Not All Chatbots Are Equal — How to Choose Wisely

Despite the hype, and the considerable investment of time, dollars and resources, many chatbots are failing to meet customer expectations.

This isn’t just my opinion. Gartner predicts by 2020, nearly half of all chatbots and virtual assistants launched in 2018 will be abandoned. Forrester expects, in the coming year, that we will see more “failed chatbot experiments offering degraded self-service experiences.”

Build better voice apps. Get more articles & interviews from voice technology experts at voicetechpodcast.com

While the promise of an AI-powered chatbot experience is very much worth pursuing, it hasn’t been realized in a material way to date, by most organizations. There is growing consensus that the chatbot ‘experiment’ of recent years — that is, the home-grown, automated, self-service FAQ widget — has introduced more customer frustration than positive engagement.

Why these challenges? Why has true, AI-powered automation not been introduced across the customer journey? Why are chatbots failing to meet customer expectations? And most importantly — what can organizations do today to make sure they are able to leverage fast-evolving technology and be ready for tomorrow?

I have a few thoughts (largely influenced by the successes of our customers at Ada, where I head up marketing):

Customer automation, chatbots and virtual assistants should be built, managed and tracked by the teams responsible for driving engagement.

The good intentions of IT resources notwithstanding, the developers and architects don’t really understand the needs and interests of your customers. If technology is meant to be empathic and approach human interaction it must reflect a genuine understanding of need at each and every stage of the customer journey.

Who is better equipped to tailor and customize that engagement than the business stakeholders (think Marketing, Sales, CX) who own the customer relationship?

The problem is that most chatbots require deep technical and coding skills, and so teams have become dependent on IT to provide customization and personalization. This broken development chain results in a disappointing, frustrating and often disjointed user experience.

The good news is, there are platforms that can be easily configured, personalized and tailored to client needs by marketing, sales and CX team members themselves. An intuitive, easy-to-use and highly customizable backend empowers marketing, sales and CX resources to train, launch, manage and track a complete automated customer experience. And indeed, customers using Ada are building Automated Customer Experience (ACX) teams to this end.

Companies misunderstand what true AI really is and where/how chatbots fit

Every customer is different and AI is only as good as the data that drives it. You cannot flip a switch and suddenly introduce an automated customer experience that genuinely reflects a customer’s unique needs in a complex and intelligent manner. Companies need to invest in data, integration and personalization to reap the benefits of AI today, but even more so long term.

Not all chatbots are created equal, and most are not designed in concert with a scalable, enterprise-class and data-driven AI strategy.

Homegrown, unsophisticated chatbots are a dime a dozen and usually nothing more than glorified FAQ platforms without any smart capabilities behind the scenes. Companies are failing because they are looking for quick fixes rather than embracing chatbots as an integrated part of their broader AI strategy.

To avoid these pitfalls, organizations (especially large, complex organizations) need to invest in platforms that persist the data, leverage enterprise data and systems out of the box and offer a highly personalized user experience that improves over time (as a result of machine learning).

Most chatbots don’t generally get smarter over time

It is somewhat inevitable that a chatbot or virtual assistant is going to misunderstand a customer’s inquiry. It can become a slippery slope, however, and businesses need to protect against alienating users when this happens.

With that in mind, when a chatbot gets stuck in a loop and is unable to meet a customer’s expectations, there must be a way to prevent this from happening again. In other words — the interface must learn and improve over time.

Rarely is this achieved with the tools on the market today, and never with those that have been home grown and disconnected from a broader AI strategy. Moreover, machine learning tends to rely on the expertise of the development team, and business stakeholders are not generally a part of the process.

Business stakeholders should be able to review past interactions and easily (without any code) retrain the chatbot to overcome those hurdles going forward. With the right tool, this process can literally be achieved in an instant and if done correctly, becomes an iterative and sustainable process that ensures the viability and continued improvement of the automated customer experience long term.

Companies are still embracing the promise of automated chat and adoption will certainly grow in 2019. However, a thoughtful, strategic approach to AI and a broader automated customer experience is absolutely necessary to ensure that fast-growing organizations are well positioned to execute against a digital transformation imperative while prioritizing the needs and interests of their customers.

Not all chatbots are created equal and companies would be wise to think through the broader application of automation across the customer journey and make their tooling decisions accordingly. To learn more, I encourage you to read the latest e-book we published at Ada.