Hey, everyone. I'm Pinjalim Bora, mid-cap software analyst at J.P. Morgan. I'm delighted to have here with me, Dennis Woodside, CEO and President of Freshworks. Dennis, welcome to the conference.
Thanks. Thanks for having me, and thanks, everybody.
So let's start with a little bit of an introduction about yourself. You're new into the CEO role. Maybe talk a little bit about yourself and a few words on Freshworks for the people in the audience that might not know about the story.
Sure. So I'm Dennis Woodside. I've been at Freshworks now for about 18 months. I've spent a lot of time in technology, spent about 20 years, 11 years at Google. I ran sales in North and South America for Google. Ran a company we bought called Motorola Mobility, and then was the COO of Dropbox. Helped take Dropbox public, grew that business from about $200 million to about $1.4 billion in revenue when I was there. Freshworks: so Freshworks was started in 2010 in Chennai, India. The first product that Girish Mathrubootham built was a Freshdesk product, which is for customer support. That was kind of in the earlier days of SaaS and really took off among SMBs. So the first customers for Freshdesk were international from day one.
As that business grew, the company realized that Freshdesk was being used in IT departments, so Girish built a separate product called Freshservice to handle help desk needs in IT, and eventually moved from ITSM, added operations management, added asset management, and then more recently added ESM for teams outside of IT. So the business initially started as SMB, but companies dragged us into bigger and bigger accounts. So today, we have 67,000 customers around the world. We have a business that's basically equally balanced between our ITSM business for our service and our customer support business, which is anchored in Freshdesk. We have large accounts, like Sony Music, Vice Media in Europe, many others, but we also serve SMBs, with larger accounts now being about 60% of our overall business.
That's a quick thumbnail sketch of me and of Freshworks.
That's great. Thank you. I intended to say this before, but I forgot. Congratulations on the new CEO-
Thank you
CEO role. You said 18 months about in the new role. Maybe talk about what is working really well at this point within Freshworks, and where do you think there's some room for improvement?
Yeah, so what attracted me to the company in the first place is that we build software that every company in the world needs. Every company that has a customer support team, a sales and marketing team, or an IT team, needs to automate the business processes that those teams have, it needs to automate the workflows that those teams use every day. So, so Freshworks has successfully moved from single-product company to multi-product company, and one of the big, shifts, which I talked about earlier, is this shift upmarket. Most of our growth now is coming from accounts that are spending more than $50,000 a year with us. That cohort, which we disclose, is growing at about 29% year-over-year.
We categorize our customers into SMBs, which are 250 employees and under, and then mid-market enterprise, which is 250 and up. Our business is increasingly being driven by that larger account segment, 250 and up. We've successfully made a shift from an inbound-only sales model, relying entirely on web, try to buy type of sales motion, to one that's both inbound and outbound, with a global field sales force, primarily based in the U.S. and Europe, but also in Asia-Pacific. That motion up market is creating lots and lots of additional opportunities for us to expand, to find new opportunities in business to serve them.
So an example there is our recent acquisition of Device42, which provides very advanced IT asset management capabilities, which larger and larger customers are asking us to bring in when we're competing. So what's exciting to me is the move upmarket is working quite well. That shift into from a single-product company to multi-product, and in particular in IT, continues to work really well. That IT business for us is very profitable. All the dynamics are great in terms of expansion, retention, so we like that a lot. And in the meantime, the support business continues to chug along, very profitable business as well, and you see that in the cash flow line for the year.
Yeah. So as you go into this new year, talk about some of the priorities that you have, more tactically in the near term, and maybe strategically over the next two years or so?
So one of the things, you know, when I joined, that we really focused on was that shift and is that shift upmarket, and one of the things that I realized is that that inbound business is very different than that field-driven outbound business. So one of the tactical changes that we've made is splitting leadership of those two business units and having two leaders assigned to each. Abe Smith joined us from Zoom. He leads field sales, and Mika Yamamoto joined us from F5. She leads both marketing and that inbound SMB business. So that is creating greater focus on both of those motions.
We're building out all the teams, if you think about field, that you need to support a larger and larger scale, field operation, larger and larger accounts and customers, things like our partner, business, professional services, and so forth. And then on inbound, we have some work to do to improve the try to buy experience, which Mika is leading.
Yep, got it. So let's take a step back, and this is a question that we hear from some investors. Freshworks is working both in the front office on the support side, helping your end customer, the customer's end customers, as well as working in the back office, helping your customers' employees. Right, so what, from your perspective, what is your vision longer term? Does Freshworks become like a one-stop shop across you know, both front office, back office for SMBs? Or would you lean more into the back office because ITSM is really working well for you? Maybe talk about that.
So we are leaning very much into the back office because IT is working very well for us. We see a huge opportunity there. In that market, you have ServiceNow, which competes very well in very large, complex enterprises, and then you have legacy on-prem solutions like Ivanti Cherwell, who have become more cloud-focused, and you have Atlassian, who's coming at it from a developer standpoint. But there's a big space in the middle for a company with 20,000 employees or so that has a sophisticated IT department, has sophisticated needs, but doesn't need the complexity that a ServiceNow brings, needs something that's easier to deploy, faster time to value, overall lower total cost. That's what we provide, and that's where we're having great success, and that market is very large.
So we're very much focused on driving success there and continuing to expand there, investing there, you know, with, with Device42 as a, acquisition that will help us, win more share there, as well as provide a natural upsell to our existing accounts. In customer support, that business for us skews more SMB and much more inbound. That has different dynamics. We are earlier in the shift upmarket there. We are having success in moving into larger accounts, like a Stitch Fix, which has about 5,000 employees, moved from a competitor onto our system for the same reasons: easier to deploy, faster time to value, AI-enabled. But that business has different dynamics, and that's been more affected by macro around SMB for certain.
Yeah. Interesting. I wanna go into a debate that we are hearing more on the CS side, but maybe a little bit on the ITSM side, too, but this debate around bots and seats. And it seems like if you have... You're largely seat-based at this point, that's my understanding, but seems like there's a leaning towards, at least from a customer's point of view, with GenAI being the topic of discussion, towards bots over time. Better bot-based deflection lowers the need to have more seats. But then, on the other hand, you can probably capture more value using bot sessions versus seats. But the flip side of that is, do we see more competition on the bot-based solutions from third parties outside of the incumbents?
That might lower the ability for you to capture more values on the bot side versus the seat side. How do you think all that plays out if you look at, I don't know, the next couple of years?
Yeah. So let's step back and just give a quick refresh on our AI strategy. So we have two products in market now. One, our moniker for AI is called Freddy. So we have Freddy Self-Serve, which is the bot product that a consumer would encounter if they're looking for, you know, "Where's my order?" That sort of question of a customer of ours. And then we have Freddy Copilot, which is agent-facing, which suggests answers to the agent, enhances the tone of those responses, summarizes conversations for escalations, and so forth. Copilot went into GA in the middle of Q1, and Freddy Self-Serve has been available since around the summer of last year, but we've had bots for much longer than that. We're just building, or we just started building in GenAI functionality last summer.
So both of those have traction in very different use cases and markets. In, for bots and Freddy Self Service, we're seeing great traction in consumer-facing businesses, as you'd expect. So customers of ours that have millions of customers of their own, that have frequent, frequently recurring types of queries, like a "Where's my order?" or "I need a refund," and the AI-enabled bots handle those, those queries quite well. The customer satisfaction actually is as high or better than human, answered queries. The time to resolve is faster, typically. The way we monetize that product is through consumption. So we sell bot packs. You buy 1,000, bot sessions. A session equates to a back and forth with a customer over a 24-hour period. And so as, as, usage of those bots and that AI scales, so does our revenue stream.
On the Copilot side, we charge $29 per seat adder on top of the seat license. So we launched that product in beta last summer and offered it for free up until February, saw quite a bit of adoption, and now we're converting those customers to paying customers. So we're in terms of the debate, like, how is it, how does this play out? Most of our customers, you know, they're looking for a single solution to provide ticket deflection, and we've been in the ticket deflection business for years. We've had bots for years. They're looking to continue that relationship with a provider that provides that kind of support, but they also need the typical escalation capabilities that Freshdesk provides on the customer service side. They need the answer center.
They need all the analytics that are built into that, and they want their AI to also be enabled and specific for them and trained on their data, and that's what we do. So you'll get some customers who might say build their own or might separate the ticket and capacity or ticket functionality from the bot itself. We're not seeing a lot of that right now. Like I said, our customer base tends to want one place to buy their customer support software. And the way we've set up our pricing, as and when seats decline, because more and more of that activity is automated, we benefit from the consumption pricing that will be increasing as a result. So that's how we're seeing the market right now, but it's still quite early.
If somebody is leaning in on bots versus seats, from a unit price perspective, is that a little bit positive for you?
Yeah. So we've structured the pricing such that, if a seat is eliminated, and we've calculated what we think the number of queries that a human can handle in a year would be, and set the price for the bot so that we actually wind up making more money. It's basically, you know, from the customer support leader's standpoint, you know, they're looking at a $60,000-plus agent, right? Or they're looking at another $2,000 in software spend. So the trade is pretty straightforward for them, and the economic value is there for us as well.
Yeah. Not to harp on this point too long, but recently, you talked about some pretty good stats, 40% deflection rates among your customers, I believe, using Freddy Self-Service, 30% productivity increases among agents. What are you hearing from those customers when they see such deflection rates? Are they reducing agent count at this point, or are they slowing down hiring of agents? What is their customer behavior when they see such deflection rate?
Really, I mean, so we serve every industry and customers of all different sizes. So the... I wouldn't say that there's a singular behavior that we're seeing amongst all of those customers today. So you have some customers that are saying, "I want to redeploy my my service reps into more sales-facing roles. I want them to be able to drive upsell and cross-sell." So when a customer calls in, they make that customer more aware of all the other offerings of the company. You have others who are just saying, "Nope, I want to reduce my investment in support or in people and rely more on machines." And then you have others that are going through massive business transformation, either growth, consolidation, and it's a combination of the two.
I don't think that there's any one singular approach that you can generalize across basically the entire economy at this point.
Yep, understood.
And then there's a class of customers that have very technical products, where they're, what they're looking to do is just use the AI to enhance the ability of the agent to answer those technical questions that often would take a very long time researching and responding to, and provide higher level of service to their customers.
Yeah. This entire debate started because one of your customers, who, who's still a customer on the L2 side, created something on the L1 side on their own, and we can discuss about that. But the point I'm trying to figure out is: How difficult is it for somebody to go that route? Because most of your customers are mid-market customers, they don't have the capital or the resources to focus on this problem entirely. So talk about how difficult is it-
Yeah
Yeah.
Well, let me, so, let me talk about what we've built. So we have close to 2000 engineers and product managers who are focused on building and supporting all of our products. We've been building AI products since about 2018, and the current crop of AI products consist of 60+ separate services. So an example is a summarization of a conversation or suggesting an answer based on what the AI knows about our answer center and prior inbound queries. So 60 discrete services that add up to Freddy Copilot or Freddy Self-Serve. We have built 30 machine learning models to actually deliver those services. We partner with not just one, but I think it's something like 15 different outside providers, both open source and, like, Azure, for our GenAI capability.
So to deliver a service, it's not just, you know, build once and you're done, and these models are changing all the time. The models that are available open source are changing all the time. So that's what our engineers do. They are on top of that, and whenever a new model pops up that's accessible to us, that we can use to improve the service that we deliver, we can pop that model into our system. We've architected it such that it's very easy for us to do that, to improve the service that the customer experiences. So yes, someone can go out and build a one-time project to deliver a bot, but to keep it current, to keep it secure, to keep it safe, all those things are very hard.
Lastly, one thing that I don't hear investors who are thinking about this as a bear thesis is the data that you have, right? How... Talk about that data being a moat. Is. How, how significant is that for the L1 side of the business?
Yeah. So we have, you know, every single day, 1 million-plus kind of customer inquiries, queries coming in in the form of questions from our customers' customers. And for our models, we have access to that data to tune them and to make sure that the answer that any specific customer is getting is specific to them and is trained on their answer center data, their prior query data stream. That's not accessible to a startup. That's not accessible to, you know, someone who's trying to do this for the very first time. That platform really matters, and it allows us to execute quite well.
Yeah. So moving on to macro, in SMB prints have not been really great overall in software. In Q1, it seems a little bit of a downtick, frankly, and a lot of surveys are saying the same thing. So maybe talk about kind of your sense of the macro environment at this point, from talking to customers, how do you characterize kind of the demand environment?
Well, I would say, remember, 60% of our business is not SMB. It's, it's mid-market and, and up, and that buying behavior really hasn't changed. We continue to see traction, in particular on that IT side, where IT departments are looking at whatever they've deployed, maybe they made a decision three years ago, they're at the end of their contract. Three years ago, our product wasn't as mature. We're now in the RFP process, in those deals, competing for those deals, and winning those deals. So that remains the same. SMB has been challenged for a while, and we've, we've talked about this in the past, and we're seeing it in terms of lower net adds, we're seeing it in terms of slightly elevated churn in SMB in particular, and 40% of our business is on that SMB side.
Yep, understood. So let's talk about the go-to-market changes that you talked about, right? You have two different leaders. Talk about what has changed exactly in terms of processes. Are you doing more of the same in the outbound channel? In the inbound channel, what needs to be changed? How long it's gonna take?
I think on the outbound field sales side, we're just in more and more larger deals. So a year ago, a $100,000 net new deal, new customer, never worked with them before, that was a big deal. Now, it's pretty routine, and $200,000-$300,000 lands are more the norm. A year ago, a $200,000 expansion was a big deal. Now, it's more... We're seeing more and more of those as well.
So that requires a different level of skill, different reps, so we've changed out about a quarter of our overall team over the last 18 months at the leadership level, bringing in people with more experience at, you know, the Salesforces, the Zendesks, and, and ServiceNows of this world, because that complex sale is very different than where we grew up, which was inbound. That is working well. We're now focused on building out our partnership ecosystem. We did a deal with AWS, we're one of their preferred partners. They're bringing us into more deals. We can offer retirement of AWS credits when someone purchases software from us, which is a big deal. You can buy through their marketplace.
And we're expanding our set of service providers that can help with post-sales implementation, so that's also something that's super important when you get into that enterprise. All those things are helping us push into larger and larger accounts, like an A+E Networks or Sony Music, and win deals that otherwise would be competed for primarily by ServiceNow or like an Ivanti, Cherwell, player. On the SMB side, you know, Mika has been here now just a couple of months, but we're very much focused on the try-to-buy process. So we have plenty of trialists every quarter. Where we need to improve is converting a higher percentage of those trialists to buy, and that means reengineering our product to make it easier to discover value and to get to value faster for typically a smaller business. So that's another area of focus for us.
Then lastly, on the CX side, moving up market more aggressively. We've created an overlay team that just focuses on CX deals, that's had success in winning deals like Stitch Fix. Last quarter, we had a customer called Big Lots, which is a retailer here in the U.S. Those are the kinds of successes that are getting us more and more up-market success. So those are some of the tactical things we've changed.
Yeah, that takes me to the cross-sell point. A quarter of your customers, I think, use two or more different solutions, I believe. But that stat I think is more around CX than CX plus ITSM, is my understanding. You are starting to see, though, some of these ITSM customers buy CX as Stitch Fix, you're talking about. Is there a different motion there? Because those seems like two different buyers, overall.
Yeah, they are two different buyers. So, over half of our top 20 customers are both IT and CX customers, and typically, what happens is, a customer will come in, either the CIO will buy us for Freshservice, or the head of support will buy us for Freshdesk or the customer support suite. They'll have success with that product, and we will learn that their contract for the other product set is coming up for renewal. We'll ask for an introduction, and we'll get into an RFP that way. So it's more of a referral sale than the same center of power within the organization making the decision. But we are having success, Stitch Fix being an example of a longtime Freshservice customer. I think they were a customer for 7 years on the IT side, referring us in.
They weren't happy with their prior provider of customer support software, and us getting involved in that deal, getting into a competitive RFP, and winning that deal, and that's fairly becoming more and more common as well.
One thing we have not touched is the sales and marketing part of the business, which is pretty small for you at this point. Talk about that part of the business. What needs to be done for that to ramp more quickly?
So sales and marketing is, or CRM is a, is a newer product line. It's our newest product line. We shared at our investor day, it's about 5% of our overall business, so it's quite small. We think that there's promise there in that a lot of our support teams are asking for, a sales and marketing tool. They're asking for something that's integrated into the support, system, single customer record. There's a lot of work that we need to do to continue to build that product to make it ready for those larger and larger enterprises that are, that are in the support side as well. And it's still early, but, I think we have some opportunity there.
Yeah. 10 minutes left. Let me see if people have questions. Okay, we need the mic, please, just hold it.
That Stitch Fix-
Oh, sorry.
Who did they leave?
Can you repeat the question?
Yeah, I can't share who they left.
You can't? Okay.
Yeah, one of the big players, though.
Okay.
Yeah.
Thanks a lot for taking the time. Just on the CX business, what insights can you share that help us understand why you characterize the weakness there as an SMB macro issue, when we haven't seen a clear kind of SMB signal across SMB-focused software businesses this quarter? How much of that is a product category issue versus macro, and how do you differentiate the two? Thanks.
Yeah. So I think it's a bit of product category and a bit of macro. So for that business for us is over 60% from SMB, from 250 and under employee companies, so relatively small customers. We do think that it is partly macro, and we have seen other companies report similar weaknesses. So, if we look at the reasons for a churn, for example, a large reason for churn for us is just going out of business, and that, to us, indicates that there is some macro element there.
There's definitely a category issue as well, and there's an execution issue on our side, which I've talked about, which is we have to get better at helping customers understand early what the value is from the software, getting them into the trial, getting them up and running faster, and having them see the real value of the software earlier. That's something that Mika is very much focused on, and we've sort of lost sight of over the last couple of years while we've been focused on up market.
Anybody else?
So, thanks. I think you were touching on it a little, but the speculation is that, you know, this seat-based model is going to be vulnerable over time. Because if you think about ITSM and customer service, if I'm a company, my goal is to reduce, use AI to reduce what I'm spending overall, whether it's people or software. How do you know, you know, this guide down, that that may already be happening in your business? Do you have enough data with your customers to determine, "No, that's not it?" 'Cause I think that's a speculation.
Yeah, we've not seen that to date. And at the end of the day, the arbitrage is between the $60,000-a-year customer support rep and spending $2,000 a year in software.
That is a seat. So if he's gone, that seat-
But there's still... You still need the software.
Yeah.
So there needs to be some way of monetizing. The way we monetize is based on consumption. So in that model, if that seat goes away entirely, that's where we make our money on consumption. You know, I think we're going to have to see how it all plays out, but so far, that's not what we're seeing in our business at least.
Anyone else?
Could you-
Hold, can you, can you hold on a minute for the mic?
It's just a continuation on that question. Earlier, you were walking through the concept, how you're pricing your bots to make sure that you make more money if a seat is cut. Could you just lay it out in simple terms for me?
So the pricing for the bots is $100 for 1,000 sessions. And the way that we arrived at the $100 is, and our estimate of how many sessions would a human be able to serve in the course of a year. And then we kept 10% of that value, so to speak, for ourselves and shared the rest back with the customer. That's how the math works at a rough level. Now, we'll have to see how all this plays out, right? Because the customer base that we're dealing with is very broad and wide. Some, in some cases, customer support reps, you know, may be handling far more queries than what we estimated, and some far less. But that's how we came up with the pricing at a high level.
So the subscription you lose from the head loss, you make up more than 10%. You get a 10% net gain?
No, no, no. The way that we set the price, right, is that we're going to share value with the customer such that they're saving $60,000 on a seat, right? We're gonna drive about 10% of that back to us in terms of licensing and consumption revenue associated with the bots.
What does that mean on a net basis, seat?
It depends. Again, it depends on lots of different variables. The rough estimate that we come up with is it's somewhere between 10%-15% lift on a per customer basis. But again, we're dealing with hypothetical estimates before we've seen the actual data. Yeah. And w-
When might you see, or when will we see from you that actual data?
We'll have to see how adoption goes, and I think it's going to be some time because we have a very broad set of customers. Some are jumping right in and want to automate right, right now. Others are much more kind of backward-leaning and waiting for others to take the lead. So I think it's going to be some time. And I think the whole industry is going through this, and there hasn't been much that's been sort of shared at this point for that reason.
Sorry, just, just one follow-up on the same issue. Those economics assume kind of a 100% capture from existing customers into your self-service bot, how, how do you think about monitoring that conversion rate, and what have you seen thus far?
Yeah, so we look at a couple things. We look at conversion. Remember, there's two different aspects. On Copilot, we look at conversion from our beta program into paid, and then on the self-serve side, it's converting from the non-AI bots into AI bots first, and then ultimately, how does that reflect in seat count? So again, it's still pretty early for us to share any meaningful data there as to how that's gonna play out, but we think that equation actually looks pretty good in the long run.
Anyone else? Okay, we haven't talked about one thing, which you have not released yet, which is Freddy Insights. Which seems like almost a way to let the software create bots, which a manager can look at queries and say, "Okay, here is a query that is getting..." or a use case that is getting a lot of queries and is being handled by agents, and these answers are pretty standard, and you can actually convert that into a bot in a very easy way, is my understanding. Help us understand how is that gonna feed into kind of this AI story and your consumption of bot sessions?
Sure. So the problem that Freddy Insights is solving for is really, think of the manager who's coming into work on Monday and trying to understand, let's say, at an e-commerce company, what happened over the weekend. Maybe it's a big holiday weekend. How did we handle our overall support load? Where are there weak spots? Where were there strength? Now, today, that manager would have to hunt and peck through a bunch of Power BI reports or spreadsheets to try to understand the dynamics of the business. Freddy Insights is designed to solve that problem. Because you have all the information already coming into our systems about performance of different regions, reps at the product level, which products are coming back with lots of issues, and which products are not.
So we can ingest all that data and then surface insights to the manager that don't require a human to go hunt and peck around a bunch of dashboards. So that's in beta now. Early signs are promising. It's a hard problem because you have to train the AI on what would a manager actually want to understand, but we're excited about where it could go, for sure.
Yeah. So, I'll ask you about the growth equation for the company from here, right? You have one business which is probably growing in the 30s, would be my guess, and then you have another business which is slowing down. I don't know what the growth rate is. Maybe you highlighted about, in the analyst days,
Yeah, we shared-
5% or so.
Yeah, we shared, we shared at our Investor Day that our IT business was growing in the high 30s, and our customer support business was growing in the low teens.
Is that still the case?
Generally, yes. It's-
Okay.
Yeah.
Okay. So then, looking forward, how do you see those two kind of trend? Are you saying at this point, it's been one year since the Analyst Day, would you still say you... That framework that you gave us, that CX business will slow to, I think, mid-single digits, and then ITSM-
Yeah
Will kind of continue at 30%. Is that, does that still hold?
Yeah. It's been nine months, and that framework still holds. So on our Analyst Day, we said that our goal is to be a billion-dollar business in 2026, and that goal still holds. Our business will increasingly be as IT, and our business will increasingly come from the mid-market and enterprise customer, and that's those trends are all true.
Okay. And lastly, this is more tactical, but this year, seems like there's a ramp in billings in the second half. So, I've been getting a lot of question, what gives you confidence on that? Maybe talk about that.
Well, we already have insight into renewals that are coming up. That gives us some confidence as well. We have good insight into the trends that we're seeing in our book of business, our pipeline, those sorts of things, especially on the larger account side. Those things give us confidence.
Great. We have 10 seconds. Anybody wanna take a shot at a question? No? I guess we can end it there.
All right.
Thank you so much, Dennis.
Thanks, Pinjalim.