Leverage The Power of AI For Happier Customers

Artificial intelligence (AI) is no longer the work of science fiction it used to be. We interact with the invisible hand of AI almost every day — like the curated playlists delivered to you weekly on Spotify, or Gmail’s autocomplete which finishes your sentences before you do.

But for companies today, the most exciting applications of AI lies in customer experience. How do you leverage AI to simplify large, complex amounts of data into something actionable? Or automate the rote, time-consuming activities that would otherwise free your agents up to build those all-important relationships with customers?

In this Handbook, we’ll explore —

The What
The key areas where AI, together with humans, can help companies elevate the customer experience.

The How
Tips from Gladly’s AI and machine learning technical lead, Alice Li, on setting the right foundation for your AI strategy.

AI might still be in its infancy, but it holds tremendous potential to elevate the customer experience even today. We hope this book will be a useful guide in your AI journey.

Of the myriad opportunities that AI offers, its potential to elevate the customer experience and power real differentiation for a brand is perhaps the most exciting and relevant for B2C companies today. We’ve identified three key areas where we believe AI can truly move the needle when it comes to customer experience:

1. Powering Personalization
2. Augmented Intelligence
3. ‘Connected’ Support

Customer experience has become the most important differentiator between one brand and the next. For today’s leading companies, delivering on that great customer experience comes down to making their customers feel recognized and valued; providing them with an experience that feels bespoke and personalized to them.

Offer proactive help.

Sometimes customers may have questions, but be hesitant or unmotivated to ask them. Checkout pages, for example, are notorious for losing customers with last-second nerves about their impending purchase. With an AI solution, you can identify the right moment for an agent to step in and save the day, and automatically trigger a proactive chat session with a hesitant customer. According to statistics, customers who engage in chat are almost four times more likely to buy your product.

Show similar or related items.

While you’re there to help your customers already in your stores, it’s a challenge to provide that one-on-one assistance to your online shoppers. With AI, you can give your customers their own ‘virtual shop assistant’, leveraging it to identify what a customer is currently looking at on your site, and suggest similar or complementary items. If a customer is looking at a lamp, for example, your AI can identify and suggest similar styles of lamps from your inventory, along with the right type of bulb for that piece.

Get specific with marketing.

You’ve likely experienced that moment where, all of a sudden, you see ads for something you’d searched for appear everywhere you go — tucked into the article you’re reading, or right before your music plays on Spotify. Each time a customer goes online, they leave behind a trail of breadcrumbs — from the brands they shop at, to the searches they make on a site. AI has made it easier than ever before to process all of that data, and transform it into a profile of who your customer is: where they shop, what they like, and what they’re looking for.

With that information, companies can move beyond the generic marketing effort, and provide offers that are tailored to a customer’s needs, delivered to the device they’re on, with just the right message to help capture their attention. Just be sure to keep your messages helpful, and steer clear from feeling intrusive.

Tune In

On this special episode of Radically Personal, we sit down with IT and CX leaders from BARK to discuss how their team leverages AI for personalization at scale.

Like the Batmobile is to Batman, augmented intelligence is all about enhancing human effort — not replacing it. It’s especially powerful when it comes to customer service where there’s an immense value in keeping that human-to-human connection between your agents and customers, but which needs to be balanced with the equally important need your customer has to get quick, efficient responses to their questions.

Suggest the best answer.

Over time, a company’s knowledge base can build up to thousands of articles, and become a lot for an agent to have to manually search through. But what if your knowledge base could do the hard work for agents, and offer the most-suited article to an agent at the right time? You can leverage AI to learn from your agents, to identify which articles are most often used to answer questions, and suggest it to an agent when similar questions arise.

Simplified dispositioning.

Dispositioning (the process of labeling a customer conversation according to its issue, like billing or returns) is another time-consuming activity for agents. Take the legwork out of dispositioning by having your AI identify and suggest to an agent the most suitable topic to the issue. AI is getting more accurate and powerful each day, but it’s still not right 100% of the time. The benefit of leveraging augmented intelligence in these cases — where the impact would be either customer-facing or on the important data and metrics in your contact center — is that it leaves a human gatekeeper in charge to ensure the right decision is made at the end of the day.

Automate the manual work.

Every day, managers and supervisors spend precious time sifting through thousands of customer communications, and manually assigning them to agents based on the issue, urgency, or the need for specialist team help. But you can leverage AI to do that tedious, manual routing for you. For example, AI can identify when a customer with a high lifetime value is reaching out, and automatically route them to your specialist team. Or immediately push customers whose flights have been canceled to the front of your phone queue. That way, you not only automate the manual, time-consuming aspect of routing, but ensure that your customers are helped by the agent or team that’s most-suited to their individual needs.

Whether it’s asking Google Home to switch the lights on, or setting a reminder on Alexa to pick the kids up in an hour, home devices are becoming a big part of people’s daily routines. According to a recent report from NPR and Edison Research, there are nearly 120 million smart speakers in homes across the US today.

So considering their popularity and ubiquity in homes today, there’s a truly interesting opportunity to leverage these AI-powered devices in customer service.

Voice-powered customer support.

It seems like a natural step from asking Alexa the time, to checking on your order status. Companies today are leveraging the growing popularity of home devices as another avenue for customer support, allowing customers to do things like find what’s playing at their nearest theater, check on their tire pressure, or even pay their bills. Increasingly, customers want to be able to help themselves before reaching out to a company for help. And home devices offer a quick, easy way for customers to do just that from the comfort of their couches.

Voice-powered shopping.

Home devices are no longer just a place to keep a virtual shopping list — consumers can now use these devices to do the actual ordering for them. From laundry detergent to late-night pizza, home devices are offering consumers a brand new avenue to get their shopping done. Plus, when they’re done shopping, they can turn right back for updates on where their order is, and when it’s expected to arrive. Consumers value simplicity and ease, and home devices are fast becoming a real choice for companies looking to bring their customer experience to the next level.

The task of selecting an AI solution can be a daunting one. We asked Alice Li, who leads the development of AI and machine learning at customer service software company, Gladly, to share a few tips on what to consider before taking the plunge.

AI can be a real game-changer for enterprises, but it’s important to have a sound strategy in place to help you truly maximize the returns on your investment. Here are a few things to think about before, during, and after, you’ve implemented your AI solution.

Before you begin evaluating AI solutions, it’s important to understand and be specific about your goals.

Is it about creating a more personal experience for your customers? Or driving efficiency and productivity amongst your agents? Knowing your goals can help you narrow the list of the many AI solutions in the market, and might even lead you to question if you need a standalone solution at all.

And when thinking about your goals, think beyond just your current needs. What are the 5 or 10-year goals for your business, and how can you incorporate the building blocks to that today, so you’re not doubling efforts down the road?

How effective your AI solution ultimately is depends almost entirely on the data you feed into it.

As any data scientist will tell you, the hardest part of creating AI isn’t in writing the algorithms the best algorithms, in fact, are all published and open source. The difficulty really lies in putting together the data to train those algorithms to ensure they’re as accurate as possible.

Think of bad data like putting diesel in a car instead of gas your car will run, but it’s going to be unreliable, and will do a lot of damage in the long run.

Here are a couple of things to note about gathering data:

Start early and go long. Start collecting data early, and for as long as possible the more data you have, the more accurate your AI or machine learning algorithms become. Certainly at some point there will be diminishing returns, but data storage fees are relatively cheap, and it’s well worth the potential gains in accuracy. Also, collecting data over a long time range means your AI will be better equipped to account for changes due to seasonality. For example, it’ll be able to pick up on the impact of weekends or holidays, the fact that you have more returns after Christmas, or that you can typically expect more flight delays in winter.

Leverage Domain Experts. In an ideal world, you’d have the resources to identify and review every bit of data that goes into training your AI algorithm. But while it may not be fiscally feasible to hire a dedicated team, you can take advantage of your existing resources. At Gladly, we leverage the expertise of a company’s customer service agents to get the training data for our Answers Suggestions — our machine-learning powered feature that identifies the most suitable article from a company’s knowledge base, and offers it up to an agent to use to answer a customer’s question. Our Answers Suggestions monitors the company’s agents to see how they answer specific questions, so that the next time a customer asks a similar question, it can offer up the article that agents use most often, and save them from having to do the search themselves.

This is where the work you did in the first step comes into play again. Schedule regular check-ins to see if you’re achieving your goals. If you’re not, it may be time to adjust or reconsider your approach.

Take the example of a commercial bank that introduced online banking to its customers. While the goal was to make it easier to bank with them, it ended up confusing customers, resulting in more customers reaching out on the phone, or going to the bank, to clarify their questions.

It’s the same with any AI solution you implement. It’s important to ensure it’s working towards your goal, not against it.

While AI can replace human judgment in certain cases, for now, it’s just not at the point where it can be left completely unguarded. Take the example of Knight Capital, a now defunct financial services and trading firm, that lost a total of $440 million to a faulty trading algorithm left to run without any human oversight.

That’s why it’s important to put reasonable constraints on what AI can do automatically, or at the very least introduce thresholds for when a human should be alerted.

For example, if a customer is reaching out about a serious accident or loss in the family, these customers should bypass your automation (or chatbots) and go straight to a human agent who can handle the situation with the delicacy it requires.

There’s a tremendous opportunity to really transform the customer experience with AI, and make it truly effortless for a customer to get the help they need. It all comes down to good planning and careful execution, and being able to recognize where and how to best capitalize on it.