Public

An Interview for Change: How AI and Automation Drive ROI

Khoros Alumni (Retired)

ParkerH_1-1662575454048.png

 

Digital customer care can meet and exceed your original ROI projects in several ways. However, poorly implementing a digital Care system can cost hundreds of thousands of dollars annually in wasted time and low NPS scores. Khoros has been a social care leader for years, but a tool is only as valuable as the one who wields it. To get to the bottom of the best ways to use and prove the ROI of Khoros Care, we sat down with Khoros product marketing expert Josh Snider to get an expert opinion on the best and worst ways to use Khoros Care. 

 

Parker Hicks  

What's your background in Product Marketing? And why did you decide to come to Khoros?

 

Josh Snider  

My background is actually in product marketing, but before that, I was in public relations, and before that, I was in embassy educational policy. I was excited to come to Khoros because I was already working as a player in the space and having a lot of fun creating chatbots and conversational commerce. Khoros offered me to work on use-case solutions, and it was something that just fell into place for me. 

 

What has been your experience with the Care platform?

 

My experience has been, most recently, using [the care platform] to create real-world mock-ups for chatbots to send to customers and say, "you guys should get a bot." This is what it would look like within Google's business messages (GBM). In turn, I would build the GBM instance and assign the bot to it. I would pull it up on my phone and simulate myself as the agent within Care, chatting with myself on the phone and recording the whole thing as though I was interacting with a customer bot, so we could help them visualize what it is. The agent UI is excellent.

An example would be the proactive web chat we built. It has the ability to create and manage 10 or 15 different chat widgets. Depending on your website and what kind of knowledge base your site knows, you assign different widgets to it. 

 

Can you share some of the best and worst use cases you've heard of?

 

Sure. This one's pretty straightforward. A number of companies will want to put a bot on public social media channels. My question has always been, why do you want to do that customer? Their answer is something along the lines of, "Because the brand is getting buried in volume." Sadly, a lot of that volume is bull. A lot of the volume is really easy questions. Obviously, it's the sort of stuff that a bot could handle. However, every time a major technology company has launched a bot on a public social channel, it has immediately melted down or been proven to be useless days upon doing it. I like to ask companies what makes you think you will be able to outperform the big name brands that have tried before.

 

What's the worst way you could use a bot on social media?

 

You don't want to let it train itself, which takes away a lot of your opportunity for very intelligent bots. The amount of effort you'd have to put in through supervised training and optimization of the bots would be an immense effort. You're putting yourself at very high risk for a public high-volume display of your company's customer experience or service excellence. You could spend a bunch of time and money, and you'll end up with a problematic bot, or you can spend a bunch of time and money, and it'll still be unimpressive in the general bot and AI field. The internet is very much “Lord of the Flies,” and as a corporation, you're going to struggle with having a public-facing bot in the divisive world of the internet. You're screaming to the Twitter and Facebooks of the world, look at me, I'm a smart bot, I can help with basic things like x, y, and z, but in reality, you're not. Best case scenario, you're going to get embarrassed, and at worst, it could seriously harm your brand. 

 

What's a best use case or real best practice? 

 

The real best practice is when the only thing the bot does is a basic routing feature based on sentiment detection. For example, if an at mention sounds high risk or dangerous, the bot can spit out, "please don't tweet us, call 911; otherwise, visit the support portal here for anything else." When a bot gets an at mentioned under certain conditions, the bot should be spitting out a general welcome message and the same welcome message every time. For example, "thanks for contacting this company's customer support. Hit the button if you want to DM a support agent." Boom, you can hit the button. With a good DMS (direct messaging service), you can do all the various bot actions you want. However, you have to be cautious with how you choose your routing options. In the past, I've had companies come to me with something like $500,000 and ask for the sexiest, most fantastic bot. In their mind, they think that everyone is going to love us for it. Next thing they know, they'll solve customer service entirely, and we will go on vacation for half the year. Sadly, it never works that way. Until natural language programming can spit back out something worth anything, you will still need to use humans for more complex questions and be mindful of your bot use and customer routing. 

 

Let's say you have a customer that comes to you and says, "Hey, we're gonna get rid of all of our humans. We're only gonna use bots." How long until they return to you, say it didn't work and ask for advice?

 

I've never seen anyone substantially large do that. Any business above a specific size knows that that's impossible to do entirely. Generally, a large company's customer engagement and care systems are convoluted. There will be five or six engagement teams on four plans. Times that companies attempt to go bot and A.I. only happen more often in the B2B sales and marketing teams. They will want to throw web chat on their site's enterprise sales and support sides, but nothing in the consumer business. I might need support to turn on the internet for 10 warehouses and 400 locations. They have chat and support messaging channels open, but the volumes are not there. It's because we're all working in an email on our phones and occasionally, in Slack, if they're a little more of a modern business. You're not going to get the same support chat volume that you would with a $70 a month residential customer. Your two businesses are more integrated. You've got people who know each other emailing each other back and forth. You're not coming in through the top of a support flow and having to get routed around. That department will like to look at web chat after a year or two and realize that they don't have any volume here. So they just turn it off. My recommendation to them is to never just turn it off. Just because you're not doing any volume on this entry point doesn't mean you should turn it off. It's better to figure out where people are having those conversations and if you want to hear them or not. 



What's the best real-world use case you've seen?



The brand uses bots, text agents, and messaging channels in Google, Apple, and WhatsApp is an excellent example of using a bot effectively in Care. SMS is cool, too. When you can figure out how to keep it very simple because looking at a bunch of plain text is annoying after a while. When a bot sends you a text wall on a phone trying to solve a problem for you. It's still just as irritating.

 My personal favorite use case is automated scheduling. When you're trying to book a haircut appointment, for example. You can schedule your appointment through a bot that gives you a time picker that can immediately add an event to your calendar and reserve that time. It's so simple that if you ever have to go back and ask, wait when my appointment with the mortgage reviewer or home appraiser or whoever is? You can go to that conversation in your messaging center on iOS or Android and text that bot or escalate to a human if you have any questions or need support. It's pretty darn simple and impressive when it's done correctly. 

 

One of the most significant customer pain points that keep coming up is that customers are struggling with too many clicks to handle the conversation. What would be your advice on how to deal with that problem?



Let me start with Khoros. Our modes of engagement are primitive but not in an unsophisticated way. You'll have to consider how you engage one-to-one customers or one-to-many at a high level. Take time to review a lot of different chats. From there, see how they are built, and you add on a bunch of power or action items. You definitely add complexity that can make it harder for an agent, right? If you want to add routing and tagging, okay, those are additional clicks. An agent needs to say, "It looks like they're a service person. I'll click tag as service route to service." Those processes don't necessarily need to exist. Companies like ours can solve them with elegant UI, enough automation, and AI. We've been advancing that science. To the customers saying this, just know that we hear you, and we're on the same team here. For what customers need to do, you have to have AI and AI-powered automation specifically. If agents have too many clicks, it's likely because there are things that they have to decide or do that a bot or a rule or some other system could easily do for them.

A simple solution would be to let bots tag routing and prioritize whatever time or click-heavy tasks you're agents have to do. Take clicks off the agent's plate. Now sometimes, doing this can be expensive or complicated. You have to remember that's the nature of the beast. For a real-world example, I think there was a telco company with 100 Facebook Messenger agents. They saved roughly a million dollars through intelligent automation and routing. I think it was a 13% efficiency improvement, on average, after like two months of it. And over the course of a year, it was something close to $980,000 in labor savings just by using our Push Next automation system. I think it's important to ask yourself if it's worth 100k to save that million.

 

 

 

At the end of the day, the Khoros Care platform is immensely powerful and can provide well over 100% ROI. However, as Josh Snider put it you’ll have to know how to use the platform and how to use automation to best suit your business needs and in turn, find record-breaking success in customer care. 

 

For additional resources on the Khoros Care Platform, please visit these resources below! 

1 Comment