When Facebook launched its Messenger API for chatbots in 2016, established companies and aspiring entrepreneurs could hardly contain their excitement. The app was in the pocket of 900 million individuals across the globe, and suddenly offered them the unprecedented opportunity to communicate with their customers in a personalized and automated fashion.
Despite the early hype around the technology, though, chatbots have yet to prove their worth for many businesses. In a recent survey, over half of chatbot users found the technology to be “not effective” or only “somewhat effective.” Not to mention a similar proportion of chatbot owners themselves were found to be unsatisfied with the technology’s performance as well.
But it would be naive to assume that these findings point to a grim future for the technology. High-profile chatbots for companies like American Express and Capital One have seen huge success thanks to their slick designs and ease of use, demonstrating that the technology’s slow start has really been the fault of uninformed creators – not uninterested consumers.
The truth is that chatbot usage is on the rise, and that 80 percent of businesses expect to leverage this type of AI technology by 2020. With that in mind, companies looking to adopt the technology should keep a few things in mind when looking for a vendor to build it.
When designed well, chatbots can offer businesses significant benefits in regard to customer relationships. The technology automates the resource-intensive role of customer support, which for consumers, makes communication with businesses easier than ever.
However, for chatbots to meet consumers’ complex needs and achieve widespread adoption, companies must first gain a deep understanding of their customers – and the conversations they would engage in. This means that companies should know exactly what their goals are with the chatbot before beginning development, as well as how exactly the technology could best serve their customers.
Accordingly, businesses need to find UX designers who take the time to research these customers, create a prototype, and test it with them. This will enable deeper understanding of the specific conversations that customers will have, which can then serve as the basis for the chatbot’s design. Depending on the chatbot’s purpose and type of end user, there will often be a number of subtle differences in language that designers must take into account for the technology to function effectively.
Additionally, businesses must consider what type of personality and tone the bot will have. Since chatbots are often used by younger individuals, many bots, such as the one from Whole Foods, rely on emoticons to engage their users. For other bots like the one from American Express, which deals with serious financial matters, the same approach wouldn’t work; more technical or formal language better suits that specific audience. At the end of the day, bots are ambassadors of their companies, which is why they require careful consideration from the branding perspective.
Ensuring a positive customer-to-chatbot experience ultimately hinges on to what depth the company and developers understand the language that will be used. As such, a great starting point when building a chatbot is to look at transcripts of conversations from a business’s existing online chat feature – that is, if it has one. Most worthwhile platforms make these transcripts easily available, which designers can later mine for a trove of customer language intricacies.
However, even for businesses without an online chat feature, experienced UX designers should be able to propose dialogues from scratch to get the ball rolling with the project. Designers who are up to the task will have numerous projects under their belts, and will have the hands-on experience required to understand customers’ language habits.
Businesses should never skimp when it comes to protecting their customers; however, considering last year’s Equifax data breach of 143 million Americans, the recent Facebook scandal involving Cambridge Analytica and the European Union’s redoubled commitment to data protection and privacy with GDPR, data security has taken on more importance than ever.
There’s no doubt everyone is already on high alert regarding data security. But it doesn’t help that there have been incidents in the past where automated assistants, such as Google Home Mini, have, unbeknownst to their users, mistakenly recorded everything to the cloud. And while chatbots are distinct from voice assistants like Google Home or Alexa, the risk still exists that data stored on the cloud could be accessed by hackers and used for malicious purposes.
As such, companies today must be wary of where their customer data goes, so as to avoid violating customer trust, as well as the crippling fines and public relations disaster that could result. Depending on the types of conversations for which a chatbot is designed, businesses must carefully decide whether they want to deploy their chatbots on-premise or in the cloud.
Successful chatbot endeavors may start with a strong design and secure data, but they don’t end there. No matter how well a bot is designed, customers will find 1,001 different ways to talk to it. It is critical to plan for this by ensuring that vendors keep a team on the project to monitor the bot after it’s gone live. Only once the bot is streamlined should the team be let go. In other words, its prudent for businesses to enter into a long term partnership with a solid vendor, which will provide end-to-end cognitive services – from design, to development, to quality testing, to maintenance.
Chatbots should be able to log all conversations and provide analytics about the customer journey in a way similar to Google Analytics. There are numerous interesting insights hidden in the language that isn’t recognized by the bot. A well-designed solution – and with the help of a vendor – businesses should be able to track this data, as well as the number of unique users, their return rate and the number of dropouts. And if the bot has an option for human to take over the conversation, it should also track how often user falls back to it.
With hyped-up technologies like chatbots, many try to take shortcuts in a rush to go to market. In doing so, however, they tend to threaten the technology’s longevity by breaking consumer trust. But by engaging experienced UX designers, leveraging existing conversation transcripts, protecting data and following up with improvements post-deployment, companies can be sure their chatbot solutions will succeed.
The article was originally published on Hackernoon.com Read also: How Intelligent Automation Changes the Game for FinTechs