news

Are you ready for chatbots?

Posted: 12 July 2017 | By Adrian Benić, Infobip

Chatbots have been the tech du jour over the past year. Fuelled by the growing usage of mobile messaging applications and advancements in AI, they look set to change the face of messaging forever. But bots aren’t actually a new technology.

Pioneers in the field have been experimenting with machine learning and language processing since the 1950s. So what’s changed and why have they risen to prominence now?

Well, chatbots are no longer the exclusive domain of computer scientists or tech giants with endless budget and developer talent. There are now a growing number of platforms, like Facebook Messenger and Viber, armed with tools that will allow much smaller businesses to jump on the bandwagon and create their own bots with impressive results to their bottom line.

Many brands including Domino’s Pizza, Burberry, Uber, Sephora and even the Guardian have launched chatbot services in the past year. But not all have been successful, and more often than not it’s the chatbots that go wrong which get the most attention. You only need to look at Tay, launched by Microsoft last year. It took less than 24 hours for Twitter and the Internet to train the bot to become a misogynistic racist.

Tay offers a reminder that we’re still figuring out AI and chatbot technology and there’s still work to do in relation to fine tuning best practices. Every company, from start-ups and smaller business to larger corporations, needs to understand the dos and don’ts of creating chatbots.

With that in mind, here are some guidelines on how to nail this interface:

Set goals and stick to them

Firstly, there is no one-size-fits-all approach. A chatbot and its features should be based on your company’s specific communication strategy or business needs. The questions you should ask yourself when building your bot are: What is the core purpose of my bot; is it a bot that handles routine banking enquiries, retail customer service, or food delivery? What features will your users therefore benefit from? And what features aren’t necessary?

However, setting the development criteria is only one piece of the puzzle. You also need to consider how it’ll be discovered. Most people recognise the immediate response benefits that a bot can offer, but they’ll still need guiding towards using it or discovering that it exists in the first place. Making your bot discoverable is, therefore, as important as ensuring people visit your mobile app or website – consider bots to be an extension of the overall experience you offer, rather than separate function.

Equally, you should be flexible to your users’ needs, but don’t overload a bot with every function possible as this will only complicate the process of building and maintaining. The more capabilities you add, the longer it will take to build and deploy and the larger the possibility of an error occurring. You should have a clear idea about the service you’re aiming to provide. The saying ‘don’t run before you can walk’ rings true here: nail the bot’s core function and then you can expand by collecting feedback and monitoring metrics.

Use clear and natural language

One of the most important aspects of chatbot user experience is the language a bot is programmed to use. As important as any business messaging, its dialogue should be welcoming and human-like but also clear so that users can understand and can complete their desired tasks.

Avoid overloading the user with too much information by keeping messaging below 100 characters and only making use of features that are appropriate to the bots function and potential conversations it has with users. For example, the functionalities of a pizza delivery bot will need to be different to that if a bot that books healthcare appointments, both in complexity and the available feature set. Finally, repeat inputs back to users by asking them to confirm to check their understanding and allow them to correct a misunderstanding.

Manage user expectations from the offset

Once your bot is up and running, make sure its purpose is clear to the user. It’s better to make this known from the introduction than to have a consumer, who is time and patience poor, becoming disengaged because their expectations don’t match your chatbot’s abilities. Let them know upfront the core function as well as limitations of your bot so they know what they can use it for during initial conversations. Keep offering them quick and easy reminders of this; including what tasks your bot supports and how they can interact with it.

You should also make users aware that they are talking to a bot. Don’t try and create the illusion that they’re talking to a human. They’ll work it out eventually and could potentially become distrusting of the service you’re providing. People are more forgiving and tolerant if you are honest with them in the first instance. Be sure to have a mechanism in place to pass the conversation over to a human representative should your bot experience problems – this is paramount considering that chatbots aren’t quite there yet just yet.

Insert humour (but not too much)

Chatbots and AI are certainly new and exciting to consumers. Take advantage of this novelty and instil some personality into your bot. Bots have the potential to represent your brand’s voice and the brand messaging you are trying to convey, and will therefore be as much a part of representing your products, services and brand values as a human representative. This means your chatbot needs to speak and type like a real person.

A chatbot that meets robot stereotypes and interacts with users like a machine will disengage quickly, meaning users may not return to your services. Throwing in a few cheesy or playful jokes, perhaps even an emoji or two, will add to the user experience but be careful these don’t become an annoyance and deter from the bot’s actual function. Users will appreciate a chatbot that can help them in some way, rather than funny one.

Offer lots of hints and help to aid users on what to do next

The pace of modern life has made all of us rather impatient. Nine times out of 10 users will give up and leave a conversation if the next step or a solution is not presented to them almost immediately. Guide the user through the conversation so they are able to easily get the information they need, for example a recommendation or an answer to a question, without wondering what to say or do next. Creating seamless interactions makes for happier and more engaged users who will return to your chatbot service in the future.

Estimates predict that chatbots will be commonplace within five years from now, with Gartner forecasting that intelligent automation will manage 85% of business’ customer relations by 2020. Messaging bots are a big deal, and experimenting now with what works for both your business and your customers will put you in good stead for when chatbots become a staple part of any brands’ communication strategies.

Once your bot is ready for deployment, you’ll need to connect it to your desired channels. Partnering with an established messaging specialist with the APIs that can provide simple connections to Facebook Messenger, Viber, and a variety of other chat applications, can ease this process and help you beat competition in the long run.

Also, make sure you get the experience and functionality required to drive the shift from transactional to conversational chat bots, funnelling users towards a better experience and also helping to bring consistency across all channels you are using in your communications and workflows.