Chatbots seem to be all the rage these days in marketing. Some are good, some are, well… not so good.
They’re supposed to be a quick and easy tool to help increase visitor presence on your website.
They promise everything from increased time on page to simplified processes, eliminating humanity at the installation of a program and skyrocketing profits in the process.
And they’re becoming more ubiquitous by the moment. Google says there are ~33,000,000 website hits for “chatbots”. Now, those aren’t all bots themselves, but articles about them and self-promotions, too. But you know what I mean.
For your business, you can buy one, rent one, develop one, probably even lease one.
You’ve run into hundreds of them, perhaps thousands.
Yet…
Most of the time…
They’re totally worthless.
Well, maybe not totally worthless.
The companies who make and distribute them still get paid. So I guess they got that going for them.
But for the rest of us? Those who are thinking of using them in our business? Or who have to interact with them as an end-user? “The juice isn’t worth the squeeze,” as they say.
I’ve got two up on my computer right now. And they’re pretty bad examples of chatbotting. Let me explain, and hopefully help you to avoid these same user experience (UX) mistakes in the future.
CHATBOT ERROR #1 – TOO PUSHY
The first bad example is for a software company. I went to their website, and the thing popping up says:
“Hi 🖐️, can I set you up with an exclusive 14 day free trial?”
Basically asking if I want their product right away.
Before they know anything about me. Before they know anything about my needs, my budget, my clientele, anything!
I might be interested in a free trial. But I don’t know that when I first get to your website!
Offering that free trial is all well and good, in due time.
“In due time” means at the appropriate point in the sales sequence.
Maybe after you’ve already had some conversation with me, to investigate whether or not this software would actually be right for me.
You know, the old “Hi, how are you today? Do you come here often? Could I buy you a drink?” Not, Hey, I just met you, and this is crazy, but, wanna go out to my car and fool around?
Seriously, this is too much, too soon, and it’s going to turn people off.
You want to know what’s even worse?
This is a company I already do business with.
Yep, that’s right. I’m a customer.
They don’t really know me, do they? If they’re offering a trial of a service I already pay for, that tells me they don’t do a very good job of tracking who their customer are, now does it?
[Shaking my damn head over here.]Don’t do this. Don’t pretend you know what your audience wants (free trial) without asking (support for an existing customer, learn more about the product, find information about careers, etc.).
And don’t be so stand-offish that you don’t even track who your current customers are! And here I thought I was dealing with a software company. I guess that’s just a myth.
But enough about that. Let’s get to the second example of bad chatbotting.
CHATBOT ERROR #2 – INSUFFICIENT PROGRAMMING.
You know how it goes.
The chatbot, all dressed up in a fancy avatar with a pretend person’s face on it (because, let’s be honest, we all know at this point that there’s no real person on the other end, we could save all of us the trouble and just put up a picture of a motherboard rather than an actress), asks us what we’d like to accomplish. We type it in, sometimes the chatbot works, and sometimes it doesn’t.
In this case, it looks like this. Please note, this is a different company from the prior example. As it should be:
Hi! I’m Olivia, your virtual job assistant at *******! You can ask me anything about our business, culture, team and more.
Not so bad. I’m confident it’s a bot, but I expect that my question will get redirected to the appropriate party once I ask. So, I reply as such:
Hi, consultants generally have a reputation for a lot of travel. Would this be true for a “******** Consultant” role as well?
See, I’m using real-people speak and a question that I truly want answered. What’s the response? Well, instead of recognizing that this question is probably out of its standard programming, and telling me she’ll direct me to the appropriate party immediately, the chatbot honed in on a keyword (“consultant”) and provided a stock answer.
On our career site, you can search jobs by keyword or location. You may also enter a requisition number as a keyword. If a position has been closed or placed on hold, it may not appear in the search results. We would also highly recommend you join our talent community to stay informed about news, events and opportunities at ******** by clicking here: *******
Hm.
Did that answer my question?
Nope.
Did it recognize that it wasn’t answering my question, and direct me to a better way to get what I wanted?
Nope again.
Did it at least consider that it might not have answered my question, and confirm whether or not it did?
All together now: No way, Jose!!!
How do I know such truths about the inner workings of a chatbot? Because I tried to continue the conversation. Here’s my reply:
thanks. what’s five minus eighteen?
Pretty simple question, right? If you’ve got a real person. Or if you’ve got a well-programmed chatbot. In this case, you’ve got neither. You’ve got this:
My pleasure! 🙂
Multiple problems here.
First, the chatbot is masquerading as a real person, which is deceptive. We as consumers tire of this charade quickly. If you had a real person in there, she (“Olivia”), would see the question, answer “negative thirteen”, and then ask whether or not she’s been able to help you. Since she didn’t, I immediately conclude that it’s a bot, that I’m not worthy of a real person to this company, they don’t care about my problems or my questions, and I can write them off as unhelpful.
And secondly, if you had a well-programmed bot, you would be able to recognize when your standard script is insufficient, and you’d be able to code up a quick redirect to get them on the path. You’d keep the visitor happy, you’d actually solve their problem, and you’d be more likely to get a future client, applicant, or business partner.
All because you actually thought through an appropriate UX design.
Pretty simple functionality that’s been lost amid the desire for automation. And that’s a shame.
I mean, I’m not a UX designer, or a broadcast marketer, but heck, even I can see this isn’t working as it’s intended.
Sure, you cut out some people (and their requisite salaries and benefits), but at what cost? Reputation, ease of use, applicability, and humanity, for starters.
CONCLUSION: DON’T DO THIS!
Don’t set up your chatbots like this.
Just…
Don’t.
It’s like, “Do you even know me, bro?” No, no you don’t. [Even more shaking of my damn head. I’m getting dizzy.]
It would have been better with absolutely nothing, rather than creating either one of these negative value experiences.
So in the future, don’t expect your chatbot to be the savior of your business.
Make it better. Actually put some thought into it. Don’t just expect to get the make-out session without the get-to-know-you session first.
Think through how people use it. And, when they do, review what they did, and whether or not they got what they needed.
And when you do, you’ll get better results.
I guarantee it.*
*better better looks, unfortunately, not guaranteed.
***
P.S. If you’d like help creating automated messages for your chatbot, why not talk to a communications expert? A copywriter, perhaps? Send me an e-mail to stephan@sjmcopywriting.com and let’s figure out how to make your chatbot, your website, your book, or your autoresponders sound human again.