JournalArrow left Journal
Technology

Chatbots must die

by Rob Hall
19th December 2017

Everyone seems to be working on a chatbot right now but very few have created mass scale examples which have reached the market and achieved commercial success.

Most chatbots on the market currently fail to solve big problems for real people. To consolidate my thinking on this topic I spoke to Dean Withey, the ‘chatbot’ pioneer behind ubisend. They are one of a handful of conversational messaging companies leading the industry by creating conversational user interfaces fast and effectively.

One driver behind ubisend’s success is their own conversational messaging platform which underpins live chat agents across multiple industries and applications.

Peak Chatbot

The first thing that struck me when talking to Dean was how weary he was of hearing the word ‘Chatbot’. Mass scale applications remain rare and using chatbots for frivolous marketing campaigns add to negative associations around the term. Most people’s first encounter with chatbots is in a meaningless and sometimes frustrating context.

As it stands, they are for show: gimmicks to support digital marketing campaigns, or experiments by big IT conglomerates that don’t work (cast your mind back to Microsoft’s Tay faux pas in 2016).

Alternatively, consider two ‘award-winning’ and yet whimsical marketing chatbots such as ‘Hendrick’s Cucumber Adoption Service’ and BFF Trump. Although more serious bots have been built for e-commerce such as chatShopper or H&M’s Kik based bot (which o ers out t advice) their eld of application remains limited.

These interfaces, their application, features and design remain a novelty. Although, businesses like ubisend are testament to the fact that the landscape is maturing. My instinct is that we’re going to see a rapid development of conversational UIs’ contexts of use.

However, in Dean’s view, we won’t be making progress until we drop the term Chatbot and focus on the wider idea of ‘conversational computing’. This is because what’s going on with conversational UI is far more interesting in terms of application and design than the term chatbot suggests.

In sum, our collective understanding of conversational UIs’ capabilities and the best practice to adopt when designing them is limited. In reality, the future of chatbots lies elsewhere.

The future

As is often the case, some of the biggest early wins for conversational computing will probably happen away from the frontline of customer experience.

Imagine a conversational UI which stores and instantaneously relays hundreds of pages of HR documents in one place. And delivers this information via a well-designed interface which falls back to a human follow up after a series of failed conversational responses. Not only would this save countless hours of company time, it would also streamline your employees’ interactions with core business functions whilst saving you money.

Chatbots will come to serve as workhorses which alleviate pointless and repetitive information exchange tasks: how to locate parts, how to complete your expenses claim, the right dosage of a medicine. Long term, this will benefit businesses and individuals alike meaning they can do less, to do more.

These examples illustrate some of the many potential use cases of conversational UI in the future. The real application will come in any context where there is currently a drain on human resources when attempting to connect people with mundane and relatively easy to find information. This fits with the prediction that by 2020 80% of businesses will have a chatbot that tackles practical issues for customers and employees alike.

Conversational computing

ubisend do conversational computing through the creation of platforms which facilitate the application of bespoke conversational user interfaces. This is a much more interesting way of describing the emerging process behind the chatbot phenomenon.

Conversational human-computer interaction is an exciting emerging field, but the question remains; how do you design the right kind of conversational UI?

The role of the designer in the creation of conversational UI

The key to success with most new technologies is to draw on the well-established principles of user experience design. But, there are some specific additional ideas I got from talking to ubisend. 

If you are looking to excel in the field of conversational UI you will need to adopt all the good practices from the field of user experience design. Having reflected on our meeting I’ve distilled my thoughts on this topic below.


1. Don’t do it yourself

Too many chatbot worthy projects start with building the conversational platform from scratch. There’s no need for this, you may be better off recruiting someone with a pre-existing platform to help you. ubisend have built a ‘small talk’ module which is adaptable to any context in any industry. This allows you to train the bot quickly in the narrow field that you want it to converse in.

2. Make it clear you are talking to a bot

If you want to create a successful bot make sure you notify the user that they’re dealing with a bot. This will help people know what to expect, as well as safeguarding against any disappointment which may manifest itself with exasperated customers leaving your sales journey. 

3. Make it clear what the bot is for… and what it can’t do

Make sure it clearly defines the field that it operates in early on. If it’s a bot dedicated to delivering in depth weather reports about Norwich Cathedral then state this. This clearly defines the user’s field of expectation for the bot, reduces the amount of questions asked to the bot which lies beyond its parameters and further informs user expectations. The narrower the better - people will understand.

4. Keep on topic

Conversational aligners are a language tool which help bring the user back to focus on the narrow field of knowledge that the bot has. Asking questions like, ‘Do you have a question about our range of cars?’ helps to steer the conversation and provide better outcomes, with greater expediency for the user. 

5. Let the user hit ‘esc’

Adding a ‘fall back to human’ feature to the conversational UI after a series of failed attempts provides a safety net for your customers. It also allows customer service teams to swiftly follow up with the individual user by taking advantage of the locally stored chat log. This adds a human element to the process whilst also providing a failsafe uncommon to more traditional chatbots.

6. Intercept traditional forms of interaction

Don’t make customers come to you with their queries, be where they are when they’re browsing. This means that you’re failing to deliver a good user experience via your conversational UI if it’s buried within an app. Close the loop, place your bot where your users are – that’s on social or on your website. Conducting some user research around this will also help fine tune your placement and allow it to be better integrated within your overall product/service design. 

7. Build the bot into the experience

Consider the placement and presentation of your bot. How does this nest on, or next to, the digital experience the person is engaging with? How are you going to design the interface, iconography and visual elements of your conversational UI in a way that mirrors your brand identity and increases advocacy whilst driving conversion. 

8. Get someone experienced to write and edit your dialogue

In conversational UI the words become the design. Copy is crucial; users will be more likely to engage with an interface with personality. Some companies are making use of actors to help build a character, you can also carefully select the language to represent the subtle nuances of human interaction, balancing the needs of the business with that of the user.

In sum, the role of UX designers, researchers and UX writers is to build and enrich the personas of conversational UI. This includes tone of voice, language, structure, as well as taking in the design considerations mentioned above. 

The real problem of banal and lengthy tasks presents businesses with an opportunity to save time and costs. Let bots take the load and cut cost for you and save time for the customer. But, be careful about how it plays out in front of customers. The days of chatbots winning awards are probably already behind us, so focus on creating mutual value for you and your user.

Related articles

View All
Is Facebook Messenger about to eat your app?
Interconnected graphical globes around a central square representing marketing
Brand & Marketing

Is Facebook Messenger about to eat your app?

By Tom Wood

Our recent ethnographic study of WeChat usage in China gives several pointers about how payments, customer experience, and mobile apps are set to evolve in Western economies.

What's happening with WhatsApp in India?
WhatsApp Indian payments interface on a mobile device
Brand & Marketing

What's happening with WhatsApp in India?

By Rob Hall

WhatsApp has launched in India and looks set to become its leading digital payments vehicle.

What WeChat’s domination in China tells us about mobile banking in the coming age of the super app
An illustrated graphic containing particles representing a computer system
Technology

What WeChat’s domination in China tells us about mobile banking in the coming age of the super app

By Tom Wood

In less than a decade, the speed, simplicity and mass adoption of mobile payments has turned China from a cash-only society into an increasingly cash-free society.