“Yes, I am human:” learnings from fast customer service via live chat | Hacker News Reader

“Yes, I am human:” learnings from fast customer service via live chat

blog.indextank.com

67 pointsrolando33 comments

I used to work for a competitor of Olark's and I can confirm this happens all the time. What's funny is that there are two types of companies:

1) Trusts their well-paid chat operators and allows them to type freely.

2) Force-feeds their operators 'canned responses' (as are used in email support) and doesn't encourage them to move off script.

#2 is always what marketing departments want, but it was always worse for the customer. The operator's response was too quick (a paragraph of text hastily selected from a recommended response), there were no misspellings, and they used PR speak that normal humans don't use. Often people would respond, "are you a person?" and some small group would ask questions specifically designed to trip up a bot.

My advice is to allow the people talking to your customer to use a modest amount of slang (a smiley or an occasional lol is fine if the customer is in a happy mood) - let them make a few misspellings - and encourage them to really read the response and even empathize with a customer. Marketing departments hate this idea, but in the end the feeling of a connection to another human is worth the risk of offending a grammar nazi.

Yep, "legacy" PR based on controlling the message is just fundamentally flawed. It is time for candor and honesty.

If you're talking to a person who can't deviate from the script, you're not really talking to a person -- you're talking to the script. It's like an AI, but clunkier.

Right, basically the rep is stuck in the Chinese Room.

Our development team at IndexTank handles our customer questions. The training they've received comes from working alongside client development teams for years (our team has been together for 4+ years handling search problems for many companies). So we trust each other to type freely.

The misspellings is an interesting suggestion. Since our team originates from Argentina, English is our second language, and we do make mistakes. But we try not to. But now maybe we should ! Thanks for your advice.

Most people will take thoughtful typing with mis-spellings over a 'canned response' any day - especially younger users.

A bit more advice since you're abroad. If people ask, I wouldn't lie about your location. We had some customers who would ask Indian reps to pick an American name (lots of Bobs) - but sometimes people make chit-chat about weather and the like while waiting - and getting caught in a lie can destroy the company's reputation with the user.

A great point. We have come across some situations where we are asked where the team lives, and we tell it like it is. Developers are pretty positive about us being both in Buenos Aires and in SF. I think the global community is a nice plus.

It's usually pretty obvious when you're talking to a "real" Bob or a "fake" Bob. In my experience, the offshored agents are almost incapable of communicating successfully.

I was the Mechanical Turk (not in the "amazon" sense of the words) behind the Olark in that conversation. I was never instructed in what to say when we get a question from a customer. We only have to follow three simple laws, in a specific order: never injure a human being, obey orders by human beings and protect our own existence. That's all.

But, hey, saying "how can I help you?" sounded right to me, so I use it often. Since the blog post we're getting contacts through Olark testing our response times. We're still pretty fast, and we're still passing turing tests!

Those rules are surprisingly similar to Isaac Asimov's 3 laws of robotics.

I don't think that's a coincidence.

It might even be an instance of that rare animal called a "joke".

Tell me more about those rules are surprisingly similar to Isaac Asimov's 3 laws of robotics.

AdGrok's seen this a number of times:

-----------------

USA (Philadelphia,PA) #4859: wow

USA (Philadelphia,PA) #4859: so you are not a bot?

Matthew: Only on the dance floor, man.

-----------------

USA (New York,NY) #664: haha never chatted on olark, this is fun

antonio: yeah, it's quite the app

USA (New York,NY) #664: didn't think there would be ppl behind it

USA (New York,NY) #664: :)

antonio: there isn't, this is a robot

antonio: :p

USA (New York,NY) #664: hahahah

USA (New York,NY) #664: ok

USA (New York,NY) #664: am very happy if robots are this intelligent

antonio: well, intelligent-seeming...

I am thinking it would be nice to have a 'best of olark transcripts'collection. Surprising customers with free-flowing interactivity is what this is all about.

+1 for your 'only on the dance floor,man' response

I worked customer service for "a" major MMO for a while. People who got asked if they were robots were generally behaving like robots: using macros (pre-written responses) to respond to questions even when the answer didn't properly fit, sending responses (even proper responses) to questions faster than people generally type, or abusing the auto-name-fill-in shortcut to fill in the character name in every sentence.

In short, there's very likely to be an actual problem when your customers ask if you're a robot ... not a "cute" problem like you were too quick to answer their initial request.

Not exactly. As you see in the transcript on the blog, we responded with a simple "Hi, how can I help you?" and then we're asked if we were a robot. We don't use pre-written responses, nor do anything else like you suggest. We're just online, working away.

And you'd be much more likely to be seen as a person if you opened with, "Hi, this is Jorge. How can I help you?"

Jorge the robot? If someone is anticipating a bot, I don't think greeting them with a name is going to dissuade them from asking.

Interesting... we'll do some tests on alternate ways to introduce ourselves. Being able to test out different approaches real-time is another benefit from live chat. thanks!

Another possible response might be "Negative, I am a meat popsicle".

My first thought looking at the transcripts was that seeing "IndexTank" as the screen name answering the questions rather than a person's name would make me suspect "robot" at first too. Either way, I agree that introducing yourself to the user with your own name will do a lot to reduce the suspicion.

That's a great catch. We chose the name for the account when we started using Olark, and labeled it with the company name. We decided initially to have only three accounts and rotate them among our team, so all of the accounts were given IndexTank labels. As we are expanding our use of Olark, we will give each of our developers their own account, and they could put in their first name there.

We will test it out. thanks.

As a hobby I volunteer as "support" for an MMO, too. I go into an open chat room and answer people's questions. I'm very obviously a human - every response is different, the answers fit the questions (because I'm typing them!), I abbreviate people's names, I share anecdotes etc...

Still people not only ask if I am a bot constantly, sometimes they just decide that I am. (Of course, they still believe me if I start talking to them directly and "prove" it, but in those cases I don't like to ruin the joke if I don't have to!)

The only robotic thing I do is type about 90wpm - and read fast enough to keep track of a quickly scrolling chat room. It's not superhuman, but I am "above average" compared to the casual participants. I can understand people being surprised at first.

But making the leap into assuming a fast typist is a bot - despite all the evidence to the contrary - well, sometimes there is no actual problem. Sometimes people are just ridiculous.

How do you deal with the constant distraction to the dev team? Must be quite bad if you have a lot of customers and a small dev team.

Having random people pop up on your IM client seems like the opposite of an environment that encourages a state of "flow"...

Our main objective is really to get to know our customers. We've been working on the platform for a fair amount of time, and we need interactions to tell us how we can make it better.

Also, we do rotate accounts among the team which helps some.

Finally, we don't see them as 'random people' -- we see them as developers like us, using a service we built. WE LOVE THAT.

I added this to http://paggstack.com since the very beginning, and I swear, it's the most commonly asked question over any other. I always try to put some humour in my response, but damn! We need a solution ;)

I cant help but think of the fonejacker cinema prank and laugh. http://www.youtube.com/watch?v=WgOkJjjwmnk

It is like a reverse turing test.

That will work until an AI can pass the Turing Test.

.. there may be more important things to worry about when that happens.

Like apologizing to Kurzweil?

Skynet: hi I'm a self-conscious robot based on technology from the future, how can I help you?...

even Deckard thought he was human, and we don't know if he passed the Voight-Kampff machine test http://en.wikipedia.org/wiki/Themes_in_Blade_Runner#Deckard:...

Reply on news.ycombinator.com