Building a digital expert to power client self-service
Why have so few companies managed to give us a Google Assistant, Alexa or Siri experience in the customer service arena?
As we fast-track into the digital era, the term 'artificial intelligence' is increasingly being bandied about. It truly sounds like magic. Having technology that can learn from millions of pages of documents and answer questions related to this content within a matter of minutes. Or review thousands of hours of video footage and through face recognition and behavioural patterning predict which person in a crowd of thousands is most likely to cause a disturbance, and do so more accurately than highly trained and experienced agents.
It's Hal on steroids, and it sounds like it is coming to a theatre near us, very soon.
In many areas it has already arrived. Where information sources, both written or in image formats, are available in both quality and quantity, we have artificial intelligence (AI) operating in incredible ways. Just consider Google, Facebook or Amazon, which are powered by AI that feeds off global data on the Internet.
Ask Alexa, Google Assistant or Siri a question. If the information is documented somewhere online, you will in all likelihood get the correct answer. It could be a simple question such as 'what is the capital of Bhutan?' or 'what is the top-rated Indian restaurant within 1km of me?', or it could be a very specific instruction such as 'book an appointment with Jeremy Jones on Wednesday, 13 March 2019, and send him an e-mail notification'.
Log into Amazon and type in a product category that you have bought from before, say, 'business books'. Now ask a friend from another city who has a well-used Amazon profile and is more of a psychology and self-help book reader to do the same. You will be amazed at how, in all probability, you will be offered a different list of books with possibly even different recommendations and prices, largely because of the data Amazon has about you which it uses to tailor your experience.
AI is alive and well, and shaping our digital experiences on a daily basis. The ability to not only understand our accents, but also our requests, and then have these actioned for us, is truly astounding. It's like having the coolest personal assistant ever. You just fire off instructions and viola, your wish is his/her command.
Why then have so few companies managed to give us a Google Assistant, Alexa or Siri experience in the customer service arena?
AI is alive and well, and shaping our digital experiences on a daily basis.
Customer service logic is different to the logic powered off data on the Internet. It is typically bound and shaped by complex product, policy and procedural rules. These rules dictate not only a specific logic sequence, but also a large set of rules that must be followed in a certain way. It is what we call prescriptive logic, while the logic that powers most AI is predictive, generated from huge sets of unstructured data.
Secondly, the quality and quantity of content that describes the customer support rules is seldom enough for an AI engine to learn off and accurately apply across multiple customer situations and contexts.
You tend to get different levels of product, policy and procedural documentation with different versions and conflicting rules stored in a bulging knowledge base. As a result, it requires people to train the logic.
Given the explosive nature of customer variables and possibilities, however, this is proving very difficult to achieve. Too often the digital advisor offers you a 70% or even 80% accurate experience, which is not good enough to go live, especially in regulated industries.
Finally, customers are not patient users. Unlike the millions of people who have unwittingly volunteered their time to train Siri, Alexa and Google Assistant, most companies only have thousands or, if they are lucky, a few million customers they look to serve. Few of these customers will be willing to train the company chatbot so it can more effectively understand what it is they are actually asking for, so that it can then respond more accurately.
This means most companies are still stuck with generic support offerings, such as frequently asked questions or heavily decision tree'd customer journeys that seldom solve the customer's contextually-rich challenge. And as a result, customers continue to default back to contact centres and retail consultants.
That is until recently. Technologies now exist that allow non-coding teams to rapidly build a digital expert based off structured (as opposed to unstructured) data. These digital experts can ask the right questions at the right time, and based on the responses, consistently guide the customer to the right solutions and trigger the right actions.
To achieve this, the developers of these solutions have had to move beyond the limitations of knowledge bases and decision tree scripting logic, and have instead found smart ways to build complex logic off the soul food of digital intelligence: data.
This breakthrough is allowing companies to build powerful digital experts that will do exactly what they want them to do, in context of their customers, knowing the engagement will be in line with all their prescribed business and compliance rules, and that they will have a detailed record to prove it. And they are achieving this in weeks, not months.
This shift from trying to build a digital expert off outdated knowledge bases using machine learning, to one where the expert logic is built off structured data tables that can handle millions of predetermined customer journeys, is unlocking customer service within complex, regulated organisations. They no longer need to spend millions trying and failing to train their own Alexa off limited training data.
Now, there are no more excuses when it comes to offering customers consistent, compliant and context-relevant support experiences, when they want them, and where they want them. All delivered flawlessly by your very own digital expert.
Ryan Falkenberg is co-founder and co-CEO of CLEVVA, a company that specialises in artificial intelligence for people. Prior to founding CLEVVA, he co-founded Hi-Performance Learning, which created new ways for staff to learn more in less than half the time. In 2014, Falkenberg co-founded CUDA Technology, a company that developed enterprise learning and knowledge management technologies. He and his brother and long-time business partner Dayne sold both HPL and CUDA to form their Human Capital business. The two partnered again to start CLEVVA in 2011.