Today’s customers are impatient. They want to — and have the right to — get
answers to their questions and concerns about a company’s products and
services without being in the need to perform lengthy searches or to dig
around. This holds true for pre-purchase questions as well as to
post-purchase questions.
Enabling this is an important part of good customer service. And good, or better yet, excellent, customer service is one of the core drivers
of a positive customer experience.
But how does a typical service flow look like, from the customer’s point of
view?

I have an issue. Hmmm, let’s Google this. Maybe there is a solution on the
web, or in their community? Bummer, nothing. Do they have self-service, or
can I search the web site? Ahh, here, something comes up! Oh, dear, that is
not quite what I look for. Is there a chat? No? Ok, then let’s call them…
There are scores of studies confirming that customers want their time
valued and that they will abandon online purchases if their questions are
not answered quickly and easily, or that they will stop doing business with
a company altogether because of bad experiences, sometimes even after a
single bad experience, e.g. a shocking 89 per cent according the 2011
Rightnow Customer Experience Report; or 62 per cent according to the
2015 Parature Global Customer Service Report.
On a more positive note, Forrester Research finds in their 2016 report on The Future of Customer Service that 73 per cent of customers consider valuing their time as the
most important way to provide good service.
At the same time, customers are
starting off with search and self-service, while talking with a
service agent on the phone is still vastly popular. A non-representative
poll that was conducted during a webinar that I recently attended also
confirmed this.

Source: What Should Your Business Be Thinking About to Grow BIG in 2017?
The number of customers turning to chat after self-service fails is
steadily increasing and has reached 65 per cent already in 2015, up from 38
per cent in 2009, according to Consumer Technographic Lifestyle Data 2009 –
2015.
So, as per the scenario above, customers first try to educate themselves
before contacting a company via phone, chat or mail.
It is the Customer’s Way – or No Way
As I have
written before, an important part of providing good service is being available to help the
customers on their preferred channels, at the
time of their choosing, and at their pace.
Theirs, not the company’s!
This includes that a customer, who initiates a conversation, or engaging in
a conversation that is initiated by a company, may not respond in a while,
or chooses to continue using another device, or both. On the other side a
customer will not accept the company being unresponsive or losing
information she already gave during handovers between different service
agents. Customers expect the company to remember information about them –
when it suits them, not only the company.
The conversation between a company and a customer is both, asynchronous and
asymmetric. Just it is no more asymmetric in favor of the company, but in favor of the
customer instead. And the company needs to deal with it.
The way to do this is an
integrated, intelligent, and efficient service offering that helps customers getting to the information that they
want, at their pace. Efficient, integrated and intelligent are the key
words here. The chat equivalent of holding music is not an option. Scaling the call center infinitely isn’t either.
Then how to achieve or maintain positive customer experiences in service
engagements?
What Does it Mean?
The customers’ choice to increasingly prefer self-service over assisted
services and then proceeding at their own pace is a huge opportunity for
businesses.
It allows them to build communities and enable service automation via bots,
with both being backed by a learning system. Knowledge contributed by human
agents trains the bots, which in turn free them to do the hard work and
support them in the ongoing solution finding process. Communities and their
body of knowledge help reducing the number of calls to the call center
altogether and bots help call center agents to scale beyond one
conversation at a time.

At the core of the technical part of the solution are knowledge bases,
machine learning and bots. Service agents obviously still need to be well
trained, perhaps more than ever, as they will get an increasingly higher
share of difficult to answer questions. Routine matters need to be taken
care of by the machine.
In 2017 we can take it as a given that chatbots will become mainstream in
service solutions. As covered by Gil Press, Forrester Research’s recent AI technology radar has placed
virtual agents, machine learning platforms, robotic process automation and
text analytics/NLP into the growth stage of their life cycle. Speech
recognition and semantic technologies are out of the baby throes while
natural language generation is still emergent. All these technologies are
relevant.
The future of assisted customer service clearly sees bot-supported
interactions with the bots being the first point of contact of assisted
service. The bots gather — or already have access to — relevant contextual
information coming from the customers’ current and previous interactions
with the company. They then, based upon the available data and information,
and the internal knowledge base plus other sources, e.g. communities,
determine answers for the simpler questions, and seamlessly hand over the
too difficult ones to human operators.
And they learn from the proceedings of the ongoing conversation.
Making it Work
Putting all this together there are four main courses of action that
business leaders need to pursue to gain both, effectiveness and efficiency
in customer service, thus creating lasting positive customer experiences:
1. Think of creating a community. Not every company is big enough or has
sufficient resources, but a community is a strong helper when it comes to
self-service – and self-service is clearly on the rise.
2. Consider the knowledge base, covering internal as well as external
(community) knowledge, a very high priority. Data and information needs to
be attributed and converted to knowledge so that external (Google, Bing,
etc.) as well as internal search engines can find relevant information and
put it to the top of the result list. This is a learning system utilizing
an intelligent agent that learns from the queries and their results. This
system will also help the call center agents. Knowledge management is
absolutely the foundation for self-service as well as well as assisted
support.
3. Strongly consider chatbots that use contextual information and act as a
first line of interaction, both in customer initiated as well as company
initiated interactions. Minimally they keep the customer engaged and do a
seamless handover to a human agent. This does not only help the customer,
but the agent, too, as she is prepared and can dig right into the problem.
As the chatbots are supported by a learning system the problems they solve
can become increasingly complex. Additionally the bots do support the agent
with their ability of continuously suggesting good solutions based on the
knowledge base, suggesting tags for new articles, etc.
4. Human agents need to be well-trained, as the first level service will
increasingly be taken care of by automation. Focus on effectiveness, not efficiency. The chatbots will benefit from the knowledge they create as
their search patterns and recommendations become learning material for the
bots.
Some really interesting thoughts! Would love to see this play out in reality – have you got any examples/case studies of companies using bots for customer support? What issues did they run into, and were they able to measure an increase in customer happiness and retention?
There are multiple really helpful use cases available for AI Chatbot+Human integration, where up to 90% of the traffic was handled by the Chatbot.
I did one similar project a few years ago, the key there was that the best results were measured in environments with relative simple Q&A’s (eg NAP change, product upgrade, send brochure); I question if there are any examples of AI handling really complex, multi-input use cases (eg outage analysis). Tend to think we’re still in the “little AI” phase. Certainly promising, but for me not the end of the physical customer care agent, yet.
Helena, thanks for your comment and question. I do think that we are very much talking about an emerging market. Down here in NZ I do know of two proof of concept implementations, one geared a bit to tracking and tracing of deliveries, accessing an OLTP-like system, and another one answering questions based upon a knowledge base. In Australia is a big live governmental implementation that even uses Natural Language Generation. Some contacts in helpdesk software for smaller companies tell me that they see not much demand but that they are anticipating a rise, preparing to build a solution. A fairly big retailer (well, for the ANZ scale) tells me that bot-support for their call center is not on their priority list at the moment. But then, knowing them, their priorities can shift fairly fast …
Putting a vendor name here (and no, I am not paid by them ;-) ) – I know that Microsoft has a bot that connects to Dynamics CRM and is capable of e.g. creating cases right out of the box.