, is to fool the human into thinking he’s chatting with another human.  Ever since it was proposed in a 1950 paper by Alan Turing (yeah, I’d name it after myself too!) the programmers of the world have been trying to pass this test.

While some programs have come closer to this goal than others, the reality is that it’s an *extremely* difficult test.  Many a Phd thesis and late night theory have been spent trying to map out practical theories and implementations.  In the end, your automated chat has a very HIGH chance for failing at some point in the conversation.

Boldly Go, Where No Program Has Gone Before

Get yourself a live chat solution, I’m a big fan of Bold Chat. For under $30 / month, you can get 2 operator seats and start interfacing with your customers.  Think of it as manning the phone, except that people are typing instead of calling.  Depending on the size of your company (and how confusing your website is  or isn’t) you won’t necessarily get inundated with chats.  It’s very manageable and there’s more value to it than just answering questions.

The customer experience insight gained through live chat is significant.  You’ll find out what your website isn’t communicating properly, how you stack up with the competition, and even new product ideas.

It’s really not so much about the dollar cost as it is the time investment. Sound familiar?  Social Media, buzzword for our decade, is also free.  Your cost is the time spent doing it.  While Social Media ROI tracking is still teething, live chat conversion results are build into most systems.  Additionally Chat can be turned on and off during the day based on availability of an operator.  It’s often a great idea to have the Chat button disappear when no one is available.

Allocate the time and get in front of your customers. You’ll be amazed by what you learn!


  • Pingback: Tweets that mention Ensure a Human Pulse in Your Web Chat | Casey Cheshire --

  • Shea

    Awesome post Casey!! I can't even tell you how many times I've answered a chat and received the same series of “tests” to prove I'm really a human. The point is that live chat should be like having a phone call. Only you're typing. And no one wants to go through an automated phone call, so why would they like an automated chat? Is that good customer service? Will you actually LEARN anything about your products, business, website, etc. by automating chat? Nope.

    (Disclaimer: I work for Bold Software and that's how I met Casey, but I also consider him a friend thanks to social media.)