Waiting for Godot, is a play written by Samuel Bennett, that premiered in English in London in 1955 and in a poll conducted by the British Royal National Theatre in 1998/99, it was voted the “most significant English-language play of the 20th century.” Without wanting to be a total spoiler, the main theme is that two men are waiting to meet the mysterious Godot, who in the end never arrives. As I normally write about customer experience, you may well wonder what this has got to do with anything.
Vivian Mercier described Waiting for Godot as a play which “has achieved a theoretical impossibility—a play in which nothing happens, that yet keeps audiences glued to their seats. Very much like many of us when waiting on hold in a contact centre queue. This was brought to life In the 2008 best seller Nudge, by the legal scholar Cass R. Sunstein and the economist Richard H. Thaler who marshaled behavioural-science research to show how small tweaks could help us make better choices. An updated version of the book includes a section on what they called “sludge”—tortuous administrative demands, endless wait times, and excessive procedural fuss that impede us in our lives and is referenced by Chris Colin in his recent article in The Atlantic entitled That Dropped Call With Customer Service? It Was on Purpose. The article and the myriad, frustrating, teeth grinding examples of sludge, may want to make you to lose the will to live, but stick with it, it has an reasonably happy, (well satisfying) ending.
Just ask yourself this question. How many times in the past month, or even the past week, have you had a customer service interaction where you waited patiently, or not, if you were lucky enough to find a phone number, or a chat link, and would have been delighted if someone or something had answered your seemingly simple inquiry. Yes, I mean a chatbot or even better, an Intelligent Voice Agent (IVA). (even though the article above is not particularly friendly towards our robotic friends)
The more astute among you will probably see that I’m taking you down the Conversational AI road and as the cacophony of noise about Chat GPT, Gemini, Co-Pilot etc., reaches peak volume, there are equally or more voices in the Customer Experience (CX) world who will sniff and then say that “we only want to talk to humans.” Naturally we do, but what if there aren’t any people? Or if they can’t answer questions accurately, confidently, articulately, or at 2:00 am? Then what? Nobody is handing out refunds for wasted time!
I recognise that for those of us who have experienced poorly designed and underperforming IVR systems and chatbots whose most frequent refrain is “I’m sorry I don’t understand that”, are still sceptical about these tools. This is primarily because many of our experiences have been with chatbots that are programmed to handle general questions and are not able to handle the depth and breadth of inquiries that modern contact centres handle. These ‘general’ models cannot handle diverse intents and different ways of asking questions and lack the contextual sophistication needed to respond to even simple inquiries. As a recent article in the Financial Times noted, “chatbots can perform limited transactions, but they can’t have a real conversation. Humans, supported by new technology, still have a lot going for them.”
Herein lies the difference. A well designed and well-trained IVA that continues to learn, just as a human does, can handle myriad inquiries, answer questions, complete tasks, and, apart from delivering superior customer experiences, eliminating mundane tasks for employees, and reducing costs, an IVA has other benefits.
Being heard. Automatic Speech Recognition (ASR) engines ensure precision transcription by customising IVA models for your business. This ensures exceptional levels of accuracy, enabling digital agents to understand your callers, whatever their accent or dialect. Trained for customer-specific use cases, digital agents can understand a diverse range of accents and dialects. They can also handle poor audio quality and provide precise transcriptions with over 90% accuracy, well ahead of general language models that aren’t designed specifically for call centre usage. This is critical for those operations such as banking and insurance, where clear and accurate wording is paramount to successful interactions.
Being understood. Natural language understanding (NLU) enables digital agents to identify your callers’ needs quickly and empowers them to follow the flow of conversation and offer accurate, empathetic responses. Digital agents are trained on your customer-specific historical data and deliver a natural conversational experience, both in understanding and replying to your customers. They have contextual understanding with memory and recall and can work through tasks logically – just like a human agent would. They can also handle multi-turn conversations where a customer may ask several questions simultaneously. “ Can I make a payment on my phone bill, and can you let me know if my salary has been deposited and the amount.” Human often speak this way and it’s vital that an IVA can both understand and respond effectively and accurately.
Convenience. Around-the-clock availability, no long hold times and a streamlined path to query resolution ensure that your callers’ queries are resolved quickly and effectively on their channel of choice. What’s more, any interaction can be seamlessly transferred to a human operator if required when complex, or when requested by the customer. The IVA can provide a transparent and auditable conversation path. This path is established in real time to continually drive the conversation to achieve accurate query resolution in the shortest possible time frame.
Sentiment and emotion detection. Advanced sentiment and emotion detection tools allow digital agents to create personalised caller interactions that feel natural and human. While this is not to say that IVAs can express empathy in the same way that a human can, they can provide you with valuable customer behavioural analytics on what may have triggered a specific customer emotion such as anger, disappointment, relief, happiness.
While an IVA can be deployed in many applications and business segments here is a summary of the industries and use cases that are worthy of consideration:
Industries:
Banking, Insurance, Health Care, Public Sector, Housing Associations, Debt collection, Logistics and Food Delive
Example use cases:
Inbound: Caller identification and verification (IDV), customer self-service, end to end processing of insurance claims, logging tickets and schedule maintenance, order tracking and returns, gathering information before handover to specialist agent.
Outbound: Sales validation, customer satisfaction surveys and after sales follow-ups.
Simplicity and Customisation:
With most callers still preferring to communicate with contact centres using voice to get the “right” answer, a voice-based digital assistants can offer a human-like caller experience, especially if staffing constraints or inconvenient opening hours restrict availability of human agents. Just as with a human agent, the IVA does require some training, and conversation design becomes a vital element in successful IVA deployments. Conversation design is based on human conversation. moves beyond pre-defined scripts and uses natural human language to simulate how people communicate in real life. The more an (IVA) can simulate human conversation, the less users need to figure out how to use it.
Current IVA models from a number of providers are custom trained for each specific use-case and the varying dialects of callers, to provide unprecedented levels of transcription accuracy and language understanding – fluency. The tools to empower customers to build customised deep learning models and sophisticated conversational dialogues, are no-code/low code drag and drop programs that don’t require teams of developers.
The first use case for an IVA doesn’t need to be complex and in fact, despite the capabilities listed previously, the best route is to start with low complexity, high volume tasks and consider automating internal tasks to gain support from colleagues who know what will work best in their scenario. Then you can evolve that via active learning as the IVA collects more caller data and expands human language understanding.
Developing and deploying the IVA:
There are a number of actions that an organisation can take in coordination with their front-line teams to determine if and how an IVA can be effective for both customers and colleagues:
- Identify, through measuring the end-to-end customer journeys, where demand failure exists and why. (e.g. waiting on hold to change contact details, reset password or other basic tasks)
- Explore where Conversational AI (CAI) and/or an IVA could reduce demand failure.
- Ask advisors what calls do they hate taking, what takes the longest and potentially upsets the customers most?
- Be very clear on your use cases and the measures of success for each one
- Have them describe how AI could help them, and customers, to deliver better customer experience
- Engage the troops – Don’t do this is a dark room
Organisations shouldn’t be investing in an IVA or other types of AI just to save money or cut staff. It must be about making life better for customers and colleagues. The bottom line to any IVA deployment must be whether the customer has achieved what they want, with a minimum of effort and is as easy, perhaps even easier, than speaking to a human. It’s not for every interaction and the customer will always be the best and only arbiter of whether it’s been a good experience. But if the alternative for many is wasting precious time via a long Godot like, slow moving queue, while being “consoled” with “your call is very important to us”, and then finally getting through and finding out that the computer still says “no”, and an answer never arrives, then perhaps it’s time to have a chat with an IVA – and it won’t just be all talk!