fbpx

AI vs. Humans: What Does the Future of the Call Center Actually Look Like?

AI vs. Humans: What Does the Future of the Call Center Actually Look Like?

Goodbay Technologies

To say that there’s a lot of hype surrounding the rise of artificial intelligence (AI), at this point, might be something of an understatement.

According to a study recently conducted by Stanford University, the total number of active startups operating in the AI space has increased by a massive 1400% since 2000. Adobe estimates that the number of jobs requiring AI has increased by 450% since as recently as 2013. Things have gotten to the point where, according to Monster.com, the three most in-demand skills for new hires today are machine learning, deep learning and natural language processing – all of which tie directly into the larger category of artificial intelligence.

As one would expect, this isn’t just changing the way businesses operate in the present – it’s changing the shape of their futures, too. A significant 83% of businesses say that artificial intelligence is one of the biggest strategic priorities for their business today. 61% of business professionals say that machine learning and AI together are their company’s most significant data initiative for the next year, regardless of the type of business they’re running or even the industry that they’re operating in. Based on all of this, it should come as no surprise that the global value of the AI market will balloon to a massive $190 billion by as soon as 2025, this according to the research firm Markets and Markets.

Of course, all of this demands a fairly important question: what does the rise of artificial intelligence mean for the future of your call center? The answer to that question requires you to keep a few key things in mind.

Artificial Intelligence and Call Centers: The Story So Far

These days, everyone is talking about the myriad of different ways that you can use AI and similar techniques like chat bots to not only answer support tickets, but to also drive down costs across your enterprise. But the fact of the matter is, experience has shown that even with all the advancements in machine learning that are taking place around us every day… AI is still just not very good at these types of tasks.

Sure, it’s absolutely true that bots are fine for the most basic of tasks. They’ve been able to help call centers reduce incoming ticket volumes by between 20 and 30% in many cases by giving customers an easy way to interact with their accounts and make basic adjustments without the intervention of a living, breathing human being. Based on that, it’s safe to say that there is very much a home for bots to drive overall efficiency.

However, that undeniable advantage comes with a fairly big caveat: using bots to manage a customer support experience and using them to create a better one are two entirely different concepts and need to be treated as such.

The more complicated the task you hand over to artificial intelligence, the less likely that task is to get done correctly. At that point, you’re talking about worse than a lateral move – you’re talking about something that harms your long-term efforts at support and retention, not helps them.

The Trouble With Artificial Intelligence in the Call Center

For the purposes of this discussion, the main type of artificial intelligence that you need to concern yourself with as applied to the modern day call center is called the Front End Deflection method. As its name suggests, this is a type of proactive support that usually involves knowledge capture, a chat bot and similar techniques that often enters into the equation prior to the customer ever writing a support ticket in the first place.

The goal of this type of solution is generally to inject some logic and intelligence into the situation as early as possible, in an effort to deflect potential issues up front. In call centers, this is the type of AI that most people know about and have the highest degree of experience with – it’s also the type that is largely responsible for a reduction in call center volumes, as outlined above.

The problem with this is that it’s also the least up to date and least dynamic type of solution out there. The “brains” of the solution are fundamentally restraint to the content that has been manually created by humans and made public on your website and/or support site. If that type of content doesn’t exist, the solution cannot draw from it.

This, of course, is where the problems begin.

The tangible “help” that this type of solution is very basic. If a customer loads an interaction with a chat bot and asks a question, the active assistance that it can provide is limited to “if you have a problem with your bill, go HERE. If you want to find out more information about a charge, go HERE” and things of that nature.

Because the content that solution is drawing from is essentially incomplete, if a customer were to replace words in a ticket, the system is unlikely to pick up on it. The major issue is that nobody actually writes this content with SEO and AI logic in mind – meaning that critical synonyms are often lost. Full and complete answers are often not written in these locations, which limits the overall efficacy of these solutions to somewhere around 15 and 20% – in the absolute best case scenario.

Another issue is that most companies want to be very guarded with this information, as it essentially amounts to intellectual property. The majority of it is set to be used by an internal macro and/or support article, to be given to trained and proficient customer support agents to answer a specific question from a customer. This method is usually preferred instead of putting detail into a larger “playbook” that is available to everyone in the general public.

More Intelligence, More Problems

To get an idea of what this looks like in practice, consider the example of a player for a mobile video game interacting with a customer support service to assist with a problem they’re having. Once they write and support a support ticket, must current customer relationship management platforms will offer automated responses to answer tickets BEFORE an actual human ever has an opportunity to do so.

The problem is that the player is having very specific issues and they’re not necessarily outlining those problems in a way that the AI-powered system can “understand.” Lingo, buzzwords and unique words that are used in the “world” of the game get looked over as unrecognizable – the system has a hard time linking them to the best article to feed to the customer. A human customer support employee who plays the game themselves would pick up on these words – a computer, however, won’t.

Likewise, if the player writes something to the effect of “I played in the weekend tournament promotion a few days ago, but I still haven’t gotten my rewards” – the player didn’t write which tournament, which means that the system has no idea what articles to offer up. A human would absolutely know which particular tournament they were talking about based on the date and the context of the conversation, even if multiple exist within the system. Artificial intelligence won’t be able to recognize this because oftentimes time-related content (think: events, games, specific promotions, etc.) are all handled by the Live Ops teams that do all event-based promotions.

Equally complicating things is the fact that oftentimes support tickets are not as straightforward as artificial intelligence “needs” them to be. Once people start to ask three or four different questions in a single ticket, AI is easily confused. Depending on the specific phrasing and word choices of the player, it might serve up several support links or it might only offer one – thus leaving all other questions unanswered.

Finally, one of the major issues that comes along with most modern day artificial intelligence solutions is that it treats all customers the same – which is a problem, particularly in situations like gaming companies who depend on things like micro transactions as a form of repeat business.

An AI-powered system, for example, essentially cannot “acknowledge” new users. This is a person who problem hasn’t spent any money on the game yet, as they’re likely using virtual currency that was given to them when they originally signed up for their account. They may be a big spender in the next week or two, however – it’s just that they haven’t developed this pattern yet.

Humans can give particular types of responses that cater to the needs of these new users. By treating all customers the same, artificial intelligence cannot. It won’t be able to acknowledge the number of logins that player has had, for example, and it won’t have information about whether or not that person participated in recent tournaments, etc. Non-spenders absolutely have the potential to become spenders if they’re given the appropriate amount of care and attention during these fragile early days of their relationship. If you create a positive experience – if you welcome them to the community in the right way – you’re in a position to help make this happen. While an artificial intelligence response is better than nothing, it does little to actually further that relationship during this pivotal moment in its existence.

The truth is, even most sophisticated clients have maxed out on an accurate artificial intelligence response rate of between 25 and 30%. Average clients hover around 10 to 12%. Neither of those situations is ideal, and neither proves that artificial intelligence is the “magic bullet” we’ve been waiting for.

Most clients dedicate between 12 and 18 months of time, money and effort to get artificial intelligence to work properly – and between 70 and 80% of them end up giving up entirely. At that point, the facts are clear: artificial intelligence isn’t an opportunity cost, it’s a cost of opportunity in a time where you can’t really afford it.

Why a “Best of Both Worlds” Approach is Critical

Even as we move into 2019, customers are still not ready to have artificial intelligence and bots handle EVERY support issue that they have – as the issues outlined above go a long way towards proving. They understand these limitations, too – which is why many of them still insist on human support more often than not.

However, that doesn’t mean they want you to get rid of AI entirely – and you shouldn’t want that, either. Customers like the fast and easy answers to simple questions they can get from a bot. However, they also prefer that human touch to more serious issues. In those situations, they often find automated responses cold and inhuman and, worst of all, unhelpful. This makes them feel like your company doesn’t actually value their business enough to allow them to speak with a living, breathing person – thus harming the customer experience, not helping it.

Again – AI does have a place in the future of your call center, it just might not be the one you originally thought. Gone are the days where you can get by hiring low skilled customer support agents to handle monotonous transactional work. Instead, agent skill levels will only need to be further elevated in the not-too-distant future to handle critical thinking and judgement calls during the fragile post AI/bot interactions.

This is going to increase the cost of these agents, certainly, as they will be in a better position to command a premium hourly rate or salary – with expensive benefits packages needed to attract and retain top talent inflating an already loaded cost.

Luckily, there’s a better way – and it’s a lot more attainable than you might think.

The Goodbay Approach

At Goodbay Technologies, our intelligence customer support team is excellently equipped to handle tickets and calls post IVR or bot interaction. We’re a collection of passionate, critical thinking, empathetic problem solvers who use our best judgment to handle all tickets and other issues that a bot simply cannot.

Our highly skilled agents are perfect for lowering your costs in the ways that you thought AI would, all with the same quality of support you’ve come to expect on a 24/7 coverage basis.

In the End

Ultimately, nobody is saying that artificial intelligence, machine learning and related concepts will not one day become a ubiquitous part of life in a call center. It’s true that significant advancements are being made all the time, and every day things like chat bots get a little bit better than they were the day before.

Having said that, in the present day, they’re still not quite “ready for prime time.” If your company does not care about important metrics like CSAT scores and the quality of the customer experience that you offer on the balance of the 70 to 80% of tasks that cannot be automated, by all means – automate away. At that point, it’s clear that driving down costs are your only priority and to that end, AI will help nicely.

But if you do value a strong and positive customer experience – and make no mistake, you should – you need to think about artificial intelligence differently than your competitors. Remember that the quality of the customer experience you offer and the level of satisfaction you’re able to bring to people every day is the major driver of your competitive advantage in an increasingly crowded marketplace. It’s what allows you to stand out in a crowd and, more importantly, it’s what empowers your customer retention efforts.

Even though technology has come a long way, there are still a significant number of support tickets that absolutely require careful thought and the type of decision making that only human beings can provide.

For the best results, at least for the next few years, most call centers will likely want to experiment with a mixture of human support professionals and AI/machine learning-powered solutions like chat bots. This is particularly critical if customer retention is important to you, and if the lifetime value of a customer is significant to your business.

For the vast majority of all businesses, be they mobile gaming companies, e-commerce brands, or businesses focused on consumer electronics, the future of your organization depends on your ability to build strong, intimate and organic relationships with your customers where they keep coming back for more. If they keep coming back for more, they keep spending more – and at that point, the cost of retention with a high quality support system in place is far lower than the cost to acquire a new customer.

So while it’s true that a bot may save you the aforementioned 20% of new tickets by volume, it’s going to cost you dearly on the other 80% by way of customer attrition.

Instead, you need to look at the current state of artificial intelligence for what it really is: an opportunity. Not one that will allow you to replace your people, but instead to empower them – giving them a chance to work “smarter, not harder” in an appreciable way. With the amount of time they’re saving on that 20% reduction in ticket volumes, they now have more time than ever in a day to devote to the most important thing of all: getting out there and providing that intimate level of support to customers that, at least right now, the machines simply cannot match.

If you’d like to find out more information about the never-ending battle between artificial intelligence and humans in the context of call centers, or if you have any additional questions you’d like to discuss in a bit more detail, please don’t delay – contact Goodbay Technologies today.

Keywords: