Think you’ve seen one chatbot, you’ve seen them all? Now’s the time to think again…
2020 has been a pretty turbulent year so far. The Covid-19 pandemic brought with it unprecedented months of lockdown, projections of 1980s level of unemployment to come over the next stretch and mass payment holidays for customers struggling to make their repayments. For the asset finance industry this has meant a sustained reduction in new business (figures released by the Finance & Leasing Association (FLA*)) show that total asset finance new business fell by 11% in September 2020 compared with the same month in 2019 and overall in the first nine months of 2020, new business decreased by 27% compared with the same period in 2019.) At the same time demand increased on customer services as worried customers sought to contact their lender.
Having weathered the initial storm, many are eyeing the next few months with a degree of understandable trepidation. With local lockdowns affecting increasing swathes of the country, a no-deal Brexit lumbering ever closer and rising unemployment as furlough winds up, finance companies who have already been running at 110% supporting their customers can expect to have even more asked of them. With costs rising, increased debt provision squeezing budgets and new business falling, just throwing more people at the problem isn’t a viable option.
The capability of conversational AI’s (e.g. chatbots, voice assistants etc.) has made a quantum leap forward in recent years thanks to Natural Language Processing (NLP). As the name suggests, NLP works much like a person listening to speech by breaking down the sentence structure, using knowledge of idioms and grammar, and recognising patterns to parse and interpret the underlying intent of what the user is trying to say. The old-school style of glorified FAQ bots have been superseded by the next generation which, incorporating NLP, can now conduct complex multi-turn based and multi-topic conversations and also perform actions like taking a payment or giving a quote. This means they can now be applied to a much wider variety of use cases than their more primitive predecessors.
One powerful example is as the initial ‘greeter’ on a website (just the way a person in a physical showroom would come and say hello and offer assistance) to customer services. Retail salespeople have long understood the power of engaging the customer in conversation and using this approach online, Škoda have increased website conversions an impressive 400% with their virtual sales assistant Laura, resulting in a big uplift in sales for the brand and is now available in multiple languages and countries.
The sophistication of voice assistants like Google Assistant and Alexa can also be applied to responding to more traditional channels like the telephone via Speech to Text conversion and back – for example Bank of America’s Erica: a customer representative who is contactable 24/7 by voice or text through the bank’s mobile apps. This allows for a real conversation between a customer and a virtual agent, rather than the classic ‘press 1 for x’ IVR (Interactive Voice Response) system, which so many of us find so frustrating. Using an omni-channel and multi-language platform means you can expose the same conversational AI agent via WhatsApp, SMS, Google Assistant, telephone/IVR and classic web chat for no additional effort and in multiple countries with minimal additional effort.
Integrating that virtual agent with other systems like CRM or a central contract management system can then give it the power to look up information and perform actions as required – for example, verifying a customer’s identity for DPA purposes, enabling them to self-serve changing their address and phone details, getting a settlement quote or even extending their contract. Authenticating a customer can be further simplified by use of biometrics – e.g. ‘My voice is my password’ – making for a greatly enhanced customer experience.
Another value-add would be in replacing cumbersome and inflexible IVR platforms with NLP powered virtual agents. Common requests can be handled by an easily scalable virtual team of AI agents while human customer services operatives can give more time to the customers that really need it with more complex tasks. Reducing the load on those human agents also means reducing hold times as more calls can be dealt with by the virtual agents and even those calls which do need to be transferred to a live agent are shorter as the customer has already been authenticated by the AI.
What about virtual assistance?
Virtual assistance can also be a powerful tool in collections, allowing customers to be contacted by SMS or WhatsApp by a virtual collections agent. This increases both the efficiency and reach of collections contacts and can be less uncomfortable for those being contacted.
These are just some of the interesting use cases we are seeing for NLP powered AI and over the next few weeks, my colleagues and I will be exploring some of these use cases in more detail starting with the use of conversational AI to support customer services agents.
The technology is finally catching up with our aspirations of a Star Trek style computer that can answer questions and perform actions. With the turbulence set to continue for the foreseeable, now is definitely the moment to augment your digital sales channels and customer services with some virtual assistance.