Hello or Hell-no? — Why Everything You Know About Chatbot Greetings is a Lie
That “Your chatbot’s greeting is its most important response” thing is a myth, and I can prove it.
If you speak to chatbots on a regular basis, chances are you’ve been (unwillingly) welcomed by many of these:
The crucial starting point to any chatbot designer’s project, the illustrious Conversation Starter (or greeting message) is an essential way to open the conversation, guide the user through the topics and teach them how to use the bot. Except, well really, is it?
If you’ve ever answered your door to a heckler, you know how unwelcome a Hello Invasion can be. And an over-talkative, unwanted address complete with a list of PLEASE PUSH ME buttons screaming at my door uninvited?! That sounds even worse.
So, why are we heckling our poor users? Now, I haven’t forgotten my manners and am not suggesting that we do away with social convention entirely. But, when it comes to saying hi, I’ve been thinking — shouldn’t chatbot greetings have the same etiquette as a human hello?!
After all, I don’t walk into a store and expect to be bombarded with a bunch of different ways their staff can help me:
“Hi my name is Heather! And I can show you where the jeans are, or the tops, or the dresses (take your pick from the above options!) and then I’ll meet you at the cash register where I’m physically capable of processing your payment with your choice of credit card, cash, coin,…”
…you get the point. So why are we forcing actions on our users before they’ve had the chance to say ‘How are you?’, let alone tell us what they’re looking for?
Users should do the talking
As I’ve said in my other articles, when it comes to customer service (or any user interaction for that matter), a chatbot that’s too helpful is just as annoying as one that’s not helpful at all. Especially in an era where (dare I say it) we are constantly invaded by the over-eager, well-intended popup reminders from apps we install, or the I-Swear-I’ll-Throw-My-Laptop-At-The-Wall-If-It-Happens-Again alerts from anti-virus software we don’t remember installing.
To put it plainly, the Siris and Silicon Valleys of this world have conditioned our exhausted users to resent all pop-ups and un-chosen content, despite (or perhaps as a direct result of) how cheerful it may be.
So when your lovingly crafted chatbot pops into their life, you better make damn sure it’s not as intrusively helpful as its over-talkative peers. Instead, I propose that we give our users a little credit — not to mention some cognitive breathing room — and allow them to lead the conversation. After all, they’re the only ones who know what they want.
How not to say hello to your users…
Let’s take a look at the Anatomy of a Terrible Greeting Message to identify the common culprits (I’m allowed to say that because I designed it):
Here, you can see our unfortunate user has gotten the works:
- A greeting so long it requires scrolling to read. Or worse, it’s not noticed at all. Your instructions are gone, and confusion ensues because your content is based on the assumption it’s been read.
- Making assumptions about questions that might be asked. (BONUS POINTS for asking questions you don’t allow users to answer. I wouldn’t do that in this article, would I? So glad you agree 😋). Taking words out of the user’s mouth can make them feel frustrated and ignored. It’s similiar to talking over someone during a spoken conversation and leads to disengagement.
- A bustling list of buttons which requires more reading and thinking than any old typed sentence would.
- Too many topic cards. The point of a bot is to find your answer faster than you would on an FAQ page. Scrolling through unnecessary menu options defeats this purpose.
But self-deprecating sass aside. You’re probably thinking:
What if my bot doesn’t know what the user is asking? Aren’t menus and buttons helpful then?
To that, I say — Tough terabytes, my friend. In the realm of chatbots, the only way to solve Conversation Design problems is to create a Conversation Design solution. If your fallbacks (or failed message data) show that the programmed keywords aren’t finding the right content, uhh, dude, fix those keywords. If your users ask two different questions in similiar ways, geez that’s tough. But with enough strategic thinking, you can find a copywriting solution to separate those queries (like a disambiguating question):
“But what if the content they need isn’t programmed into my bot?!” I hear the Conversation-Design-Anxiety scream within you. “Shouldn’t I lead the user in the right direction towards content that exists?!” *Cue heavy breathing*. Too bad. That’s the point of what we do. Until we’re at Skynet Singularity levels, our bots won’t answer for themselves, so we have to think for them. In my mind, the best results are gained from one failed message, from a single user which informs the next answer I‘ll add to the bot. Not a weird smoke and mirrors loop that pretends we understand them because we’re too afraid to admit that we don’t have a Content Crystal Ball, so we didn’t write that answer yet.
Buttons spelling out the functionality of your bot is the opposite of user-focused design and quite frankly, a big clickable bandaid for bad design. The better your bot’s architecture is, the less important your greeting message becomes. So, don’t worry about it so much and just focus on making your content the best it can be. 🤗
Pssst..people don’t use menus anyway!
Here’s a secret straight from our analytics:
People rarely click on greeting menus and buttons. They may look flashy, but our experience is that majority of users overlook greeting menus because they prefer to type their answer in the widget. And if that menu’s cut off by the widget? Well, engagement gets even worse. You can bet your sweet buttons that your users are far too busy to go scrolling in search of that content:
I’m talking 4% of users couldn’t give a flying flux capacitor about your delicately crafted menu options. Still, one shouldn’t ignore a menu’s merit as they neatly present pics in the bot and showcase products beautifully. All I’m saying is, it depends on the use case and whether the user has requested the content that they’re seeing. Ask yourself: Are you a business like a fashion store, where images are important to your users? Or can the message be conveyed to them with text alone? And then, does the user need to see these images before they get to say “I have a question.” Or could it wait until they ask “Show me 5 dresses that suit my style?”. The temptation even I fall into, is to PACK IN THE FUN in your bot content with a technicolour cornucopia of dopamine-inducing images, emojis and buttons at every step. But I’m sure this focus on visual stimuli tends to annoy rather than assist the user (unless the sole purpose of your bot is to entertain).
The right way to say ‘Hi’
In my humble opinion, minimalism is key when it comes to your Greeting Message. The safest option, in terms of appealing to the user and retaining their interest, is to let them do the talking. Everyone likes to be asked questions and to have their opinion heard. A simple, “Hi, how can I help you?” suffices for the majority of cases. Take it from the masters:
1. Google Assistant
Google, Apple and the like don’t slather their AI with buttons. In fact, it’s about as minimal as you can get.
This is the most human-like your greeting can be! Not to mention, the lack of reading and assumptions of a user’s needs means this message isn’t likely to irritate your users into disengaging.
However, If you must guide them through the content, do it in a way that is interesting to the user. Give instructions that feel as conversational as possible, ideally in the form of a question. Allow them to interject as many times as you can so it’s a truly interactive experience. And keep the introduction process short, so you can get to the REAL answers ASAP:
2. Mica the Hipster Cat
Mica has one function, so it makes sense that she tells you about it first.
She lets you do the talking and gets the convo started FAST. But doesn’t waste time with a boring list of buttons to scroll through.
There are options for different interaction methods to suit different user types and best of all, IT FITS IN THE WIDGET!
This conversational command is much more engaging than a lengthy ‘POPULAR HOTSPOTS’ menu.
3. DR.AI
Text-based options would be even better, but at least Dr. AI gets straight to the point.
The greeting doesn’t tell you what to do because the rest of the content speaks for itself.
There are buttons, but that’s ok. They’re minimal, and you don’t have to scroll sideways to see each one.
If you must explain your bot’s purpose, do it in one sentence like this! It’s never necessary to waffle in a widget.
So, now that you know what not to do, I pass the hello torch onto you. Let’s put our waving hands back in our pockets (unless it’s to give me claps 😉) and get ready for the revolution. Now is the time for chatbots to triumphantly stand up* and say — “Hi, how can I help you?” rather than, “Hi, this is how I’m helping you whether you like it or not.”
*metaphorically speaking.
👏 I promise I won’t say hello if you give me some claps!