Guru’s Rick Nucci on using automation to scale your customer experience

Some think customers will see the advent of AI as a welcome way to get self-help quickly and get back to their task. Others worry that AI will worsen the customer experience as more and more companies use it to save costs.

For Rick Nucci, customers will win as long as AI empowers employees to do more of what they do best – that is, dealing empathetically with other humans – while skipping tedious busywork that could be handled just as well by a bot. Businesses also win if, as a result of AI, teams are able to create more white-glove moments that turn users into customers for life.

Rick started Guru in 2013, after successfully founding Boomi (a cloud integration and data management company) and selling it to Dell. His new venture focuses on creating a central knowledge network for scaling businesses so that employees aren’t wasting hours by chasing down important resources and company history scattered across a host of apps like Google Docs, SharePoint sites, Evernote and more. If there’s someone who fundamentally understands the power of automation and how it should be balanced by human empathy, it’s Rick.

He joined me for a conversation on tackling issues that come up as you scale your customer experience. Short on time? Here are five quick takeaways:

1. Businesses focused on scaling need to think of customer support as a revenue team, not a cost center. With the advent of the subscription economy, support is increasingly responsible for fostering that relationship over time (which can lead to more long-term revenue).
2. The first steps to transforming your support team into a revenue team are to rethink your metrics and incentivize teams to ask “Why?” Instead of focusing exclusively on how fast you can move through support tickets, dig deeper on the pain points that created those tickets in the first place.
3. When scaling your customer experience, remember AI can’t simulate human empathy. It’s always better to think about how AI can make the humans in your company better at their jobs instead of replacing them.
4. Most customers either know their problem and are looking for a self-help solution (which AI can help provide), or are stumped and need a helping hand from a human. Faking the human interaction with AI can hurt you.
5. Collaboration pains increase as you scale. To solve them, you need a network of knowledge between the experts at a company and the employees who need it to do their jobs efficiently.

If you enjoy our conversation, check out more episodes of our podcast. You can subscribe on iTunes, stream on Spotify or grab the RSS feed in your player of choice. What follows is a lightly edited transcript of the episode.

Kaitlin: Rick, welcome to Inside Intercom. Tell us about your career and how you came to found Guru.

Rick: Guru was born out of a pain I personally lived at my last startup, Boomi, which I started back in 2000. Boomi is a cloud integration company that I worked on and built for 10 years and was acquired by Dell in 2010. I then stayed with Dell for three years, and left to start Guru in 2013. My co-founder Mitch and I lived the pain that Guru now solves: the problem of getting the right knowledge to the right person at the right time. As we both dealt with our own pain and looked at how other businesses tried the same problem, we were amazed at how this problem is unsolved for companies, and we just got really excited about the opportunity.

Kaitlin: As you mentioned, this is your second startup journey. Kudos. What has it been like the second time around? How did you all get to product-market fit?

Rick: I heard a great analogy that when you’re in pre-product-market fit, you can feel like you’re in a desert. When you’re walking in a desert, you don’t actually know if you’re going deeper into the desert, or actually getting out to the oasis – getting out to the water, the promised land. Folks often ask what it’s like, and at least for me, it doesn’t feel any easier the second time around.

“When you’re in pre-product-market fit, you can feel like you’re in a desert. You don’t actually know if you’re going deeper into the desert, or actually getting out to the oasis.”

As were doing our customer discovery for Guru, there were a lot of concepts we were really excited about that weren’t really kind of figured out yet – they were very blue-ocean. Just to give you one tactical example, one of the most common ways people use Guru is through our browser extension. If you go back to 2014, when we were building that part of our product, it was kind of the Wild West back then as far as browser extensions go. They were being used for personal things and for fun things, but there was just a handful of truly workplace-designed extensions.

There aren’t a whole lot of analogies to gain inspiration from; a lot of it is paving your own road, which in hindsight felt like it just took a long time. But once we found that first fit, it felt like things started moving really quickly. For us, that first fit was with growing sales teams. We were very, very focused on that fit – to be even more specific, on growing sales teams for high-tech products. That was essentially the Boomi sales team, so I could personally relate to them. But as we talked to other sales teams that fit that profile, we just heard the same things over and over again. It’s a product person’s dream when you can be in that situation and hear the same pains articulated in pretty similar ways. That allowed us to move quickly.

“It’s a product person’s dream when you can be in that situation and hear the same pains articulated in pretty similar ways. That allowed us to move quickly.”

Every knowledge worker in every company of every size has the same problem. So it’s tempting to try to go wide when you’re building a product and build something for the lowest common denominator. But we decided early on that we owe it to ourselves to have conviction on a very specific persona. We wanted to build for that persona and dedicate our focus to that.

We focused on daily active users as our north star engagement metric. So of our monthly active users, how many are using the product daily? By saying “Before we try to go any further, product wise and market wise, we need to prove to ourselves that Guru will be a product that this persona will build a daily habit around,” it just made finding the next 100 customers so much easier because the product just hit. The likelihood they were going to build a habit and engage was really, really good. That was one of our first aha moments.

Leveraging customer support to drive revenue

Kaitlin: It sounds like you really focused in on sales to start, but Guru today really seems to have expanded to think intentionally about marketing, sales, support and success teams as revenue teams. When did that distinction come about and why is important for you guys?

Rick: That shift came about around 2016. We started to realize two things. One: we saw support teams adopting the product and having a lot of success, and so we leaned into that. That’s a fairly obvious thing. We could see them using it and having success, so we started to dig in and ask: “Why? What is the pain? How is it different from what sales teams are dealing with?” And we realized it’s really all the same pain. In fact, in some ways, support teams have a more intense pain. Customer service agents spend their entire day in customer conversations, and they constantly want to have the confident feeling of being able to provide their customer the right answer quickly. That’s such a core part of the job and what gets them feeling good about their job and their success on the job. They face the same struggles that sales reps face, and it is just as impactful but in a different way. But the way a customer service person spends their day is different than a salesperson. The workflows and the tools they use to do their jobs are different, which matters a lot for Guru and how we design our products.

“The advent of the subscription economy – and the way that people buy and use products – means that the interaction with the salesperson is the start of the relationship with the customer, not the end”

Secondly, we’re seeing a shift in the industry. The advent of the subscription economy – and the way that people buy and use products – means that the interaction with the salesperson is the start of the relationship with the customer, not the end. The actual, most enduring relationship with the customer comes through the customer success and customer service teams. They actually play a big role in revenue. Historically, we’ve always talked about these roles as being costs to the business. But the really exciting part with the businesses that take Guru into their companies is they don’t think of these teams as a cost. They think of them as impacting revenue.

It’s super inspiring when you dig into it, because it’s true. You can see the impact these teams have on customers. Not only do customers stay with you, but they buy more from you. They upsell; they convert from free to a paid plan. All of those things absolutely happened asa result of great customer service. We absolutely owe it to ourselves as an industry to call ourselves revenue teams, because we are. Not in a transactional coin-operated way, but we deserve the same seat at the table as anyone else interacting with a customer. That’s why we like that phrase “revenue team.”

“The actual, most enduring relationship with the customer comes through the customer success and customer service teams”

Kaitlin: As someone leading a customer support team, you’re preaching to the converted here. I’m especially interested in how you think about that support experience. We very much share this common view that support should not be seen as a cost center – but rather with the right people, the right processes, and the right technology and tools, it can actually drive revenue.

What is your advice for leaders who have come to realize that scaling their support team means doing a few things, not the least of which is turning their gaze to how this team might actually drive revenue?

Rick: Great, great question. It starts with changing the way you define your operational metrics as a customer service organization. Look, I completely understand why metrics like “average handle time” need to be part of the conversation. They do. I think it’s important. It’s a way to manage capacity for a growing customer service org. You want to know, on average, how long it takes to resolve issues. But looking at it in isolation can be very, very dangerous because you don’t always want to get the person off the call. You don’t always want to close out the ticket as fast as possible. That’s why I think it can be dangerous to only look at that.

We have customers for example, who will literally teach their customer service teams to intentionally ask open ended questions as calls are wrapping up. The reason they’re doing that is to get it the real ‘why’ behind the ticket. Someone could write in tomorrow and say, “Hey, I need to know how to do this specific thing in your product,” and as a customer service agent, you can be prompted to just say, ‘Oh, you go here, file, drop down, click this, choose that, thanks.’ Then they close the ticket and think, ‘Cool. I closed that one really quick. I came in below average.'”

But ask: Why? Why are they trying to do that? What are they really getting after? I just hear story after story about how asking ‘why’ leads to the real thing:

“We’re really struggling with getting our team to adopt this part of your product.” “Oh, really? Why is that?” “Well, they don’t think that it’s a good way to actually do this or that.”

You could have a totally different conversation than what it seemed like they wanted to do when they wrote into you.

Starting with changing the metrics and how you measure success in that org, that’s step one. It has to be CEO-down. They have to buy into the concept because if you don’t establish that, then the conversation is just going to be about cost and how much we have to spend. What you actually want to hear is, “Hey, this quarter we started asking all of our customer service agents to ask the following three open-ended questions at the end of each call. As a result, we saw CSAT go up,” or “We saw the premium to upsell conversion go from X% to Y%.” Now, you’re starting to connect the team’s performance to more revenue-connected metrics, which I think is a great start in changing the narrative.

Using automation the right way as you scale

Kaitlin: How should we think about using automation to help businesses scale? Because most companies reach a certain point where they know they need to scale but doing it can be hard. Sometimes it can be expensive, and that’s where bad decisions can be made. How do you think about using automation to scale a customer support team, while keeping that customer experience personal and in line with those CSAT scores?

Rick: I think the first and most important thing is to understand what automation can and cannot do for you as a company. It’s very tempting, as a company to think about automating away the customer service experience and turning it into scripted responses and bot conversations. The way I always like to frame the entire technology trend in the workplace is that it is always, always, always better to think about how AI can make the humans in your company better at their jobs, versus thinking about how AI can displace a human doing the same job.

“It is always, always, always better to think about how AI can make the humans in your company better at their jobs, versus thinking about how AI can replace a human from doing their job”

I think that is the first, and most important, distinction in how the entire automation conversation, agenda, strategy should be different. Because if you commit yourself to thinking, “How do I make humans better?”, the way you’ll apply technology and automation will be fundamentally different.

For example, let’s apply this to customer service. If you think about making customer service agents better at their job, you’re going to be thinking about things like, “What do I want my customer service agent doing ideally with a customer?” And, “What’s a waste of their time?” Well, things that waste their time are: tagging tickets, looking up answers to FAQs, and analytical or data-driven intensive tasks, like data entry in general. These are all things that are non-productive or not contributing to those outcome metrics I talked about previously.

If you go the other way, however, then all your conversations are going to go, “Let’s look at the conversations our customer service agents are having with our customers, and let’s think about how to just remove them from that entirely.” I think there’s a role for that. But you need to be really, really careful for two reasons.

“By framing the automation question as ‘How can we make humans better at their jobs?’ we can set the table in a really exciting and compelling way for customer service organizations”

One, going back to my revenue team concept, you have to consider whether you actually want that. If you make customer service agents better at their job, and they can do five more white-glove moments this week than they were able to do last week (meaning that they did something for a customer that made them so excited that they were compelled to tweet about it or share that experience), you’ve got more customers for life. Isn’t that worth more than the couple of dollars you saved by automating it away?

The second thing is the consequence of not being able to do automation well. Some automated experience can take a customer, who might already be emotional or frustrated, to go from zero to 60. And now they’re just mad, because they’re being greeted with something else, when they really just want to talk to a human.

By framing the automation question as “How can we make humans better at their jobs?” we can set the table in a really exciting and compelling way for customer service organizations.

“AI is not at a place where it can simulate human empathy”

Kaitlin: It sounds like the key here is understanding how to keep the customer experience personal in an automated world where you let humans do the personal stuff, and you let bots handle low-empathy situations where that human touch isn’t required. What about chatbots? I’d love to hear your opinion on them.

Rick: I think about them in this way: As someone seeking customer support, I’m going to go down two paths. Path one is that I want to solve the problem myself. I don’t want to talk to anyone. I want a quick answer, and that’s the mode I’m in. As a technology provider and customer service team, how can we help you complete that quest?

Path two is when I don’t believe I will get self-help. For whatever reason, I perceive that my question to be too complicated to be able to look it up or solve it in some way, and therefore I want to talk to a human. So I’m going to start a chat with them or call in. As long as the technology, the workflow and the usage of bots respect that logic tree I just described, it’s wonderful.

Guru is an Intercom customer, and I see the way we’ve deployed Operator, Intercom’s conversational bot technology. Operator is wonderful, because it doesn’t break the chain of that path I described. You can really run into trouble when your customer wants to talk to a human because they don’t believe they’re going to just self-help, but you respond to that path by faking it. By that I mean that you’ve put a bot in front of a person who’s on that path and is trying to get help to their problem. That’s the danger path, because that’s when you need high empathy. There’s something going on. There’s some complexity, perceived or real, where you really need a human involved and doing what humans do best.

The other practical reason behind it, is that AI is not at a place where it can simulate human empathy. It just isn’t. I think that people can fall into the trap of thinking that it is. I love to see bots deployed in situations where you know it’s a bot. It’s very, very clear that’s what it is. It’s augmenting and helping. It’s not getting in your way or blocking what the customer wants do. That’s how we think about chatbots here at Guru. If you follow that logic path, great. You can do amazing things.

Solving collaboration pains as you scale

Kaitlin: Honesty is really the best policy. Going back to issues that come up when you scale customer experience, is the issue of outdated wikis or siloed knowledge bases only a problem for businesses that have hit scale or reached a certain size? When should businesses start thinking about a knowledge management solution?

Rick: The way we think of ourselves, and the way we talk about what we do, is that we create a network of knowledge. Think about the connections between the people who need knowledge to do their jobs, let’s say a customer service agent, and the experts who have the answers, who could be their manager, product managers, product marketers, security teams, engineers, et cetera. We all know who’s an expert, but it gets harder to know that as we scale. That’s problem one.

“That customer-facing team is always going to outnumber the relative few subject-matter experts by orders of magnitude”

Problem two is that it can be so easy and frictionless to talk to folks through chat and messaging, or by walking up to their desk and shoulder-tap them, that you start to see the pain of not having a knowledge system when your teams grow. That customer-facing team is always going to outnumber the relative few subject-matter experts by orders of magnitude.

The reason we call it a “network” is because Guru connects the experts within a company to people who need that expertise. We do it in a way where you don’t have directly ask someone for help or an answer to a question. We’ve found you need something like that when collaboration starts to really become painful. It tends to be as you’re building and growing customer-facing teams. It tends to be as you’re shipping product frequently, because with that product comes a whole raft of new knowledge that needs to be distilled out to the team. Those tend to be the drivers.

Kaitlin: How big is this problem? What’s your estimate for how much time is wasted by people having to hunt down information, internally?

Rick: There’s a famous McKinsey study that gets quoted by a lot of people. It reveals that on average, workers spend over a day per week looking for answers, looking for knowledge. We constantly need knowledge to do our jobs.

Regardless of what job you’re in, that’s roughly the amount of time you’re spending, so you have to consider the consequence of that.

• If you’re an engineer, it means that you’re not doing what you want to be doing, which is writing code and solving technical problems.
• If you’re a customer service agent, it means you’re not spending as much time responding to your customer quickly with that answer that you really want to give them quickly.
• If you’re a salesperson, it means that your prospect is talking to you and probably two of your competitors, and every minute that goes by that you weren’t giving them that answer, they’re potentially getting it from your competitor.

The urgency and pain of that wasted day per week starts to add up and become a tax. The reason why we call it a tax is that it’s so tempting to start to solve knowledge problems departmentally, but what happens when you solve it departmentally is you have a knowledge system for support and a different one for sales and a different one for HR. Every time that tax goes up, the poor customer service agent now has five places to go just to get the answers to do their job that day. All the additional knowledge system is doing is just adding to that waste a day a week. You start with good intentions, but the toll keeps going up and you end up just sprawling and sprawling and sprawling.

A company signed on and became a Guru customer last year, and they had roughly 600 sales AEs across the world. They surveyed them and asked, “What tool or system do you use to get the knowledge you need to do your job?” Then they aggregated how many distinct answers they got across those 600 sales reps. There were 68 different systems that they were using, from notepads to wikis to Google Drive folders to a SharePoint site, to you name it. When the sales leadership and the head of IT saw that, they thought, “Oh my God, we have to put an end to this immediately.” It just illustrates my point that when this stuff propagates out, it comes at the expense of the poor customer-facing employee who just wants to solve their customer’s problem.

Kaitlin: One thing that’s particularly interesting is this concept of using AI to push information to people rather than them having to pull it through search terms. I think that’s a great way of articulating it. I imagine you guys have gotten positive feedback on that.

Rick: Yes, absolutely. Our product vision is, “The knowledge you need to do your job should find you when you need it.” As you said, AI does play a huge role in that, and what we do in these situations is imagine a customer service team at scale, working all day answering questions, using Guru in conjunction with customer chats or tickets. Whatever the interaction with the customer might be, Guru is learning from every moment, and it allows us to proactively suggest knowledge. Literally as you’re having a chat with a customer, Guru is signaling to you, “Hey, I have an answer here that will help you in this conversation, in this moment.” It’s very, very focused on those use cases, so we develop these AI models specific to customer service workflows. Intercom – and how we augment the Intercom conversation – is an example of that.

Then, we have different ones for sales teams and specific use cases. It takes about a month for the system to train and get accurate, and then after that month, we can see the closed loop. 75% of the time, when we give you a suggestion, it’s the right one that users accept. That’s been really exciting and amazing to see, because I think that’s an accuracy level that’s high enough that you’ll trust it. You’ll keep coming back and looking. You can kind of tolerate that we might get it wrong 25% of time, but it’s right so often that you get really excited. We were blown away seeing that. It was sort of a revelation.

What’s next for Guru

Kaitlin: What comes next for the Guru team? I hear you have an upcoming book coming out. I’d love to hear what’s on the horizon.

Rick: We’ve got a couple fun things happening. Just from a product perspective, our statement that “The knowledge you need to do your job should find you when you need it” is an intentionally broad one. Over time, we’ll continue learning about how our customers use Guru today and the tangential lines of additional use cases. Because, again, it’s a network. We’re connecting people together with knowledge. That’s how our customers use us.

As more and more those groups start using Guru, we will deeply want to understand those workflows much like we’ve done for sales and customer service teams. What knowledge do you need to do your job? It’s different if you’re an engineer or an HR person or a recruiter than it is if you’re a customer service agent. We want to understand those workflows, the nuances of how you spend your day and how the knowledge you need can find you. It’s the same problem but a different answer, a different nuance. That’s a good way to think about how Guru will continue to evolve our products and continue to add value to our customers.

And in May we’ll be coming out with a book that talks about the AI opportunity – the notion of how AI can make humans better at their jobs. The book was born out of many, many conversations we’ve had with customers who have said: “Okay, I’m with you. I want to think about how to apply AI to make humans better at their jobs, not automate away departments within my company. Now what? Help me do that. How should I actually start thinking about that?”

The goal of the book is to give you tactical things you can take into your company and help drive those conversations and identify projects and areas where you can apply AI. It’s a distillation of things we learned as a company while we building our own AI technology. It is very hard to do. It is very different than traditional software engineering. How did we learn it? How did we focus? How did we think about building models that would actually work and be accurate? It’s also organizational in that it offers philosophical ways to set conversations.

We’re really excited about it, because again, I think it’s really just reflective of conversations we’ve been having over the last year and a half. It will be out in May online, and then we’ll go to hard copy this fall.

Kaitlin: If that wasn’t enough, it sounds like you guys are also throwing together a conference in your backyard in Philadelphia called Empower 2019. Can you give us a plug on that one?

Rick: The whole idea with this show is to bring together these revenue teams. We are bringing together leaders of any team that is working directly with a customer — Sales, Customer Success, Customer Service, etc. — for two days of tactical, hands-on community development. How can I, as a customer success leader, start to think about changing the conversation to be more revenue driven and cost driven? It really is a great set of speakers from great brands and companies (including Slack and Google and Shopify and Square), so it’s going to be a great two days.

And of course, an added benefit is Philadelphia. As I always say with regards to Philly, “The product is much better than the brand.” And so, I’m also excited for all these folks that come into Philly and check out everything going on in this city, because there’s so much great stuff happening. It’s not just cheese steaks and Rocky and the Liberty Bell. In all seriousness, I think it’s going to be a really informative two days.

Kaitlin: Sounds great. Rick, I think it’s time to say thank you so, so much for your time today. It’s been a blast hearing all about Guru, Empower and how you guys think about automation and scaling revenue teams. Thank you so, so much. Where can people can keep up with you?

Rick: Yeah, absolutely. Our website is simply getguru.com, and my Twitter is just @ricknucci. I appreciate any and all interactions, so thank you for that.