“It wasn’t a walk in the park. Today, Gong is a super hot company. But at that time, we got a lot of no’s, by not stupid people. There were a lot of objections, like salespeople are going to hate it as a big brother, and Google and Amazon will compete with you“, says Amit Bendov, the CEO of Gong.
From those early days of facing skepticism, Gong has indeed become a hot startup loved by customers and ushering its own category, revenue intelligence. It’s also had tremendous fundraising success with VCs, raising $305M in less than 18 months, including a $200M round on a $2.2B valuation, announced in August 2020.
We were thrilled to welcome back Amit at Data Driven NYC, where he had spoken a few years ago, when he was CEO of SiSense.
As always, Data Driven NYC is a team effort – many thanks to Jack Cohen for co-organizing, Diego Guttierez for the video work and to Karissa Domondon for the transcript!
FULL TRANSCRIPT (lightly edited for brevity and clarity)
[Matt Turck] Amit, I was going to say welcome, but in fact, welcome back, since you were a speaker at Data Driven NYC back in the March of 2014, and interestingly considering our other speaker today is Ashley from Sisense – at the time you were CEO of Sisense. So, six years, it’s incredible, but very excited to have you back here today. So welcome. Maybe just to jump right off, I’d love to maybe start with a quick overview of your background and your journey to founding Gong.
[Amit Bendov] Sure. So first, excited to be here. This is personally my favorite city in the US. Don’t tell the people in San Francisco, but it is. And especially, Gong, actually, the idea for Gong was born in New York. Gong is the fourth company that I’m leading. I was one of the founding team of a company called ClickSoftware that was acquired by Salesforce for $1.3 billion, and CMO and VP of sales at a company called Panaya in the ERP space, smaller outcome, quarter of a billion. I was, for a period between 2012 and 2015 or ’14, around that time, was a CEO of Sisense. Gong is the first company that I’ve actually founded from scratch, always joined pre-revenue, but this is the first one that I actually founded from scratch. So that’s my addiction, building companies.
Maybe just to start, from the very top, so what is Gong, and then we’ll talk about the idea and the why now, but what is Gong?
Yeah. So Gong is a revenue intelligence system. Now that doesn’t tell you anything, probably the majority of the world still doesn’t know what the heck it is, but it’s a new kind of system that is used by a customer facing organization in sales and customer support, to provide a layer of inside information that is 100 times better and bigger than what they’re getting from their CRM system.
How is it even possible? The information in a CRM system is based on what the reps are punching in. They’re filling in forms, how was the deal, how was the conversation, just very little and obviously subjective. Gong captures information at the source. So rather than relying on that, it just monitors conversations with customers. It could be emails, could be Zoom calls, phone calls, documents. It uses natural language understanding. This is a data forum so you understand, essentially it takes a huge amount of unstructured data, which is what people say, and extract structured data that provides the insights, both for the customer facing people to be better at their job, it gives the leadership team insights, what’s really going on, and how to get better, and all of that without anybody having to do anything. So think like a self-organizing CRM.
Every salesperson’s dreams, of how you don’t have to do the data entry, and sales managers don’t have to force you to do it or threaten you.
So what was the why now moment? Back in 2015, as you were thinking about, okay, I’m going to start a company, why that specifically at that moment?
I wasn’t even thinking about starting a company. So, when I was at Sisense overall we were growing pretty well, but then we had a quarter from hell. Or I don’t know if it’s okay to say it on these webinars, an “oh **** quarter”. Everything’s going up, up, up, up, up, and then one hell of a nose dive. And I had no idea what’s going on. Marketing is pointing fingers in sales, sales is pointing fingers in marketing. Everybody’s blaming the product. And I’m the Switzerland in the middle, I’m trying to understand what’s going on.
So in my desperation, I had my operations people start to crunch the metrics. Are we getting conversion rate, do we have enough leads, are we making enough conversations? Everything looked fine, which obviously didn’t make sense. And I had my product managers actually sift through Salesforce notes, and they’ll dump everything in Excel, and they’ll start reading, go over thousands of deals. They’ll spend a couple of weeks – there’s nothing there. We don’t know what’s going on.
And then it dawned on me that, this is pretty crazy, there are some fundamental questions that we’re unable to answer. Why are some salespeople successful and others are not? Is it a bad hire? Is it our training? Is it something like pure chance? Why are we not winning more deals? So Sisense was and still in the business intelligence space, a highly competitive space. There are a lot of players, buyers are confused, and you never really know why you’re losing, even how much you’re losing to the competition or why. Because the only way for companies to know how much they’re losing is if their reps select that drop down list, who we lost a deal to. And which often, they don’t do, and sometimes it’s more than one.
So then it dawned on me that, “Hey, the king isn’t wearing any clothes. This is ridiculous. These are some fundamental business questions that we can’t answer.” So I was looking for something that can give me a better picture? I was asking, actually, the reps to record calls and send me recordings. At that time, I was commuting between New York and Tel Aviv every week, and on the flight, and instead of watching movies, I would just listen to these because I need to get to the bottom of it. But even that was super time-consuming. And I said, that’s anecdotal. I can listen to five calls, but there are thousands. How do I even know that I got the right five.
So I started Googling for something that can give me automated insights from conversations, but couldn’t find it. And I said, “Hey, maybe there is an opportunity.”
Then what did you do next? I always find those entrepreneurial paths to be fascinating. How did you get started, building a prototype and all the things. How long did that take? How much sort of technical product building was there to do before you had an MVP?
It worked pretty fast. I wouldn’t say it was super smooth, but so first we did two things. First, I found a great co-funder, Eilon, who is my CTO. When I was at Sisense I was interviewing for a VP of engineering. And as soon as I left, I told him, “Hey, I’ve got this little start-up for you.” So we are meeting at a coffee shop once a week and trying to come think about the idea.
We did two things. First, we did the market validation. Often a common mistake for founders, if they think that something is a great idea that they would love to buy, they think that the entire world is willing to, and that’s not the case. So we interviewed about 50 potential buyers and asked them, “Hey, we’re looking at something that’ll automatically glean insights from conversations. What do you think? Will you buy? How much will you be willing to pay?” And the feedback was pretty positive across the board.
Those buyers were VP of sales types, or VP of sales ops?
Yeah, both. At that time, we didn’t know the difference. So we did a pretty broad spectrum. And again, this doesn’t tell you, because again, it’s my fourth company. I know that they can say yes. When it comes to actually buying, they may still say, no. But if everybody said, “No, that’s a stupid idea,” then probably, we would have abandoned the idea.
The other things we try to understand are the state of the art of the technology that’s available, especially in natural language understanding, speech, video. We didn’t want to spend three years in development. We want to know what’s available right now, how good is it, and can we deliver something today?
So fortunately, we started in 2015, where both deep learning and natural language understanding just made a leap that allowed us to be successful. I think if we started a company two years before, I’m not sure that we would have been successful. So with that, we said, “Okay, that’s an amazing idea. Let’s go and raise money.” It wasn’t a walk in the park. Today, Gong is a super hot company. But at that time, we got a lot of no’s, by not stupid people. There were a lot of objections, like salespeople are going to hate it as a big brother, and Google and Amazon will compete with you, and standard stuff. I think fortunately, because of my track record, people saw that, maybe we’re not convinced about the horse, but the jockey is okay. So we raised a $6 million seed. That was at the end of October 2015.
In 2016, we weren’t planning on selling anything. We just wanted companies to start playing with the product. So we’ll start to train the model, and we get some recordings that we can start testing the models. We found a dozen friends and family, Sisense was one of them. We backed them, and they did us a favor. Things started to get better pretty quickly, which was a surprise to us. So around February or March, we saw that people were actually using the product more than what we thought. And we thought, okay, let’s invest it in a few tweaks in the user experience and move some buttons, nothing major, and engagement started to go up. Then around April or May that year, only four months from a pre-alpha, I was talking to Eilon, he says, “You know what? Let’s do a trial close.” A trial close is a technique in sales, you just go and ask for the order and see what happens. And we wanted to ask for something that was a non-trivial amount, not outrageous, but not $500.
And again, we just wanted to know where we stand. Do we have something that people find valuable enough that they’ll pay us? So we reached out to everybody and said, “Hey guys, sorry, payday’s over. It’s time to pay.” And it was in the tens of thousands. And everybody balked, but 11 out of the 12 bought. And the 12th bought a year later for a lot more. But it says, “Okay, now we have something.” So it’s almost a product market fit on day one. Which, in my view, was better than expectation.
And you found that the engagement was strong. How did you navigate the social dynamics of precisely the salespeople feeling they may be listened to by some bot? Did you sell more to the decision-makers or to the salespeople? How did you do that?
We did both. So that was, the people that had that concern were pretty smart. It’s legitimate. So we built actually enough in it for the users themselves that they’ll be able to at least go through the past hurdle. And what we’ve seen consistently is that the first day or two may feel uncomfortable, because now you’re in this big open space with everybody. But actually that creates a very positive transformation within the company. That people say, “Okay, people now can help me. They could watch my game. I can watch their games.” So just like LeBron watches game tape every day.
It’s working for him. So it is, plus now it makes their life a lot easier because they don’t need the yellow pads to take all the detailed notes and then put them into Salesforce. They could just focus on the conversation. So they love Gong. Our net promoter score as of today is 80, eight, zero. Higher than the iPhone in 2008. Most of those ratings are about, it’s insane for enterprise software. So most of our ratings are by the salespeople that absolutely love the product and can’t imagine their lives without it. But we had to tweak the product to make it work for them.
Amazing. And initially, did you go after startups initially as a target market?
Are you still there? And presumably you’ve gone up market to the enterprise.
Yeah, initially it was startup companies. To us it seemed like big companies at the time, we’re just five people at the company. But now definitely we have accounts in seven figures, we have very large companies using us. It’s an important motion for us. That actually came a lot sooner than we’ve anticipated. I still remember my initial seed pitch that we build a timeline, like down the road. That was in five years we’re going to start going after the enterprise. But it actually almost on the second year we were starting to see very large companies coming to us.
Okay. Really interesting. Maybe let’s talk about the product a bit more, and I go into how that works. So presumably there is an ingestion layer of some sorts where you get data. So you started with voice, but you cover other sources as well, right?
Right, right. Yeah.
How does ingestion work?
So the product, yeah, there is an ingestion layer. So they’re probably right now about 100 systems that we support. And then every couple weeks we’re adding more. Tomorrow, by the way, we have our celebrate event. Shameless plug. You’re welcome.
This is good. When is it?
Tomorrow, it’s an all day, start morning San Francisco time.
Anybody can go to Gong.io and register?
Yeah. Celebrate.Gong.io. And we’re announcing integration with Slack, with Zoom. Zoom we always had integration, but now it’s tighter, it builds into native API. But anything from all the web conferencing, Zoom, WebEx, Google Meet, Amazon, Chime, whatever you’re using. Microsoft Teams, obviously. Most of the telephony systems, cell phones. Then email, either Google or Office 365, Exchange, reading text messages. So ultimately our vision is to get to anything that communicates with a customer, it’s going to be contracts, proposals, anything. But we started with voice because it is a very rich data source that was still not done by anyone in a very good way. And for emails, we knew that we could get to it, but there’s less information, and there are other solutions that did it. So we didn’t want our initial entry to be based on something that isn’t difficult and doesn’t provide as much value.
So yeah, once we ingest, each one of them is getting mined for information that is specific to that media. You might think that they’re all the same, right? But even conversational language is very different from written language, very different from chat. If you think about Slack messages, very short, they have emojis. Emails, usually well-written proper English, conversational. It’s very hard to even determine if something is a question or a statement, or people don’t follow correct grammar. So almost different processing for every type of media. But then Gong parses anything from which topics were discussed, what are the action items? Questions, competitors, objections, all of the things that are pertinent to customer facing applications. And then there’s a higher order analytics on top, that you could compare people. For example, at one of our customers, I’ll just use an example. They’re selling point of sale equipment to restaurants, right?
And their solution has an iPad component and an application, right? It’s for reservations or managing tables or whatever that is. Gong recommended that they actually when they present their solution they bring the iPads before they discuss the application. Now, the beauty of it, Gong doesn’t understand their business. It doesn’t even know what an iPad is, but it just saw what the top 5% sellers are doing, is they bring the iPads before they talk about the solution. And then it trains the entire team to do the same thing, which increases the win rate by 15%. So ingestions, natural language understanding, and then there are three applications on top. One is people intelligence, determining differences between people. And then levelling them up all the time…
Meaning, the people on the customer side, or on your team?
On your own team. Your own team.
So who’s performing better? How do you…?
You kind of know who’s performing better. We knew that even in Sisense, right? We just don’t know why. So what are the differences in behaviors that might explain some of those differences, right? And what are the best practices? Deal intelligence – we look at all of your pipeline and tell which customers and deals are likely to close, which one are progressing according to Gong. For example, if the deal is 30 days from closing, right? But still there was no discussion of pricing whatsoever on any of the channels, like on emails or in one of the calls, that deal is a high risk. Or if they’re not talking high enough in the organization, so Gong tracks with email, who are you talking to? How many people are engaged? How responsive is the customers, all of those signals.
So it does measure your pipeline better than any traditional pipeline management. And the third is market intelligence. Gives a higher order – you know, how much are you losing to competitors? What the people like about your solution. How did it respond to new pricing? All those things. Very popular, not just by the sales team, but also with product and marketing teams. Gives them great input about the product.
That’s great. And then the NLP / NLU engine at the core, how much of it was developing your own algorithms and doing some AI research, as opposed to using existing stuff but just fitting in a lot of data?
Today it’s pretty much our own. I think we were probably using some open source, it’s not that everything was developed from scratch. When we did start, and this is probably an important lesson for entrepreneurs. We did not have our own speech engine. Even though it was key, we used some cloud-based solution that was not great, but okay. That was with the seed money. We did not want to burn $2 million on developing our own technology. The biggest risk for our customer is not finding a product that people want to buy and pay for. So we wanted to create the experience. We want to know what the application is. And only after we raised a Series A, we developed our own that provided higher accuracy, lowered the COGS, so we had better margins. But the first thing was to find something that people love and were willing to pay for that love.
Interesting. And how did that translate in terms of teams? From a company where AI is so central to the product, did you have deep AI researchers day one? Or you started bringing them in precisely when you started building your own product?
When we started building our own. So the first it was more an application. Can we create a good enough experience? Something just, I mean use MVP. My partner Lonnie calls MWP, Minimal Wowable Product. People say, oh wow. So we had that. Nobody complained about the accuracy and all of that. But then, we have a real research team. A lot of people, R&D means engineering. At Gong, no. We have about a dozen researchers. I mean, the entire R&D is 100 people. So it’s 10, 12%. They work on real long-term research projects, some of them don’t succeed. And that’s okay. So we are taking risks. They’re always staying a year or two ahead. So there is a real team. For speech alone, I think, it’s outside of research, but we have four or five people working just on that.
How do you recruit them? Are you now able to recruit them because, as you said, you’re a hot company, it was a lot of traction and they’re working on instant problems?
We were able to recruit them before we were hot. I think that one of the things when Eilon and I started, remember, we did a survey of both the market, what people buy, but also the technology. And we interviewed a lot of people. So we started a company in Israel, and I don’t know if everybody knows. But the R&D is based in Israel. We pretty much went to companies who did similar speech technology for different applications. So we mapped everybody in the market, then we started knowing who’s good. These people know each other, so it’s a network. So once you hire one or two, odds are that they can bring their friends, and that’s how we started building. So the team was built organically, but we made it a point to really know everybody in that nearer community.
And do you encourage them to publish as well?
Yeah. They’re publishing research. I mean, I don’t know that I encourage, but we definitely don’t stop them if they have motivation. They speak in conferences. Yeah, definitely they publish articles.
Is the product structured in a way that the AI learns across all customers? How do you think about that?
Yeah. Some of it is across all customers, and some of it is customer specific. What’s unique about Gong is our audience, right? They’re not data scientists, right? These are salespeople or sales leaders. I mean, they love the insights, but they’re not going to do anything. They’re not going to crunch numbers, most of them. So we couldn’t rely on anything, that’s why the learning is unsupervised. It means that we’re not asking people to label or train the system. You just, you turn it on and it’s working. Everything that we do has to be fully automated without human intervention. As you get into a new company, Gong will start, obviously first understanding the words. That gets better over time. Words like competitor names or product names, the system starts to learn, if it’s being repeated enough. Fortunately, we have the emails as well, so that’s a good training set. Because we know that some of the words are unique to the company.
Even a name like Sisense. Because it’s hard, right? Or Gong, it could be an object or a company. So that varies by companies. So that’s on a speech level. Then there are the topics, right? So there are generic universal topics like pricing or next steps, or even small talk. Gong can recognize across all the customers. He said, okay, this part of the conversation is a small talk. We could be talking about the Lakers game, or we could be talking the weather in New York, or the air quality in San Francisco. It’s not an easy problem. But that Gong learns across all customers. Under company specific things that it learns, once it’s had a good number of conversations, for Gong, privacy is something that we get asked about. It’d be like big brother, right? It’s a common question. It’ll start to learn of things that are unique to Gong. And if you want to create deep insights, it has to be company specific.
Okay. That makes perfect sense. So what’s on the roadmap? What are you guys working on now? You raised a large round with I believe a 200 million Series D that was announced just a few weeks ago. Congratulations on that. And a reported $2.2 billion valuation, making you guys a Unicorn. So all of this is very exciting. So what are you going to do with the money? Is that a global go to market expansion? Or is it more product? What’s the plan?
We’re thinking about buying a few islands in Greece, and making an office.
I’m available if you need somebody to visit.
The money’s there. I think we didn’t need the money. In fact, we raised a $65 million Series C from Sequoia in December, and we still haven’t touched that money. So it’s just there, it’s just prudent on companies or CEOs to recruit when they can. The market is good. So it’s there for two reasons, first for emergency, in case the world turns on us, or who knows what’s going on right now. So it’s always good to have it there. Potentially to make acquisitions when the opportunity does allow you to do several things. And to give you some margin for error, right? If things don’t go as well as we hope. But the plan is to … this should get us well beyond an IPO, that’s the path for Gong, within a few years. So now we don’t need to worry about money. We are spending heavily on R&D. I mentioned we had about 110 people in R&D right now, we’re doubling that next year. Or in the next 12 months, even before next year.
And are you guys global? Does the AI work in all languages? And so on and so forth.
It does, it does. Not as well still as American, right? So that’s English. But it will. We haven’t expanded global yet, more for a marketing reason than for things. Because we want to create more resonance in a smaller echo chamber, or a big echo chamber, like in New York and San Francisco, versus spread yourself in multiple countries. But once we get the momentum going, probably next year we’ll start expanding to other countries.
Great. So maybe one last question from me, and then we’ll open it up to some Q&A from the audience. The theme of category creation is very interesting. So this whole conversation around revenue intelligence. How do you go about that? And how could other founders or teams that may listen to this, how could they learn from your experience?
Yeah. This is strategic. The first thing you need to know, what game are you in? Either you’re an existing category and the thing is differentiation and standing just above the crowd. Like BI, for example, the previous company I managed, that was it. People want to buy, right? So they have budgets and you just do need to explain to them, you are not the other guys. That’s it. So they need to hear about you and they need this. Category, it’s more complicated. It has advantages. There’s very little competition, right? And you could get people to buy, but if you want the market to evolve you have to become a must have. So the line items and budget exist, right? And it’s one thing, I mean, you could grow pretty quickly if you’re selling for $1,000, it’s not a big deal. You don’t need to worry about this. But if you want to get to seven and eight figures, it doesn’t just happen. So you need to educate the market, this is something that you must have.
Why, what are the benefits? Why is it a game changer for your organization, for the more conservative type of buyers, for CFOs? And that takes time. You can’t fully control it. You can’t do it in six months. It’s like pregnancy, right? It does take time. You can accelerate it some more, and if you like it better or worse job, you can influence it substantially. But if you’re in that thing that you’re not replacing something or not competing with something, then this is strategic… It does take time. It requires money as well.
In terms of concrete steps around that, what do you do? You create a conference around it, you create content like a podcast? Or, how do you just say, okay, this is our category?
I think that first, you can’t really create a category, right? Just technically. You can help it happen. It’s got to be genuine. It’s fundamental. There are few companies that have done it. You got to create a great product, get people passionate about it. And start with a smaller circle of people that would advocate for your product. That’s the core. Without it, nothing happens. And start growing, it’s like a fire. Yeah, conferences can help. But if I just do a conference, that doesn’t do it. It’s one of the things.
It needs to be the full story, the product that solves a problem, people that are super excited. It might be a small group initially, and that’s fine. You don’t need the entire world. Just preach to the converts. Not the Yankees. And grow it over time. You need to have a story. You need to understand people why this is a big deal. It’s important, both for investors and the buyer community. There’s actually a pretty good book by Anthony Kennada, he was the VP of marketing at Gainsight. He had done a pretty good job, so I think he explains it way better than I am. But those are techniques that can help you, but get something that people don’t just say is good. This is a good product. You need to have an amazing product, right? That has enough power to create a fire.
Great. Wonderful. All right, Jack, folks have some questions?
So a couple of questions around your comment about Gong being an advanced CRM. In that world, how do you differentiate from a tool like Salesforce, and is Salesforce even necessary?
Yeah. That’s a great question, and we refer to CRM because that’s what people understand. Gong actually does not replace CRM. It’s a new system, just like CRM did not replace ERP, but nobody thinks about managing customers with an ERP today. So it’s a new kind of system. CRM is a system of record to see who your customers are, what’s their contracts and your contacts. It’s a database. It doesn’t tell you what’s really going on with your market or your customers unless people put in. Gong is an autonomous system. It’s a self-driving system that helps you manage your customers 100 times better than those forms.
You mentioned you sold to enterprise customers earlier than you’d thought you would have. Can you give a bit more detail as to how you landed these first large enterprises, inbound, outbound, previous relationships, et cetera?
It was actually inbound. They came to us. So the first customer was LinkedIn. They were interested in someone they knew from a previous relationship. And they told me they like to play with new technologies and we said, “Okay, that sounds great.” And now they have like thousands of people on that. And then we had another one, which I can’t mention public, but they have like several thousands of people. And again, they’re intrigued, they tested the software, they loved it. They prepared a very meticulous POC with metrics and all of that. And following that successful POC, they bought and they’ve been a customer for three years now or maybe four.
How have you addressed training the team on best practices learned? For example, the iPad lesson and others, that might be more complicated.
I’m guessing that the question is, how does Gong train the system? We focus on things that are kind of easy to train. The system can’t, or at least we don’t know how to teach you how to be funny or tell jokes. But it is simple changes that are very easy to train and easy to detect and are pretty deterministic. So what should your call sequence be or who to reach out to the organization? You can tell you’re speaking to the wrong person, you need to reach out to this. So those things are very easy.
For example, one of our customers, it’s a big brand and part of the presentation was that they speak about who they are as a company. And Gong found, if you go over two minutes, up until two minutes, okay, but after there’s a huge drop in conversion rates. It found that actually the younger reps tended to talk more about the brand, which is ridiculous. You don’t need to tell who you are, people already know. But maybe, I mean, we don’t know why it’s a correlation, not causation. Maybe because they’re not as secure. So they’re leaning on the power of the brand versus like having a real combo. So it’s the topics that we discussed and the actions, the interaction. Are you being a good listener or not a good listener? It’s things that are very easy to train.
What machine learning framework do you use and what motivated that decision?
I have no idea, I have complete ignorance here. But I can-
That’s all right. We can follow up if needed. One other question. So another question from Alan, it’s similar to the question about early enterprise customers. But can you talk a little bit more about how you got your first beta customers? Were you targeting companies that were using certain tools already? And how long did it take you to convince them of the effectiveness?
Yeah, it’s just like all start-ups. You beg, steal and bribe, just that’s what you do. Call companies from your networks, beg them, they’ll do me a favor. I don’t have anything intelligent to say, just find people that you know, or use your network and they’ll do you a favor.
What did Sisense do differently to almost create a mafia of sorts with some ex-employees starting Gong, Firebase, any special sauce?
I’m not sure that I understand the question. What did Sisense do?
Oh, I think the question is there’s a lot of great companies coming from ex-Sisense employees.
It’s a great company. Great company, great culture and great people. And we hire well and you help people then that’s what happens, you have got good genealogy.
That’s great. Thank you.
All right. Well Amit, thank you so much for sharing all of this candidly and the journey and all the things and congratulations on everything so far. It would be an understatement to say that we keep hearing great things about Gong across our portfolio of companies are using the product and there’s a lot of very happy customers out there. So congratulations on all of this and thanks very much for coming back to the Data Driven NYC family, and sharing this story.
Excellent. Well, thanks for having me.