Influential architect Cedric Price once asked: ‘Technology is the answer, but what was the question?’ He was lecturing on architecture at the time, but it’s a useful maxim to consider when thinking about artificial intelligence (AI).
In recent years, the technology has swept through the customer contact community as companies race to deliver the most innovative, competitive customer experiences. Using AI to predict how many calls will come in the next shift, what their intent will be and who should answer them (from type of agent to type of machine) or allowing it to monitor customer calls and offer real-time recommendations, is no longer science fiction.
But is AI worth the hype? Many organisations leap before they look, eager to chase the latest trend, but AI is like any technology investment – it has to be treated with care. Organisations first need to understand what problem they’re trying to solve, and then decide if AI fits the bill.
The right solution for the wrong problem
AI is at the zenith of its hype cycle, and for good reason. In recent years, machine learning has allowed AI solutions to learn from and adapt to their environment, making them suitable for customer-facing roles.
However, hype inevitably creates more hype. Sadly, misconceptions around AI and its capabilities are starting to create unnecessary change in the contact centre.
Consumers are becoming more accustomed to having their calls answered by sophisticated, AI-powered chatbots. Yet a chatbot can’t solve all your communication challenges. In fact, if deployed carelessly it could actually magnify them.
If you’re only implementing a chatbot because the competition has one, it’s time for a rethink. Far from easing the pressure of customer demand, you may only be creating another channel for customers to contact you on. This only adds more to your agents’ caseloads, as they need to monitor and respond to these automated channels. Ill thought out implementations cause greater customer journey friction.
From another perspective, some companies believe AI will reduce their dependence on call agents and allow them to service more customers, faster and at a lower cost. However, when out in the field it has clear limitations and can’t respond as emphatically and helpfully as a human can.
When using AI, the objective should always be to enhance the customer experience and better utilise your human talent, rather than replace it. If you’re looking at AI purely as a way to cut costs, you’re asking the wrong question.
Stop the friction
It’s not the technology you have, but how you use it that counts. As with any technology investment, your decision to use AI should be problem-driven.
Before deploying AI, you should first be trying to spot where the friction is in the customer and user journey. This could be anything from long customer waiting times, a lack of customer information or a multitude of repetitive manual tasks for an agent to complete. Once identified, it’s then time to test if AI can solve the problem.
When deployed strategically, AI can help share your call agents’ load by reducing the number of inbound queries to existing channels. In this respect, AI is at its best when automating low-level customer requests.
The majority of queries agents receive are predictable and routine. Beyond answering these questions with a predetermined response, there is little value they can add. AI liberates employees from these mundane tasks. Instead they are freed up to focus on tasks that benefit from their specialist knowledge and creative-thinking.
AI is advanced enough that it can process the overwhelming majority of these customer queries quickly and effectively. Customers get the information they need while the frustration of queuing to speak with an agent is eliminated.
Ultimately, companies shouldn’t just adopt new technology for its own sake. First and foremost, you must evaluate a solution based on how much it will improve the customer experience. All too often we are asked for customer contact to be underpinned by AI or Machine Learning with no data to quantify the problem (if it exists) to be solved.
AI adoption, similarly, must always be targeted at a well-defined problem before it is trialled and deployed. After all, you can’t solve a problem that doesn’t exist.