Page 28 - EngineerIT December 2022

TECHNOLOGY TO IMPROVE CUSTOMER SERVICE

Virtual Agents as a Service:

why it may be the ideal model

for your business

By Ryan Falkenberg, CEO, CLEVVA*

growing number of companies in South Africa and around the globe are looking
to offer their customers significantly improved self-service. Many have tried,
A and failed, to build a chatbot capable of doing anything more than responding
to basic information or transactional requests. As soon as a customer has a more
complex query, issue or complaint, the chatbot immediately hands over to a live agent.
So much was promised with chatbot technology. With low-code teams, you were
convinced that within no time you could have an intelligent virtual agent that could
speak like a human agent and perform like one too. The truth is, it’s not that simple.
While chatbot technology has improved enormously from a natural language
understanding perspective, it’s the brain department that continues to underwhelm.
This is because chatbots are designed to learn off a knowledge base - the same generic
content we use to train and support our human agents who then take months of trial
and error to ‘join-the-missing-dots’. And while chatbots are capable of learning too,
they don’t have the benefit of drawing on years of interpersonal learning that humans
do. They are expected to learn from the responses of customers, who seldom have the
patience to teach them.
Software vendors are very quick to shift the blame to the chatbot developers, or
the content quality within the knowledge base. They highlight the almost magical
superpowers they can showcase when illustrating their chatbots off a carefully
designed and curated demo environment. chatbot that keeps many live agents in
The truth is that building, deploying and managing a virtual agent capable of business? What incentive would they
performing like a human expert, and resolving hundreds of complex queries in a have to cannibalise their own revenue
consistent, compliant and hyper-personalised way, is not easy to do. You certainly can’t stream?
do it using standard chatbot logic.
What you need is technology that can capture and automate expert-level Why a Solution-As-A-Service
thinking, and specialist teams that can build a digital brain capable of engaging at model makes sense
the level of a human expert. When you do, that’s when magic really happens – it’s Faced with a noisy landscape of chatbot
when you see your customers truly embrace your digital self-service, with resulting technologies, all using different words
lowered call volumes, improved customer satisfaction ratings and a reduction in to describe the same things, wouldn’t
your total cost to serve. it be better to approach your digital
self-service channels the same way many
Simple software delivers simple solutions approach your human assisted channels?
Before digging into the advantages offered by a Solution as a Service model, it’s worth Many companies have done
taking a deeper dive into some of the issues with most chatbot software. the maths and realised the value of
The first issue, as noted above, is that chatbot logic is by design limited. While you outsourcing their contact centres to
may hope your chatbot will self-learn to become an expert, don’t hold your breath. specialist business process outsourcing
They will probably only ever manage to offer your customers basic tier 1 support. (BPO) companies. These customer
The second issue is that most CX platforms offer chatbot modules. This allows service specialists eat, breathe and sleep
them to promise a multi-channel offering. However, these platforms are all, at their the complex world of contact centre
core, designed to support human agents and their licensing is largely based on human management.
agents. Why then would they look to offer a chatbot that successfully resolves more They specialise in handling the
queries without relying on a live agent? Wouldn’t it be better to rather have a limited complexity of aligning technology with

EngineerIT | December 2022 | 28