Reset password:

Stop Making Pointless Chat Response Systems

Written by on April 28, 2016

I want to make a point about all these new chat bots, because most brands and publishers are doing it completely wrong. And the problem is that they are all doing it like this:

So... what's the problem here? Isn't this rather fancy and the future of chat AIs?

Nope. Not even close.

What you see above isn't the future of chat AIs. Instead, it's a simple voice response system like those we know from customer support hotlines.

You know how it is. You call a brand because you have a problem or because you are not entirely sure which product to choose. And, instead of getting a person, you reach this computer who tells you to "Press #1 for [...]. Press #2 for [...]. Press #4 for [...]."

You then grumble a bit, and press whatever number you need, to which it continues: "Please hold. You call is very important to us. Remember that you can use our automatic system for most questions, press #1 now to use that instead."

Right?

And people hate this. It's the most annoying form of customer interaction that you can have, done by brands who don't care about creating distinction, or having a passionate customer base.

Brands use these automated systems to disconnect themselves from their customers, hoping that their existing needs will keep motivating them to do all the hard work on their own. And sure, in terms of tech support, where people have a simple problem that doesn't require human intervention, this might actually work. But for everything else, it sucks.

So what does this have to do with Facebook Messenger and all these new chat bots? Well, everything! Because they are doing exactly the same thing.

Take the example above from the fashion shop Spring, and just look at the options. Now, as you read them, add "press #1" in front of each. Like this:

  • Hey Thomas! What are you looking for today? Press #1 if you are looking for women's items. Press #2 if you are looking for men's items.

So, you press 2, and it continues:

  • Super, what type are you looking for? Press #1 if you are looking for clothing. Press #2 if you are looking for shoes. Press #3 if you are looking for accessories.

And, again, you press 2.

  • Great. What kind of shoes? Press #1 for sneakers. Press #2 for loafers & slip-ons. Press #3 for boots.

You press 1, and it says.

  • Got it. What's your price range for men's sneakers? Press #1 for under $75. Press #2 for $75 to $250. Press #3 for over $250.

And you go on like this, narrowing down your selection until you finally get a standard response back.

Note: Screenshot from Fashionista.

And now they ask you to:

  • Press #1 to buy this item. Press #2 to see more like this.

These chat bots might look new because they are on Facebook Messenger, but they are not. They are exactly as idiotic as the very, very old voice response systems of the past.

And anyone who has ever worked in sales knows that this is a terrible way to get you to buy something. Even more so, it's a terrible way to motivate and connect with people. It's the random supermarket approach to selling.

More to the point, it's linear, incredibly slow, and exceptionally limiting the amount of products people are exposed to. And if you ever compared ecommerce sales on websites with mobile, you will know how dramatically negative an impact that has on sales.

People buy more if they can see more, but these chat interfaces limit that dramatically.

But there is an even bigger point here. This is happening on our phones. That is, our big 5.5" phones with AMOLED screens, fast mobile internet connections and touch interfaces.

Why would people navigate a slow, linear, and inflexible text chat interface to find a product, when they could just... you know... follow a link and do it on the web?

We have our screens right in front of us. Why would people do with text that they could do much better with a rich interface? Why would they not just do this?

But wait-a-minute, I hear some say. Isn't chat what the young people want? Isn't that how they are interacting? Look at how they use Snapchat, WhatsApp, and even Facebook Messenger at times. Surely, these chat UIs are the future. Right?

Well, yes... and no.

You see, what young people are actually doing is that they are using Snapchat and WhatsApp to socialize and chat with people around them. They are doing it because of the more intimate two-way interaction that it provides, not because of the UI.

In other words, young people use chat apps to chat.

This is not what brands or publishers are doing with these pointless chat bots. What they are doing instead is to use chat apps to navigate.

There is huge difference between the two. One connects and socializes. The other is similar to the annoying voice-response systems in call systems. These brands are not actually chatting with their customers. They are not doing any of the things younger people do. They are just focusing on chat as a format rather than chat as a behavior.

That's a massive difference.

The other thing we do with chat AI today is to use it as a command line for our voices. Think Apple Siri, Google Voice, Amazon Alexa, and Microsoft Cortana.

These are all part of another important trend. But you would never design them the way these chat apps work. You would never force people through a lengthy and cumbersome step-by-step navigation.

If Siri worked like the Spring chatbot above, it would fail.

What they do instead is that they use artificial intelligence, natural language pattern matching, and extensive algorithms to give people the answer they need from the first point of contact.

You would never use Siri to navigate through a linear set of options; you instead ask it a question. For instance, you would say:

Hey Siri. I need a pair of shoes that has the comfort of sneakers, but with a look that matches a business setting. Any suggestions?

And Siri would come back, after having analyzed this question, and say:

Sure Thomas. What about these classic low cut sneakers from Rivington?

This is what the future of chat AIs is really about. It's not to design bots that are basically just a website navigation in a chat UI. That's completely pointless. It's instead to focus on the two-way interaction between a real person and a real AI (or as close to it as we can come).

It means that your bot is actually chatting with people in the way you chat with anyone else. It's not about the UI or the format. It's about the behavior.

The publishers

It's the same with publishers. Here is a screenshot of CNN's chat bot.

It's the same problem. CNN is not actually chatting with its audience. It's just sending out a number of predefined options that people can then choose from. It's the old voice-response system all over again. Press #1 to read the story. Press #2 to....

But the worst part of this is that it's random.

Why would people use Facebook Messenger to get random news? This makes no sense. People already have a ton of random content inside the Chat Apps themselves, or via Facebook's News Feed. Why would they then go into Facebook Messenger and type "latest news"?

This is not a need that people have.

But more to the point, the CNN chat bot doesn't know anything. It's just navigating predefined phrases, just like you would use a website. Let me give you a few examples:

I asked it, "What the latest on Apple's problems in China?" And the answer was that it had no idea what I was talking about, but here is a link to some random articles.

And it's not like CNN didn't cover this story. They did.

So, I asked. "How are things going with BP?"... and, again, it had no idea what I was talking about (and yes, this story was also covered by CNN).

Okay, fine. So I eased up on it a bit, and asked it: "Okay... so tell me about electric cars" (which was a random topic not linked to any specific story). And here it did kind of work. It found three articles. Two about Formula E and one about Tesla. Sure, that's about electric cars, but has no targeting or intent.

Seeing that one of these stories was about Tesla's investors, I asked it "What did investors tell Tesla?" which again it had no idea about.

The CNN chat bot doesn't know how to answer questions either. For instance, I asked it, "How many primaries are there left before the convention?", which it didn't know anything about.

So, I asked, "When is the next Republican debate?" and here it finally matched a word that it had in its database, and gave me a very old article about a mishap when the candidates were introduced at a much older debate.

I'm sorry, but this is a pointless distraction. This is not the future of chat AIs. It's not even chat, nor is it artificially intelligent.

And there is a very simple reason why it's this crappy. The CNN chat bot doesn't do any natural language processing, and as such, it has no idea what you are asking it. It's trying to match specific words, so if it finds 'Republican debate', it will give you random stories about that. But it doesn't understand the context, nor does it comprehend what has happened before.

It's just a very simplistic search engine and web navigation designed with a chat-like UI.

Why tech companies are winning

The problem with all of this (both for brands and for publishers) is that their chat bots are just a gimmick. They have no real world use. Right now, they get a little bit of traffic from people trying them out for the first time, but, because they are generalized and focused on navigation, people will quickly get bored and move on.

The problem is also that this is another example of how the tech companies are winning. We constantly hear about this in the publishing world. Publishers are asking, why does Google keep winning?

The answer is obvious. While the brands and the publishers approach new tech innovation as a gimmick, one that they can just quickly implement using some standard low-end bot service requiring little or no editorial change, the tech companies approach it as a future strategy.

In the latest Google/Alphabet earnings call, Google's CEO Sundar Pichai said this:

We've been investing in machine learning and AI for years, but I think we're at an exceptionally interesting tipping point where these technologies are really taking off. That is very, very applicable to businesses as well. So thoughtfully doing that externally we view as a big differentiator we have over others.

"In the long run, I think we will evolve in computing from a mobile-first to an AI-first world. And I do think we're at the forefront of developments.

In other words. Google considers this the future of interactions, and they are going all in to make it happen.

We hear the same from Microsoft and from Amazon.

There is no gimmick here. This is a serious high-end strategy, that will redefine what these tech companies will be in the future.

Compare this to CNN's bot, and you see nothing like this. You instead find a publisher who is approaching technology by only focusing on the lowest hanging fruits, creating something barely good enough to even be considered a bot.

CNN's chat bot doesn't change who CNN is. It doesn't impact their newsroom, their editors, or their journalism. It doesn't even alter the way people consume the news, nor what defines news in the first place.

And this is why the tech companies are winning. They look at new technology, new behaviors, and new consumption patterns as a potential strategy changing event. And instead of just doing the least amount possible about it, they set out to 'own it'. They want to own this change, not just use an already out-of-date bot system that has no intelligence at all.

You see the difference?

Mind you, I don't expect publishers to be the leading entities of AI and chat research. That's not your market. Your market is to be really good at communicating.

But what I am asking of you is to approach these things as more than just a gimmick that doesn't have any impact internally. You need to approach this future as something that will potentially change the way we consume news, exactly the same way as digital and mobile changed how people behaved.

And you can't just think of it as a new format. That's what publishers did when they saw digital and approached it by thinking of it as a simple marketing channel on a website. Look at where that put them today. Every single newspaper is now struggling to survive, because they never approached digital as an event that would change the way we define newspapers.

So, stop thinking of chat AIs as a format that you design something to look like. Just because you can make something look like a chat bot doesn't mean it's actually a chat bot.

Instead, think about this future of chat AI from a behavioral perspective. What type of news do people need, how would they need it, and when would they need it? Imagine a world where the first point of contact is people saying:

Hey Siri/Cortana/Alexa/Google, ...

What would this mean for your editorial and journalistic focus?

Does it mean this?

Or does it mean this?

Hey Thomas, the latest news is that BP had a terrible year. It had an annual loss of $5.2 billion, compared with a profit of $8.1 billion in 2014, and a 50% drop in profits (excluding one-off charges). And it now plans expect to cut 7,000 jobs by 2017 (3,000 more than expected).

"The reasons are many. There was the continuing fallout from the 2010 Gulf of Mexico disaster, the steep fall in oil and gas prices, and writedowns in its oil and gas exploration and production business.

"We had our business analyst look over the numbers if you want to know more, and we have a scheduled interview with the CEO coming up next week.

It's the same with brands. The chatbot from Spring isn't intelligent either, nor does it actually help people. It's just their standard linear web navigation designed in a chat interface.

Again, they are approaching chat AIs as a format rather than a change in behavior. So, ask yourself, what would people actually need in this type of situation? One hint is that, when people engage in chat, they do so because they have an intent. So what is that intent? How would they communicate that intent if they were chatting with a real person? And how can you then make your chat AI bot do that instead?

As I mentioned before, how would you do this?

Hey Siri. I need a pair of shoes that has comfort of sneakers, but with a look that matches a business setting. Any suggestions?

Also, think about what type of moments this applies to. One of the key things to understand about digital natives is that they are living in a multi-moment type of reality, and designing for the right one is critical. So which moment is right for chat AIs? Or will chat AI define an entirely new type of moment?

This is the key to winning in the future. It's not anything that we see from the simplistic chat bots today.

If this was interesting to you, I wrote a much longer Plus article (39 pages) about this in "Future of AI/Chat: It's Time To Step Up The Game". This is part of a three-part series about all these new forms of media, the other two being "The Future of VR and 360 Video" (44 pages) and "Creating the Future of LIVE Video" (32 pages).

Share on

Thomas Baekdal

Thomas Baekdal

Founder of Baekdal, author, writer, strategic consultant, and new media advocate.

Follow    

Baekdal PLUS: Premium content that helps you make the right decisions, take the right actions, and focus on what really matters.

There is always more...