Hello, all. This is Ophir writing with the introduction pieces and then Tiernan with a fantastic overview and interview with the CEO.

Lede

We sat down with the CEO of Top Pick Five9 (FIVN) and what we experienced was a fantastic window into a CEO and his vision. This was a truly exceptional conversation.

Preface
Call centers are a massive industry spread across the globe. The vast (vast) majority are not cloud based, and most are nothing but telephony. This leaves customers unhappy, and costs spiraling out of control with scale nearly impossible without yet more costs.

Along comes a new player with a new idea — turn the customer service portion of the business into a technology hub — insert artificial intelligence (AI), go beyond telephony, add data analytics, put it in the cloud, and turn it all into a profit center. Make customers happier, faster — make corporations more profitable. Make scale easier, faster, and less expensive.

This is the thematic shift coming to the call center and Five9 (FIVN) is leading the charge. Revenue is booming as operating leverage kicks in to show stronger profitability measures. The market is huge – it’s changing, and we believe Five9 will be the winner.

FIVN has an entire ecosystem for customer service that spans email, chat, voice, mobile, and social while putting an artificial intelligence layer on top of it to make it scale. The company already has monster partnerships with Salesforce.com, Oracle, Microsoft, zendesk, servicenow, and many others on the CRM side.

Equally, it has technology partnerships with Google, IBM, Microsoft, Zoom, and many others.

Further, 92% of the company’s revenue is recurring, as in, a subscription, and no single client makes up more than 5% of total revenue.

 

All of this growth is driven by enterprise. The company has seen its enterprise customers double from 2016 to 2018 and according to their financial disclosures, for every $1 in acquisition cost, the company generates $6 in cumulative profit over the next five years to follow.

Key Thematic Metrics
No company makes it onto the Tip Picks list without an enormous thematic tailwind that drives the business forward. Here are some of the thematics for Five9.

* By 2022, 72% of customer interactions will involve an emerging technology such as machine-learning applications, chatbots or mobile messaging, up from 11% in 2017. Yes, that is nearly 7-fold in six years.

* But, though the proportion of phone-based communication will drop from 41% to 12% of overall customer service interactions, a human agent will still be involved in 44% of all interactions. Remember this point — a human is still involved, it’s just less likely it will be over the phone.

* Further, by 2022, two-thirds of all customer experience projects will make use of IT, up from 50 percent in 2017.

* The customer service realm itself is expanding. By 2020, more than 40 percent of all data analytics projects will relate to an aspect of customer experience.

* In what would stun most people to learn, which is why we’re talking about it early, by 2019, 20 percent of brands will abandon their mobile apps as they have failed to deliver the level of adoption and customer engagement they expected.

* Artificial intelligence is coming into the product thanks to the involvement of chief technologist Jonathan Rosenberg, the inventor of the “SIP” protocol for the Internet, a significant technical accomplishment of the last 20 years.

For a fuller view of this company we highly encourage a read of the most recent Top Pick dossier:
Five9 Shines on Every Metric (paywall)

Highlights
The interview was wonderful but lengthy, so here are some highlights from the transcript to help you focus your research. If the highlight has a quotation, it comes from Five9’s CEO Rowan Trollope.

* Five9’s software and SaaS (Software as a Service) side are going up. “The company is already a SaaS company with interesting software, but we are making it even more interesting. With agent assistance, and with an AI and data play.”

* The is are clear ROI to automate a part of the customer service process, for Five9. “The potential is huge.”

* Five9’s newest expansion in its product line is the integration of a human assistant with a virtual assistant using artificial intelligence to listen to the call, translating speech to text, listening to both the operator, the agent, and the caller, and then using NLP [natural language processing] to extract information from what the customer says, and then to make predictions.

“This technology lets humans be awesome and lets computers take over what they can do well, to drive way more efficiency in the call center space. We can leverage the best of technology and humans together.”

* “Five9 has been on a really solid growth track, and now we are adding a very significant new potential market opportunity to our addressable marketing. It’s very exciting to all of us, and very significant for us.”

* “About 15% of the calls will be automated in the next ten years. I think that’s maybe an understatement. Even if it is an understatement, that’s $38 billion dollars’ worth of agent labor, is what 15% is. We spend $210 billion on support agents per year. It’s a substantial amount of savings, potentially.”

* “We are a company that is focused on innovation and there is a large amount of innovation to be done in customer service.”

* “There is lots of money spent by businesses, $234 billion per year in customer service to deliver an experience that almost no one likes.”

* “Humans are awesome, they can handle all the boundary, edge cases. My attitude is the same in customer support. Yes, we should make products that never have problems, and self-service, but that’s not a case for not having good customer service. I fundamentally believe in humans.”

* There is a generational change coming: “digital natives have much more of a preference for service, they say it is as important as product.

When they have a great service experience, they end up being even more loyal than they were before. Even when your product didn’t work as advertised, you can drive a higher Net Promoter score by having a higher service experience. Humans can accomplish miracles.”

Finally, we turn to Tiernan’s interview with Five9 CEO Rowan Trollope.

One-on-One CEO Interview and Tiernan’s Preview

Shares of unified communications mavens Five9 (NASDAQ:FIVN) notched another substantial leg up in an already impressive year of performance on November 6th, rising almost 9% after the company the night before comfortably topped revenue and earnings expectations for its Q3 report. That was followed by a well-received analyst day event on November 12th that further drove the price. Five9 is now up over 50% for the year at a recent price of $66.29.

If you’ve never heard chief executive Rowan Trollope speak, you’ve missed perhaps the epic poet of call centers. When Trollope sat down to talk to Capital Market Labs, the day after that analyst briefing, his responses to questions were initially succinct. But when he started to get into the details of the company’s approach to artificial intelligence, Trollope, who was at one time a call center operator, and a call center manager, waxed ecstatic about the role of humans solving customer service issues. 

He went on for paragraphs, talking extemporaneously like a bard in a beer hall regaling the pub folk with tales of faraway places. All of it is captured in the transcript that follows. 

The fellow is seriously passionate to the point of obsession about what his company does, it seems. 

Artificial intelligence is coming into the product thanks to the involvement of chief technologist Jonathan Rosenberg, the inventor of the “SIP” protocol for the Internet, a significant technical accomplishment of the last 20 years. 

Rosenberg gave a demo at the analyst day of how machine learning can help a live call agent, and Trollope tells CML, “The potential is huge” for the technology. “There’s clear ROI to automate this process, for us.”

More important to Trollope is that AI is a way, as he sees it, to aid live human beings on support calls to be their best selves. “This technology lets humans be awesome, and lets computers take over what they can do well, to drive way more efficiency in the call center space,” he said. Citing remarks by Tesla chief Elon Musk regarding how people handle “edge cases” in processes better than machines, Trollope maintains that “humans can accomplish miracles,” at least, from a business standpoint. 

“I fundamentally believe in humans” is his rallying cry. 

To recap the results and outlook, Five9’s Q3 report included revenue of $83.77 million and earnings per share of 20 cents, topping consensus for $78.7 million and 15 cents per share. 

For the current quarter, the company sees revenue of $86 million to $87 million, and EPS of 21 cents to 23 cents, higher than the average estimate for $83 million and 21 cents per share.

The following is the transcription of our conversation.

Capital Market Labs: At the end of the analyst day, there was a bunch of discussion of this outlook for getting to 27% Ebitda margin from 18% today, over the next five years. That was not a change but a re-affirmation, correct?

Rowan Trollope: Yes. People wanted us to re-affirm that and we did. People said you are basically almost there on some of these areas. Barry [Zwarenstein, Five9’s CFO] said there is a relatively mechanical line of site to there.

Today, gross margin is in the 60s [percentage of revenue], and we will be driving the gross margin, increasing it every year because of the mix shift from usage revenue to subscription revenue, because subscription margins are higher. And so, you don’t have to have a lot of leaps to get there.

We don’t break down the difference [between usage and subscription margins], but they’re [subscription] higher — usage margins are in the 50s, which is good, but it’s nearly no R&D expense there.

CML: Barry was asked about R&D, about whether the current level of spend is correct for you with a lot of new products coming forward, in artificial intelligence applications, for example. Whether 8% to 10% is the right percentage of sales at which to have R&D spending. Do you have anything to add on that point?

RT: My answer to that is that, you shouldn’t look too much at individual line items because those will move around, there will be puts and takes.

What we are committed to is the 27% EBIDTA goal. You could see us hitting that while still spending more on R&D. And then, the other point is, our R&D as a percentage of total revenue, if you back-out the usage revenue, which is not costing R&D, we are actually in the teens as a percentage of subscription revenue, so, it’s bigger than it looks. Usage revenue is about a third of total revenue.

CML: And so, you have the right level, because all these new product initiatives…

RT: Yes, definitely, I understand the question. The question is one that makes sense. Look at our R&D as a percentage of subscription. If we shift that in the model, it won’t be coming out of the 27%, it will be other line items, we might see R&D go up and see gross margins go up at the same time. If we were to do that it would be slowly over time.

CML: As for these products, you’re sort-of steadily moving beyond selling minutes—

RT: I wouldn’t characterize it as we are selling minutes. We are a SaaS company today. Minutes is really, as a percentage of the total, is going down. The software and SaaS side are going up. The company is already a SaaS company with interesting software, but we are making it even more interesting. With agent assistance, and with an AI and data play.

The capability that we have today, that almost no one has today, is because we have the SaaS software. We are recording those voice calls, and that is dark data today, that is not being used. And we can use it to do AI.

We got great questions from the analysts, and we did a demo of the product. That’s the exciting new development, I would say, over the last year, the recognition of the ability to do this, that the data actually works, and that it works at scale, and works in real time.

And when we took all that as an idea to our customers, they all said, Yes! We took it to our top twelve customers, and more than half of them wanted to participate in the alpha version.

McKesson [one of the participants during the analyst day presentation] is just the one we had on stage; there were several others. There’s clear ROI to automate this process, for us. The potential is huge. We have 160 people doing quality management on this product alpha.

CML: So, explain, exactly how does this AI stuff work with your products? From the demonstration yesterday, it sounds like the system is assisting the human operator, is that the idea? It’s helping by retrieving information that the human operator would usually have to look up manually?

RT: Yes, that’s totally it, you got it. Now, the nuance is, if you describe it that way, one’s conclusion could be that you are just pulling up a customer record, you know, just looking up the fact that the customer, Tiernan, is taking Advil, or that his DVR just rebooted more than ten minutes ago. But it’s more than that.

To eliminate any doubt about why this is so interesting, it is a human assistant, and there’s a virtual assistant listening to the call, translating speech to text, listening to both the operator, the agent, and the caller, and then using NLP [natural language processing] to extract information from what the customer says, and then to make predictions.

Say I’m the customer, I call in, and I say, “The Web site isn’t working.” When people have forgotten their passwords, they may most often say, your Web site is not working. They are convinced they know the password.

So, the computer prediction is, when they say the Web site is not working, ask if they have changed their password recently. The system will go and check automatically if they have changed it recently, and if they had, and they are saying it’s not working, the most likely recommendation is for the human agent to ask the customer, “We noticed you’ve changed your password recently, are you entering the new password?” Most times, they will say that’s exactly right. That’s the prediction pattern recognition.

The reason that’s interesting is that humans could do that too, but it takes a lot of training. It’s not obvious to an agent, it’s not intuitive. And if the agent has to look up the customer record, and click on a button, that unique information would take them time to go get.

What the computer does really well is, it does all that in a flash. It’s not only predicting; it’s automating those other steps like the last password change date. And it’s much more complicated when you get into drug interactions or mortgage applications and all the complexity there. Most [human] desktop agents have 10 or 20 desktops open at a time. They are looking at what you are calling about, and putting that into one system, getting a result, reading that, putting that result into another system.

So, this is a bit of prediction, workflow, a bit of AI. You put all that together, it’s maybe what used to be called an expert system but operating in real time. We call it a genius, someone listening into the call.

CML: It’s like a coach on the call?

RT: Right. The best real-world example of this is, I used to be in support, I took 8,000 phone calls, that was the total that I did. After 8,000 calls, you get really bored because most of them are the same. It’s no longer challenging for a human, it’s no longer interesting. That took me eight months, I became the supervisor.

I was 18 years old. I had nine or eight reps I was managing. I asked, “What do I need to do to be successful as a leader?” You need to make sure your team has lots of calls, and satisfied ones, at least twenty-four calls per day per agent. So, I had to figure out, “How can I get them to take at least twenty-four per day?” The answer is they have to be more efficient.

So, I sat down with Val [one of the call agents], I plug in my headset, and my brain is trained now: of 8,000 calls I took, there were probably 100 common calls. I know these calls, I’m good at predicting things. I plug in the headset, she’s listening.

The first call she takes, it’s something I’d taken dozens of times before. It was something like the password change example, the Web site is not working. I did a quick prediction in my mind: this person is calling because they probably recently changed their password.

What Val did was, she said, I’m so sorry, we’ll get right to the bottom of this, all kinds of filler words, meanwhile, pulling up her training manual, and looking at it, looking for the case where the Web site is not working. And she was pretty bewildered.

I’m almost jumping out of my seat; I know how to end this call very quickly! I put mute on to her, and I said, “Val, this is most likely what’s going on: they probably changed their password recently, just ask them, real quick, Let me just ask you a quick question, is there any possibility you changed your password?” The customer said yes, and I short-circuited the call. What would have been ten minutes, took 2 minutes.

Now, I couldn’t replicate my brain to all of my ten agents. If I sat with her, I could have taught her like an apprenticeship, but you couldn’t scale that. That’s what this tech does. It literally scales the learning, and allows it to replicate like an expert agent that’s heard it all before. They [computers] don’t know everything, and they’re not good at empathy, but they are really good at telling you what to say.

You couldn’t hand this all off to a computer. The human still needs to be in the loop. But you prepare them with the answer the computer has learned, and the clicking around they would have to do for these common calls.

About 15% of the calls will be automated in the next ten years. I think that’s maybe an understatement. Even if it is an understatement, that’s $38 billion dollars’ worth of agent labor, is what 15% is.

We spend $210 billion on support agents per year. It’s a substantial amount of savings, potentially. When we talk to our customers, they all think that’s an understatement. Common calls are more than 15%.

Where we are now is, let’s go through this on a few use cases. And the beauty here is, the contact centers have been doing this for a while. This concept of automating has been going on for a long time, but the rub is, the way it’s been done is not very customer-friendly.

It’s typically through an IVR [interactive voice response system]. It’s not a very pleasant experience, something like, “If you’re having a problem with the Web site, push 1!” And people get frustrated in automated workflows. When you use the agent but you pre-wire all this knowledge, in a way that you give them superpowers, you still get the empathy and the human element.

If you see the demo, we did at the analyst day, of the user interface, what you see is that it actually is a fairly big component of what we are building, this coaching piece.

The user experience for the agent is important, to get that right. It has to be accurate and helpful. We spend a lot of time on the user experience, so it will be present but not obtrusive. The guidance is easy for them to use. These are agents who spend lots of time on calls, we want to make sure that experience is a productive one.

CML: How does this change the business for you, this AI capability?

RT: We are a company that is focused on innovation and there is a large amount of innovation to be done in customer service.

If I back way out, I am a purpose-driven person, and our company is purpose-driven. I started the analyst day with why we are doing this. The “why,” here, is to make customer service a more human experience. We in general have been forced to accept an experience that no one likes. Most calls aren’t particularly pleasant. There is lots of money spent by businesses, $234 billion per year in customer service to deliver an experience that almost no one likes.

Just use your own experience: most people dread having to contact a business. You know you will have to go through the labyrinth of an IVR. There’re all kinds of cases. Your food is not here, for a food delivery service, where is it? You end up in an IVR, you are waiting, you are upset.

The way that people have thought about this for a long time, and almost the accepted answer here, is, we shouldn’t have humans doing this anymore. Information should be available on line to get answers you want instantaneously. We shouldn’t have humans in this. That’s like, hospitals are a bad experience, so let’s just make it so people don’t get sick, let’s focus on wellness. And I agree, and wellness is the equivalent of making a better product.

But we also know that shit happens, and when you need care it should be awesome. The solution to really bad healthcare is not, like, let’s make everybody well, but also solve that problem, too. I want to, if I’m waiting for my food, and see it on an online map, if something goes wrong, like they are walking around the block, over and over again, I want to call someone. Why is this taking so long?

I think it’s something Elon Musk discovered. He said he tried to over-automate his factories. The boundary cases in the real world, the edge cases, are too complex, and too numerous for computers to solve them all today. So, he put humans back in there.

Humans are awesome, they can handle all the boundary, edge cases. My attitude is the same in customer support. Yes, we should make products that never have problems, and self-service, but that’s not a case for not having good customer service. I fundamentally believe in humans.

I’ll give you another example. There’s this coffee shop near my office, it’s got a robot arm that pulls the coffee and grinds it, probably better than a barista. But there is also a person standing there, and the person is essentially superfluous, and they greeted me, “Are you having a good day?”

That human touch made me come back. The amazing coffee is nice, but the human touch, I wanted to go back and engage with that person because they gave me a nice feeling. You can’t underestimate that.

That’s not to say we should have people there in the cable company for people to have someone nice to talk to, but when things go wrong, people make a difference. In the process of automating what the people do, I want that to be awesome as well. It doesn’t make us satisfied when we dial in to an IVR, it doesn’t make us feel they care about us.

This technology lets humans be awesome and lets computers take over what they can do well, to drive way more efficiency in the call center space. We can leverage the best of technology and humans together.

This is the longer-term vision, but most people in this space want to talk about, “Health care sucks, so let’s not get sick.” I want to solve, “Why does health care suck, let’s go solve it now.” Let’s make this be one of those things like rotary phones or some figment of the past.

In ten years, imagine a world where the idea of IVR and the cable company and the horrible experience, where probably seven billion people on earth, not one could relate to contacting customer experience as a painful experience. Let’s make that zero or almost none in ten years, so my daughter is, like, “What do you mean?” I say, it was called an IVR, and she just looks at me, and doesn’t know what we are talking about.

This is with the new generation, with the coming into the world, Gen Z, their preference for experience is much higher than Boomers and Gen X. Basically, digital natives have much more of a preference for service, they say it is as important as product.

When they have a great service experience, they end up being even more loyal than they were before. Even when your product didn’t work as advertised, you can drive a higher Net Promoter score by having a higher service experience. Humans can accomplish miracles.

So, part one of my answer was the vision, and being a purpose-driven company. We’ve been hiring leaders who got this big vision. We have been hiring an awesome leadership team, so the talent is phenomenal.

David Pickering [Five9’s executive Vice President for engineering] from intuit, and Jonathan [Rosenberg, chief technology officer], these really phenomenal leaders.

The second thing is that I’ve been building my bench. Now we’ve got these new leaders in place, and it’s off to the races, in alpha with customers [for the AI capabilities]. Those 2 things are really important, and business continues to do really well.

We are very consistently executing, and we’re going to continue to do that. Five9 has been on a really solid growth track, and now we are adding a very significant new potential market opportunity to our addressable marketing. It’s very exciting to all of us, and very significant for us.

CML: Well, thanks for clarifying all that, Rowan! Keep up the good work!