6 minute read

So you want a CHATBOT…

by Daniel Mpala

Here’s what you should consider about introducing the technology to your business

If you run or work in a company that frequently communicates with a large user base you probably want a way to quickly and instantly communicate with and perhaps resolve complaints with your customers around the clock.

YOU MIGHT even have come across interesting data like findings by global research and advisory firm Gartner that by the end of the year up to 85% of customer interactions will be handled by virtual customer assistants or chatbot technology. Or the more C-suite enticing prediction by UK-based market research and business intelligence company Juniper that chatbot conversations will contribute up to $8-billion in cost savings by 2022.

The emergence of conversational AI, and by extension chatbots, has enabled those across the retail, financial services, credit, insurance and travel industries to provide personalised customer experiences at scale. As FinChatBot co-founder and CEO Antoine Paillusseau points out, this is a “booming area” due to the advances that have been made in natural language processing (NPL) and speech generation, not just in South Africa but globally.

FinChatBot’s Antoine Paillusseau

FinChatBot’s Antoine Paillusseau

The Johannesburg-based firm’s work with major African banks and insurance companies has resulted in case studies showing up to 29% revenue growth and 93% customer satisfaction.

“Most businesses understand the importance of digitising certain aspects of their business and are taking the necessary steps to investigate and implement solutions. “The success of these solutions, however, depends on a variety of factors such as suitability of the solutions and the going support and optimisation to ensure a positive customer experience,” says Paillusseau.

But, what should businesses looking to incorporate the technology consider? How do you end up with a great customer service tool and not a user frustrating gimmick?

Build to help users

Nick Argyros, the CEO of Johannesburg-based GotBot – which develops chatbots for the retail and financial sectors – says companies get the best success out of their bots if they build them with the intention of helping their users.

His firm too has seen an uptake in the use of the technology in South Africa. GotBot COO Craig McLeod believes chatbots will become a commodity-based technology that will need to be incorporated into the business customer care strategy. “Many large or medium sized businesses are focusing on efficiencies and automation in order to help with customer engagement,” he points out.

Argyros points out that chatbots are capable of efficiently processing tasks at “lightning speeds” while collecting data.

“Some examples of these tasks are answering FAQs and processing complaints, marketing tasks such as responding to targeted adverts and competitions, sales processes such as onboarding new clients and helping customers find new products, as well as financial services task such as providing statements and invoices as well as more complex functions such as customer profile vetting for credit applications,” he adds.

GotBot’s Nick Argyros

GotBot’s Nick Argyros

Decide on what to automate

For Paillusseau, the first step is to decide on what to automate. He says it is crucial to try to find a “quick win” that will have a strong impact on a company’s bottom line, like improving sales or enhancing the call centre process.

Argyros shares a similar view. He believes “a lot of thought” should be put into what the problem being solved actually is, and that it is important to identify where the technology will add most value to the business.

Paillusseau says it is also important that companies evaluate their current customer journey in line with what their systems provide. “You don’t want to change your whole internal process and system,” he cautions.

Partner with experts

“Partner with the right experts to ensure the successful building, integration, testing and optimisation of your AI chatbot,” adds Paillusseau.

He says there are scenarios where some companies have opted to develop a chatbot solution in-house, instead of partnering with an expert. “Without the necessary expertise, focus and optimisation, businesses can struggle to ensure their chatbot delivers the best customer experience,” says Paillusseau.

As GotBot account manager Nick Thomson points out, building a successful chatbot is heavily dependant on a few skills.

“If a company is looking to incorporate technology like this, the most important aspect is that the team includes someone who is passionate about innovating and using this technology to effectively connect with their audience.

“The bot needs to be built with the intention of solving a problem for the users,” says Thomson.

McLeod advises that companies need a product owner to steer the build and administration of the chatbot.

“It’s imperative that an individual with data knowledge is able to decipher the chatbot reports in order to build better and more efficient flows,” he adds.

Watch Video: https://www.youtube.com/watch?v=B0K4yFjmcek&list=PLhuckWf9CH6ugCNBIWrqV1DiENBESRndJ&index=9&t=0s

Watch Video: https://www.youtube.com/watch?v=B0K4yFjmcek&list=PLhuckWf9CH6ugCNBIWrqV1DiENBESRndJ&index=9&t=0s

Costs and ROI

Incorporating an AI chatbot into a business does not come cheap. Initial set up costs can range from R50 000 to as much as R350 000, while rental of the software can set back a business from about R15 000 a month depending on the level of integration, agents and channels required. As such most businesses will want to know how the return on investment (ROI) can be calculated.

Argyros says ROI and key performance indicators (KPIs) vary from client to client.

“With some clients it’s the rate of response that needed improvement with speed and efficiency being the measures of success. Another measure is the rate of conversion in terms of online and social commerce sales,” he explains.

McLeod says in order to quantify ROI, you first need to identify what the function of the chatbot is.

For example, in instances where the bot is primarily designed to be a sales bot, the ROI is based on sales. Likewise, if the chatbot is designed to service inbound customer requests the ROI will take the cost per agent into consideration.

Things can go wrong

As with any technology, chatbots are not without their problems, especially when not implemented well.

Thompson says with chatbots, the biggest problem is almost always around a lack of training. He points out that there are two important phases when building a chatbot. “The initial phase is when starting to build the bot, the second is refining it once there is more user data. In the build phase, the scope of the bot needs to be considered. “If there isn’t a clear goal, then the bot has huge potential to be a gimmick which ends up frustrating users instead of helping them,” he adds.

“An example was a bot deployed on Twitter in 2016, it was shut down within 24 hours - 16 exactly- because it’s training model was almost entirely autonomous, and allowed unvalidated human input.

“Many chatbots today still require a huge amount of human training in the build phase, and then again in the post launch, where the bot is refined once there’s additional training data,” says Thompson.

But, as MacLeod explains, over complicating the chatbot logic and training can also have a negative impact.

“Many chatbots are built around a specific language and if you are having to cater for multiple languages it presents an administrative challenge,” he adds.

Personalisation is the new gold

Businesses that get their chatbots right stand to gain an edge over their competition.

Chatbots, Paillusseau explains, have the ability to personalise interactions and connect with audiences in their native language which he says garners greater understanding and trust from customers, especially given South Africa’s cultural diversity.

“In marketing, personalisation is the new gold, and companies can achieve an increase of up to 80% in engagement rates,” says Paillusseau.

Chatbots are also convenient and cost-effective as customers won’t need to download a separate app to engage with businesses.

He adds that when human involvement in customer interactions is reduced, productivity, quality and speed improve.

“Furthermore, conversational interfaces also provide an easy and economical way for businesses to receive customer feedback,” he says.

He however cautions that despite the enormous potential of AI chatbots, there will always be need for “a human touch” when it comes to interactions with customers.

“This means an ability to ensure seamless agent handover and other integrations for the best customer experience,” says Paillusseau. ai