Expensify’s COO: “The Key to Achieving Both Profitability and Growth”

Ladies and gentlemen, please take your seats. Our program is about to begin. As a courtesy to our speakers and fellow attendees, we ask that you please silence your cell phones. Please welcome to the stage Anu Muralidharan, COO at Expensify. Well, good morning everyone, and here I'm going to start off with a word of advice: never book your session around lunchtime because you can't compete with food. It's really impossible. So it's very exciting to be back here at SaaStr. This year is actually particularly exciting for Expensify to be here because we are the official chat platform of this event. What that means is you'll see a QR code for every session. That QR code represents a specific dedicated chat room on our platform for that session, and any of you can join by just scanning that QR code and And if you already have an Expensify account, you can sign in. But if you don't, you can sign up totally free. This just helps you continue the conversation even after the event is over, even after the conference is over. It enables you to meet more people than you can possibly physically manage. And so I hope you'll all try it out. I hope you get a lot out of it. Please give us feedback on this particular feature that we've built. We would really appreciate it. So without further ado, I want to introduce what I want to talk with you about. The topic I want to talk with you about today might seem to a lot of you almost paradoxical in nature, a concept that seems like a paradox. It's how to build a successful high-growth company that can also be profitable. Now, I have spent you know, a few years now pitching Expensify, explaining Expensify's performance to our investors, first in the private markets, then during our IPO roadshow, more recently to our public market investors. And Expensify is a cash-positive business. We're very proud of it. We've had growth and profitability for something like 5 years now. Now, when I talk about this with investors, Every now and then, more often than not before, a little less frequently now, I'll get really curious stares in the room. And then eventually someone will ask, "Why would you want to be cash positive?" And earlier on, that used to really throw me off because isn't cash positivity the gold standard? Like, don't you want your business to grow and make money? But over time, I've come to understand why that question comes up. It's because I think as an industry, we've all kind of assumed that all profitability comes at the expense of growth and that all growth is actually good quality, that it is worth pursuing. I'm here to debunk this fallacy for you. And if the slides will move— there we go. So this mythical entity, this mythical creature, if you will, that has high growth and profitability, can be every one of your companies. It's completely doable. And you know what else? It's not exactly uncool to make money. Because look at the kinds of companies that have both high growth and profitability. So I know I put Ford in here, and it's a really strange example because, you know, it's not SaaS, it's a totally different industry. But it is a capital-intensive business, and it's pretty incredible that they could achieve growth and profitability in a mere 5 months. Right? Then we have the stalwarts: Apple, 2 years; Google, 3 years; and Facebook, 5 years. So, you know, it is possible to have high growth if you have a solid product, but at the same time aspire to profitability in a reasonably short time frame. Now, I mentioned Expensify is a cash-positive business. We've been cash-positive for many years now, and that's not something that happened by accident, obviously. Nicely. We've developed a pretty good discipline and pretty good muscle memory around how to evaluate growth opportunities. If a growth opportunity is in front of us, is it truly scalable? Can it have meaningful impact, and is it positive ROI? And if not, is it worth investing in at the expense of profitability? And then we've also always had a dual focus on both growth and cost and process optimization, which is what's helped us achieve profitability pretty early on in our lifecycle. And today, I want to take some time to talk through some of the strategies that are tried and tested, if you will, some of the strategies that really worked for us. And I hope that they'll help you in your journey and help you have your cake and eat it too, if you will. So with all that preamble out of the way, Let me go through the strategies one by one. The first strategy is make it a habit to iterate on your product using real user feedback. Now, I think it's pretty obvious and pretty intuitive if I said in your growth journey, your biggest ally is your product, right? Because it's probably your biggest asset anyway. Now, if your product just had effortless growth, if it had a word-of-mouth motion and momentum, then it just automatically takes care of one of your largest cost drivers, your sales and marketing expense. It's just automatically lowered. But for your product to have that kind of momentum, for it to have that kind of viral growth, if you will, it needs to have a pretty incredible product-market fit. But incredible product-market fit is not achieved by simply tinkering with your product and your user experience in a lab behind closed doors. It's not even really achieved by asking your customers what they want. Because you'll notice that what people say they want, what they say is indispensable, and what they tell you about how they'll use a product is just phenomenally different from how they actually end up using the product. And from their willingness to try something versus propensity to leave based on the lack of something is all diametrically just opposite. I want to tell you a story to try and highlight this point. I want to tell you about how we built one of our major features, the Expensify card. So first, I just want to show you our approach to building and launching major features. So this chart actually shows you over time the concentration of the company on building and launching the card. So all of the lines you see below the— all of the bars you see below the horizontal line in faded colors are people that were not working on the card. And all of the bars in brighter colors on top of the horizontal line are people working on the card. Now you can see something like up to July 2019 or even August 2019, barely anyone was working on the card. It was just a little pilot team evaluating the viability of this product. Then between August and December, more or less the entire company was working on it. And then after December, we launched it and had a core team sort of manage that launch, gather feedback, and set the direction on how to continue iterating on this product. This is, in general, our approach to launching new features. We assemble a sprint team. We go as fast as we possibly can so that we can get something in users' hands, because their very real day-to-day usage will teach you a whole lot. And it actually gives you better direction in terms of what's going to really resonate with your audience. Now I want to tell you a specific— example, a specific challenge. So we launched this thing from start to finish in 5 months, and that was cool, but we couldn't figure out a key challenge in that small time frame. We couldn't figure out how to support all of those customers on 30-day balances on the card, because as you can imagine, if you want to support a company for 30 days on a card, all of their spend, you need to have some pretty strong reserve cash balances sitting in an account with your processing bank. So you've got to figure out how you're going to fund that. We couldn't build everything but also figure out that pretty critical funds piece in time. So we had one option. They can use the card, all of the features are available, but they'd have to pay it off every day. But what was desirable, so to speak, was the ability to have a 30-day balance. It's the norm, right? Like all of our credit cards also, we don't pay every day. So when we did user research, when we asked our channel partners, everybody told us that if you don't have the ability to do 30-day balances, just forget it, because this thing is dead in the water. So now we had two options. We could go forward and launch this with just daily balances, or we could wait, figure out how to fund it, and then launch with 30-day balances. We went with the former, because we'd finished it, so what's the point in sitting with it? Thank God, because we launched it in January 2020, and then with February 2020 and all the COVID shutdowns, it would take us a really long time to actually figure out the funding needs we had to support 30-day balances. Almost a year, if not a little bit more. But what's cool about this is even during COVID where companies were mostly dialing back spend, we actually saw steady growth on the card. It wasn't it went through the roof, but it steadily grew in terms of adoption. So we had a lot of real data to play with in terms of how companies were using it, what they were missing, and how do we build it into their very real everyday workflows. On the other hand, if we'd waited, it would have taken us maybe till 2021 to launch this thing, and there was all that valuable feedback we got along the way which we would have missed. But let me tell you the most incredible thing of all. Daily balances is the default option even today. And if somebody wants 30-day balances, they need to switch to it, which we make pretty easy to do. But what we notice is the majority of companies never bother to switch. But people will tell you that this thing is dead in the water if you don't have that feature. But then you give it to them, and a meaningful number adopt it. But by far not the majority. So this just highlights my point that when you ask someone what they're going to do, they tell you one thing, but you never really quite know till you put something in their hands. So as a message, as a takeaway, I hope you get where I'm going with this is if you have a product that is good enough as an MVP, put it in your users' hands. Because if you wait for perfection, You won't even really get there because the way to get from good to perfect is by using real feedback from real users based on real usage. So that's our first strategy. Here's our second strategy. Think about how you can automate to broaden your margins. Now most of you, if I asked you what are the top 3 cost drivers in your business, you can probably rattle it off the top of you hit, right? Now, I think most of us SaaS companies here, so I'll take a guess. It's probably people-related expenses. And if your product is at all nuanced, then a large portion of that people-related expense is probably support costs. That was a very big driver for our business. So as we grew customers on the platform, the number of active users on the platform grew, which was amazing because that's how our revenue grew. But for every dollar of revenue that we brought in, we had to spend almost a proportional amount in support because the amount of support volume was also growing. And the only way to stay ahead of that support volume was to keep hiring into the support team. So you can see on this slide, on the left-hand side, the pre-automation kind of trajectory of revenue and support costs. What was frustrating about this was for every dollar in revenue that we generated, we are spending a comparable amount on support. So what's left behind goes towards funding all of our other fixed costs, and profitability was like just ahead of us and we could never quite catch up with it. And that was really frustrating because that's what we were trying to do. So we decided to automate this line item, and we used AI to do it. Not revolutionary at all. Everybody does. But the way in which we used AI to automate our support processes is a little bit different from the large— from the majority of the market. And I'll tell you how. Most people automate support using natural language processing. They try to guess at what the customer is asking and send them to help help articles in the hope that they'll self-service. And anybody that is still not self-serviced, if you will, will go to a live agent. Now this helps because up to a point, you'll be able to take a percent of support volume and you'll be able to send it to self-service. But what remains will still grow proportional with your usage. So long as what remains is material and is still growing proportional to your usage, you'll still never get quite ahead of it. And that's the challenge with it. The way we automated our support processes was, again, to use AI. We used natural language processing, but we used it not just to guess at what the customer is asking, but also to guess at the complexity of their question. And then we created these 3 tiers, if you will. The simplest tier where an answer is nothing but a multiple choice question. The mid-tier where the question is very simple and you'd have to write a response, but it's so simple that you don't require deep product knowledge to do it. And then the third tier, which was really where we'd already been staffing most of our support agents, was to actually troubleshoot, get into the customer's account, work with them to figure out what the problem was. What we noticed was a huge percentage of questions that people are asking, and this probably isn't even insightful, it's probably very intuitive, a huge percentage of what customers are asking actually is very simple or mid-level simple. It's not super complex. So all of it went to the lower tiers, and what remained went to the higher tier. But since we were already staffed up there for all of the volume to start with, we never needed to increase that team size ever again. And then we took the lower tiers that are super simple and staffed them with people that really don't need the level of skill that we were staffing to begin with. So we were able to dramatically decrease the costs we spent on those people because our demands of them were also very, very low. As a result, we were able to change the trajectory of our support costs. And I think that is key. So you'll see on the chart on the right, it actually changes in terms of its slope. So it still grows, but it doesn't grow with revenue. And the fact that it doesn't grow with revenue means that every additional dollar of revenue that you make now has a higher chance of going to your bottom line. So finally, we were able to use this strategy to get to profitability. So— The takeaway here is not all of your large cost drivers are made equal. You need to identify a material cost, a material cost in your books that is going or trending in lockstep with your revenue, and you need to disassociate them in order to see a broadening of margins. And you need to attack that line item in a way that you can change its trajectory, and that's the key to success. So that's strategy 2. Moving on to strategy 3, how can you use outsourcing not for cost savings but for flexibility, which in turn is going to help you optimize your business for costs? Now let's stick with the theme, people costs. It's a big— not just in terms of the actual dollar amount, but in terms of the amount of time you spend in keeping up morale, in keeping up performance, so on and so forth. So when you increase your team size, you always have to think about all of those costs because they will all multiply together. Sometimes it may be a better strategy to use an outsourced team, and certain attributes lend themselves to outsourcing more than others. So again, let me tell you a story of how we did it. Pre-COVID, we had so much word-of-mouth growth that we had an internal sales team, but we never really invested a whole lot in scaling them out, and we never really thought about scaling them out before COVID Now, after COVID, the world is a little different, and late 2021, we thought to ourselves, maybe we could play with the sales motion a little bit more, and maybe that can help us sort of bridge some of the gap in terms of the volume of leads we're seeing, um, post-pandemic. So we wanted to build a team of cold callers, so an SDR team. And then we also wanted to build a team of what we call onboarding specialists who can work one-on-one with the incoming leads so we can convert them at a higher percentage and faster. But here's the challenge. We didn't know if any of this would actually work. And we particularly didn't know if any of this would be viable for us because our unit economics are a little challenging because we, at least from an acquisition perspective, very much target a small business. So if you have cold callers and you're paying them even the most optimal rate, it's sort of— the cost of sale kind of gets ahead of you. So a lot of question marks. But here's the thing about these two jobs. They are highly repeatable. They're very trainable, and once trained, the training doesn't need to change all that much because the product isn't changing day in and day out. Because of these two attributes of this particular need, we decided to lean on outsourcing rather than hiring a full-time team. For 1.5 years or so, we tried this. We learnt a lot. We were able to dial the team sizes up and down based on cyclicality, time zone, We were able to sort of explore offshoring to save on some costs. We experimented with multiple vendors so we could get some sort of performance optimization and answer performance-related questions. All of it would have been impossible with an internal team because you would have to spend a lot of time hiring them. You can't titrate their hours up and down. You can't really play around with— their, you know, you can't play around with their time zones for sure. So it gave us a lot of flexibility to experiment and learn in a pretty short time frame. What we learned is, in fact, outbound calling didn't quite work for us from an economical perspective, even if we did it offshore. So we sunset that team, and we learned that onboarding specialists were very, very encouraging in terms of their ability to convert incoming leads. So we, we are leaning more into that. So the moral of this story is, if you're experimenting with ideas, if you're not quite sure about a strategy but you'd like to try it, so long as the role itself lends itself to outsourcing, consider outsourcing before you really commit and hiring a full-time team. Because once you hire a full-time team, there is a soft cost You can't just let everybody go because your strategy isn't working. That's a huge, you know, morale-related cost for your company and is, you know, no bueno. So think about outsourcing as a way to gain some flexibility and experiment and test many ideas before you commit to them. Because once you hire an internal team, you'll end up feeling stuck with a strategy that might not completely work for you. Now, having run through these 3 strategies, I want to say that strategies, in my opinion, are sort of like ideas. They're cheap and they're a dime a dozen. Not— no 2 companies are the same, so no one size is going to fit everybody. The secret sauce that really brought our strategies together for us is our small but very powerful generalist team. So when we hire, of course we look for the specific job skills, but on top of it, we also look for these 3 very special qualities. And we think of it as the ingredients to our, you know, unique cocktail, if you will. We look for people with innate curiosity, ambition, and humility, because we think that that's what makes someone a generalist, makes them sort of like a themselves. They can identify a problem, they can go to the problem area, they can work with the people that are involved in that area, and they can solve that problem. And that's what really helped us bring together all of these strategies in a way that could actually impact the business. Now, I joined Expensify to manage risk related to payment processes and bank audits, and I can assure you there was no single day in my life that I woke up and said to myself, 'Wow, that would be a dream job. I need to go get that.' Mostly for me, it was a foot in the door. And what really motivated me was finding the next problem and trying to solve it and working with a wide range of people. I've been at the company 8 years. I've worked in literally, you know, across every team. I've worked for sure with every single person at the company. And I think that Really cool, creative ideas come out of a deep contextual knowledge about the business and not from any textbook. But for somebody to have that kind of deep contextual knowledge about your business, they need to be generalists. And for them to be generalists, you need to have a culture that actually nurtures generalists. What I mean by that is you need to have free flow of information. If they don't know most things, they can't identify problems in those areas. There needs to be freedom so they can go to any area and work hard and show you results. If they're worried about politics and stepping on people's toes, it's never going to work. So think of your team, think of your people as your strongest tool. If you enable them to be generalists, the superstars in your team will sort of rise to the top, and you'll be able to build a really strong company with really strong fundamentals. And if you have that, there's literally no storm you cannot weather. So lunch notwithstanding, thank you for giving me your time. I really appreciate it. And, you know, we have some time for questions. And I'll, you know, keep— oh, I can't go back, but there was a— never mind.

Hello. Hi.

Hi, good to meet you.

Great talk. Uh, my name is Guilherme. I'm CEO of Espresso. We are an expense management solution like Spendify but focused on Brazil.

Okay.

And my question is what you said earlier about, you know, we follow the same steps as Spendify. We first launched the SaaS platform and then a credit card.

Yeah.

And everybody loves the idea, but the adoption rate is ways lower than we thought it would be. So how you guys solve that problem? It is customer marketing, it is in their push notifications. How are you guys solving that problem to convert the base to use this new product?

Great question. So I don't know about you guys, but for us, the corporate card is just an aspect of our platform. It's sort of like a feature. In the same way that we wouldn't push someone to adopt in-app payments if they didn't want it, or need it, we wouldn't push anybody to adopt the card either. So we want the card to grow because it makes us money. But more importantly, we want a sticky customer that is paying us subscription revenue because it's really, really predictable. And we want to solve their pain points. So the way we approach it is we kind of look at what pain point this feature solves and then try to identify what customer it will speak to. And then we try to have all our marketing language kind of talk to the pain points so that people identify with it and then come to your product. Because a user isn't looking to adopt a card. They're looking to solve a problem. So if you're talking to them in their language of problems, then they're likely to look at your feature as their solution. So I don't know if that helps, but that's really our approach.

Perfect. Thank you.

Anyone else? Cool. All right, so we have the QR code, which is gone from here, but, you know, maybe you guys can get it some other way if you want it. But— oh, okay, we have enough. Sorry.

Oh, is this on? Oh, hi. Sorry. Hi, my name is Bri. I'm working in my first job out of college at a startup that grew grew 100% year over year, then 75% year over year. Um, and I'm an ops team of one and I report directly to the CEO. So there's a lot of big shoes for me to fill in a small team. Something I've noticed being a part of Gen Z is this resistance to automation. Um, it's because I work primarily in revenue ops, so resistance in automation when it comes to sequencing, outbound, just in general. What would you recommend for navigating those conversations, seeing the advantage?

Yeah, so I think generally when companies or management talks about automation, outsourcing, offshoring, what's going through employees' minds are, my job is at risk. I think that's probably where the resistance comes from, and it makes sense. I would feel the same way. So I think if you create a culture where you show people the way to something better, so A lot of what I did has now been automated, and I'm very glad, because if they weren't automated, I'd probably still be there doing that, and I'd never have the opportunity to grow and solve other problems. So most of people's anxieties are because they have no other way to see a better life for themselves in the company. So they want to keep their jobs. They can't see other opportunities, so they want to keep this one. So if you create room for other opportunities and for growth, People are gonna let this stuff go so fast, because why would you want to keep doing the same job that you're doing anyway? So I think that's the key. It's not about getting them to adopt automation more. It's about showing them the way to better things, so then you can automate their job.

Yeah, thank you. And actually, it's for us, it's like the employees are really like grabbing onto it while it's actually executive leadership that has more resistance to that. So I don't know if you have any additional fee.

So you're saying your leadership is resisting it?

Yes.

OK. Maybe they don't believe it'll work. So maybe try— I mean, maybe it's kind of scary to do a lot of work to jump into automation. Maybe it doesn't even work. So we take a little bit more of a test and iterate approach. So don't do too much. Do a little bit. Test it out. See if it works. Because then you get excited about doing more. That may be a solution.

Thank you.

Hi. Hi. So my question is in terms of org structure. Okay. As your COO, as we're approaching $20 million ARR and growing beyond that, right now the way we're structuring the org is CFO with a VP of Ops. What's your take on CFO having VP of Ops, or when's the right time to have a COO? How does that work?

Work in your mind?

So I don't know if I'm going to have satisfying answers for you because the way Expensify works is we don't do any of that. We have titles, we have people respond, like especially at the very top, but our approach isn't to structure everybody into these teams that report to certain people because that's how you end up making specialists out of them and you silo them. And if there are certain people that are responsible for certain things, quote unquote, they're not going to be very open to someone from another org coming in and helping them solve problems or taking a leadership role. So the better way to organize your team would be keep it flat. That doesn't mean people just do whatever they want. It just means, yes, you have a day job. You're responsible for this day job. But then it's your prerogative what you choose to go and work on outside of that day job. And that's what's given us the most flexibility. So I'm the COO, but I don't have anyone reporting to me. I have ideas. I will throw them out. And people get excited by them and say, oh, I want to work with you on that. And then I'll pull them along. So depending on what projects I'm working on, I build core teams. And so do other leaders at the company. And really, anyone can try that on a small scale. And that's how they end up growing. OK. Thank you.

Hi.

Hi.

My name is Sharon, and I also come from zero-to-one building and thinking about early-stage teams, and I do a lot of rinse and repeat in that world.

Yeah.

My question is also around org structure, but on a different landscape.

Okay.

Wondering when you were talking about some of your tests with one-day charge.

Yep.

And some of that experimentation, when you're putting together an experimentation team, And more specifically in that scenario from that experience, what did the team structure look like?

Yep.

And specifically what was the balance between qualitative product led, um, digging versus data or people with a data background on that team?

Yeah.

And more quantitative. Yeah.

It's a great question. So anytime we have a project, we build sort of a core sprint team for that project. And as you can imagine, you need a project leader, right? Someone that just keeps the troops together. So we always have 1 or 2 of those depending on the size of the project. And then we have somebody or a team of people, again, depending on the size, that write— in our company, the way that projects go is we sort of have an idea. We have 1 or 2 people leading that idea. Then we write a design for it, which we call a high-level design. It doesn't get into the technical details. And I want to say this is a product person, but again, we don't label people. So a lot of people that worked on support for a really long time had very in-depth knowledge about the product and what the customers want, and so very naturally transitioned to being able to lead projects from a product perspective. So really, it's anybody that can clarify and elaborate and write clearly what the product would do, how it would look. And everybody has access to Figma. They all play around with mockups. So a group of people kind of own the product side of that. And then engineers sort of come on board. Sometimes they're involved in the product side as well. But then they come on board and kind of detailed design, flesh out how this would get built from a technical perspective. So now we have the project leader, a couple of people from a product perspective, a couple of people on the engineering team. And then we have some amount of marketing, customer success, and sales involved, because you need to know how to cross-sell it if that needs cross-selling. You need to know how to support it. You need to write help documentation. And then you need to know how to pitch it across marketing channels. And once it's launched, most of the engineers are off the project. They go on to the next thing. Even most of the product-related people, all of the people giving feedback on that high-level design are off to other things. But the project leader— and a small team of sales, marketing, and success stay back. Because what's really important is you launch this MVP. Your test actually needs to have some sort of feedback loop. That's really where we bring data. But even when we write the high-level design, we write down exactly what data we're looking for, how to query it, how to get it, what frequency, how to put it in front of people. And then we have the team that remains after the MVP sort of execute on that. So we are looking at feedback, understanding trends, and operate— and when we iterate on this design, that we're using that very real feedback to do it and not just guessing at it. Thank you for asking. Awesome. So while I was doing that, we got the QR code back on. So if anybody thinks of questions after, if you want to just say hello, like, please feel free. This is the chat room in which I'll be. Thank you again for your time. Have a great conference.