In an earlier article, I wrote about service trends and automation. At my company, we're always asked if machines will eventually take away support jobs. Do computers substitute humans or, are they complementary to existing workers?
In the last two decades, as globalization has come into the picture, we have started to compete more as humans. This often means that if there is a lower-cost resource that could do the same job as a higher-cost resource, we switch out those resources. This trend will only continue, as all humans have the same basic needs — food, shelter and luxury items. A machine does not need any of the above.
Innovation is the key to customer support. 82% of those who are in charge of making decisions for their companies believe their business' support practices must evolve to remain competitive. What will your business do to stay relevant, and how will machines be a part of that change?
What Does It Mean To Be Complementary?
Let’s look at a simple example . In 2012, teams competed to develop the best machine learning algorithms to identify dog breeds in ImageNet's library of dog photos. We engineers marveled at what the technology here could do. Then, we paused and realized that a 2-year-old could easily point to any type of dog and verbalize a description without any training. This made us recognize that humans are a marvel; we have a long way to go to make machines the same. We, humans, excel at making plans and forming decisions in very complex situations. Computers aren't quite the same. They excel when there are loads of data to process, but they struggle when it comes to making simple judgments that humans are used to. While we can understand and make sense of multidimensional situations, computers can process multidimensional data. We are categorically different. We are complementary. We have used computers for these complementary scenarios in various parts of our lives for decades — fraud detection, network switching, efficient processing power, fast transactions, etc. As we reach optimal infrastructure levels (our phones are way more powerful than the massive computers of old), many hope to bring this processing power into enterprise decision making.
Our human minds are also slowly being trained to understand the power of what it means to be "complementary." We use computers to narrow down our flight searches, find the fastest routes to a destination, discover efficient ways to find the lowest price for an item, correct our mistakes, find the words to finish our sentences and more. As millennial minds come into the enterprise world, they are questioning how we can use this technology to perform complex multidimensional data analysis and pattern recognition.
The Handshake Between Man And Machine
Let’s look at decision making in customer support and customer service. This support happens across industries and verticals. Whether the support for a physical product or a software product, it is essentially focused on solving problems in three dimensions:
Solve today’s problem, something that a customer has already pointed out.
Make sure the same problem doesn’t happen again by making changes to the product, employee training or organizational behavior.
Predict trends and problems (more applicable to physical products) before they happen.
Problem prevention and problem elimination are the core of customer support, whether it is solved by self-service, the community, an agent or field or auto-support. With smart physical machines and advanced software, the data that comes in from customers has increased considerably. Customers are also smart enough to know the nature of the problem they are facing and expect a quick response. Let’s drill into this to see how computers can help.
In customer support, we often solve customers' problems with tiered approaches and/or swarming. In both cases, humans spend hours through patterns in community forums, FAQ pages, knowledge bases, product patches and release notes, dump and log files and running proprietary scripts to reproduce or gather information from customer scenarios. Essentially, what we are doing is looking at various patterns of previous incidents to see if we have seen a similar situation before. This level of multiple-dimensional data crunching is what computers do best.
But while it is cool to have access to machine learning algorithms that can handle these tasks, computers alone will not solve all issues. Many companies are already looking at how computers can help humans solve hard problems. This process is very different from traditional service management. Businesses can search, using natural language processing and machine learning, for relevant data that can then be sent to agents. This helps make decisions across all three dimensions of problem-solving. Along with providing relevant data, businesses can also use feedback mechanisms, which shed light on what the machine predicts to be the root cause of the problem, as well as possible solutions that have occurred before. This feedback mechanism is the critical handshake between what the human interprets and what the machine predicts.
Even still, customer support automation may not be the right choice for every business, even with its complementary nature. For a business to see some sort of return on investment with this approach, it needs to be ready to invest in the solution. This may include assigning a member of management to oversee the process. Also, if your support team is less than two people — or if you deal with little digitized data — it is hard to see value in this approach overall.
The Future Of Customer Support
We will never know what unknowns the future of computing will bring. Some people are a bit wary of the future, but if we use machines the right way, we will hit the cosmic lottery. If we use them incorrectly, we can expect Skynet to take over. Nevertheless, we are at least a century away from this. Until then, there is time for you to use the complementary aspects of machines to build a vastly better support organization — that is, if the costs are worth it for your business.
Learn more,