SaaStr 2025 Partner/Workshop Stage

It. Sam it. Sam. It. You. Sa. Sa. It. It.

Please take your seats and silence your devices. Please welcome to the stage CEO and founder, Runway siki chen.

All right, thanks for coming. Who's excited to talk about financial modeling in SaaS? All right, all 10 of you. Love it. Okay, so I'm Sikhi. I run a company called Runway. We're an FPA platform for data driven finance teams. We currently help a little over 200 companies simulate how their business actually works. These include companies like AngelList, MyFitnessPal, Superhuman, et cetera. And over the course of building this company over the last couple years, obviously we've seen a whole lot of finance models all stripes. And we never met a customer who doesn't come in and feel embarrassed about their finance model. It feels like, oh, it's super complicated, my data's a mess. And that's pretty common. But a finance model usually is technically correct, right? So you have like accurate KPIs, it's GAAP compliant, it's cruel based. We also found, and this is pretty common with every finance model, is that there. Generally there's a lot of frustration because these finance models are so accurate and so clear, but they're still ignored by the people who need them most, the sales counterparts, the marketing leaders, the CEOs. And we spend a lot of time building software, thinking about how can we fix this problem? How do we make it so that other people who have a stake in the company and in finance actually understand and care about the financial model that ultimately drives the business? So it's not that there's anything wrong technically with the model. It's going to be accurate most of the time. It's going to reflect the right forecast and the right actuals. But the main problem is actually disconnected with the key decisions that a company makes in a business. And when you think about a model, one of the ways that we think about what a financial model represents is it's easy to think about a financial model as just a spreadsheet, right? And it's so much more than that. As we know, if we built a model before, the way we like to think about a financial model is as a simulation. My background is in games. I used to run product Azynga. And one of the more useful analogies for what a great financial model does is it's like a game engine. So if you ever played a video game like Rollercoaster, Tycoon or Civilization, there is an underlying simulation that simulates a part of reality. It's an abstraction of reality. And what great simulations allow you to do, especially in finance, is effectively to allow you to time travel into the future to understand the decisions that you might make today and what is the impact of it in the future. So how do you build a great simulation like this? When I think about a game engine in the form of a simulation, I think they have three parts. So if you ever played a video game, you have controls, right? If it's Mario, you have the joy pad. You can go up, left, right, down, and you can make Mario jump. Then there's logic. What's going to happen when Mario moves? What's going to happen when a Koopa hits Mario? If you increase ticket prices in your roller coaster tycoon park, what's going to happen? So those are what controls are. Then there is a logic, and then there's feedback. And feedback is okay. You press the button, you've increased prices, you put a new attraction in your roller coaster tycoon park, and there is some simulation that it affects. Then you can see visually what happens, right? There's more people at your park, maybe, or Mario dies, you get some kind of feedback. Great simulations have all three of these things. And so when I think about building a great model that actually gets used and people can understand, it starts with controls. What are the actual decisions that you get to make in a business? Is it hiring? Is it a product roadmap? Is it product decisions? Is it hiring decisions? Is it a marketing campaign? This is the most useful starting point because simulations and great models aren't just to be looked at, they're to be used, right? You want to use them to refine your thinking. So one of the mantras we have at Runway is finance isn't just accounting, but in the future, right? It is also accounting. You want to understand your budgets and your forecasts and your revenue goals, but it's also about something more than that. It's about thinking. So developing intuition for the business, understanding causal drivers of it and starting with the decisions is what helps you develop that intuition and get the rest of the company bought in and understand the impact of the decisions that you make. Oops, I'll go back. Logic. So obviously, once you have your controls, you want to wire them up and see how your controls will affect the KPIs you care about. This part is relatively straightforward. But for SaaS companies and technology companies in particular, there's a really powerful tool, which is we have the ability to build technology to increase efficiency. So one of the more useful things is to identify what are the costs to scale to different things. And generally there's two types of things of scale. One is cost to scale with people, right? If you have more people, you're gonna need more GNA. You might need more SaaS. So that's one type of scaling. And the other type of scaling is the type of scaling that comes with, as you have more customers and revenue, you might need more engineers or more customer support people, or more account executives. And the wonderful thing about technology is that it is leverage. It is leverage in a real sense, right? So if you don't have a shovel, right, and you dig a hole, it's going to take a while. If you do have a shovel, you get to dig more holes or dig the same number of holes with fewer people. That is a key leverage point. And so identifying the ratios of how your different costs scale is critical because then you can model the technology impact on these ratios. So, for example, as your product gets better, your typical customer support person may be able to support 30 customers per month or 40 customers. But as the product gets better, as it becomes more PLG, as your documentation gets better, maybe that 30 or 40 becomes 50 and 60. And it has a direct impact on efficiency and profitability for account executives. As your product gets better, maybe you start with, typically you can close about 10 deals. But as enablement gets better, as you develop new AI features, that ratio increases. So understanding and modeling these leverage points, these ratios, is super important because then you can connect these ratios to the things that you do in your product and the decisions that you make your business. And of course, then once you have these controls, you have these leverage points, then you can simulate how it affects the downstream KPIs that you care about, whether that's revenue margin, profitability. This is how you make it directly visible. And you kind of repeat this cycle until your assumptions get tighter. So just like any simulation or any software simulation, you're not going to get it right the first time. It takes some iteration. And what you want to do is, as you execute, you're going to see where you're correct and where you're off. Which assumptions you thought was going to be X and it ended up being Y, which relationships you thought drove retention or conversion or new AR and actually didn't. And then you can adjust the logic and change your simulation over time. So, you know, one of the examples of controls that we have. So I'm kind of a model nerd. When I ran product at Zynga back in the day in farmville, we created dozens and dozens and Dozens of these models. But we did a really good job of exposing controls that tied directly to what we do day to day. So we might have a decision between hey, we should do a big Christmas event and if we do a big Christmas event on Christmas, it's going to lift engagement by 20% and some percentage of that 20% is going to stick around. And then we had another decision which is instead we could have improved a tutorial at the same time. And that seemed like it had a smaller effect. It's about a 2% lift in retention. But if you have a right simulation, you can get a better answer what's going to happen in the future? And as it turns out, over a six month time range that lifts revenue by 15% because the 2% retention compounds. And so when you have the right simulation and right controls, you get to make better decisions. And over time you can make better decisions in your day to day because you have develop an intuition as you use your model and play with your model sufficiently. And so when you have this simulation and you have this model, it becomes the interface into your business. You can prioritize futures based on leverage. And then over time you monitor your forecast versus actuals weekly. You adjust your model, adjust your assumptions, and then you can even adjust your hiring plans based on how effective you were at delivering a roadmap and delivering an impact. So how do you actually go about start doing this? We have a very easy, quick, three step process. First step is identify the controls and the different KPIs and leverage points of your business. Right? What are the key decisions that you actually make week to week, month to month? Define the leverage points of your business. What are the most important ratios that impact efficiency and your ability to grow in your business. And three, outline the outputs. And for SaaS, usually this is going to be margin, CAC, new AR, and you connect all three dots. Once you have that, you connect it to real data. So everything that happens in a company eventually falls to the bottom line somewhere. Whether that's a new hire, whether that's a new customer, whether that's a new product feature, eventually it does something to the business. And usually the things that you do are mapped to sources of data. And by connecting the sources of data, you have a starting point, you have actuals, and you have a basis to compare. As you execute over time, you define your controls and your logic and then you build the actual leverage ratios and you manage towards them. And step three is you make your feedback a habit. So on a weekly basis now you review Your actuals against your forecast. When you miss or you over perform, identify why. You can drill in to understand the causal controls and see if it's a logic that was wrong or the assumptions that were wrong. So what this looks like on the other end of it is the following, which is before, and we've seen this a lot of companies before, every department has their own model. The sales department has their own model, the marketing department has their own model. None of them really talk to each other. And if you can build a useful simulation with the right controls, what you'll find is that you can have everyone use the same simulation, same source of truth, and now everyone's looking at and understanding the impact decisions you make using the same framework. So trade offs become easier and it's easier to align. The second thing that happens is, you know, imagine you're in a meeting and someone asks, okay, what if we push back the roadmap by three months or we pull ahead of marketing plan by a month. And usually that takes days. But if you have the right controls modeled, you can answer them, live in a meeting and just say, here is what's going to happen. And that is a great way to get promoted usually. And final thing is the people that don't care about your model now cares, particularly the CEO. The model becomes, instead of the spreadsheet, it becomes a single source of truth for your company actually thinks. And now the whole company can play the game. That's it. Thanks for coming and I'm ready for questions if you have any, Any questions. Yeah, you can come down to the middle if you want. One. One, Yep.

How do you, how do you handle like especially like in technology investor environment? Yeah, you have very aggressive models and very aggressive goals. And then that's, then of course, you know, you still have reality that you're living in. So you want to think about, okay, what are the like, the realistic expectations? So just kind of like how do you, how do you handle that balance? Because I found myself at times having to keep up with like two or three different models that are like, okay, CEO for investor actually running the business.

Yeah, version control is a big issue. And so typically you will end up with a couple of models. But just like when you play a video game, there's like different possible futures. Right. You can put in different assumptions and you can maintain a, you know, base case, a high case, a low case, and you get a range of possibilities and you know, not shilling Runway particularly, but there's different platforms that you can use to make that version Control and scenario management a lot easier. The problem with. And I've actually, this is the reason why I started a company in 2020. I was running a company called Sandbox VR and then Covid hit and so we were literally being triaged by Andreessen Horowitz on who's going to like get the emergency funding because no one else was going to fund any companies. And we had to create these scenarios whether Covid was going to be three months because Elon said it was going to be over by June or a year or two years. Andreessen was the only fund or people that told us it was going to be two years. And they were spot on. We ended up with these like dozen spreadsheets and when we update one, it won't update the other. And so having good version control is actually really important to keep it clean. And there's tools for this. I saw another hand raised. Hello.

Hi.

Hi.

Typically, how long does it take to create a model and see the outcome?

Yeah. Do you already have a model?

No, not really. So what you described, with everybody doing their own kind of thing, siloed. That's what we have.

Yeah.

Getting it all together. And typically, how long does that take?

Yeah, I mean, we've seen people get started even before the first kickoff meeting. So we have Y Glove support. But, you know, it's a version. What we build is a version of notion, basically that does math and connect to your data. And so having a consumer grade experience is really important. But I would say on average it's about a two to six week implementation, depending on the complexity of business. Yep. Yes, sir. How do you handle transparency of your financial data to the rest of the company? That's a great question. So some context. This is my first B2B SaaS company of any kind. My entire career has been in consumer software and there's so much to learn about B2B and I really enjoyed it. But one of the more useful things is being able to take learnings from the consumer world to the enterprise world. And so one of the stories I think about is, I think it was around 2005, 2006 or so when Facebook was just getting started and they did a newsfeed and their main competitor was MySpace, which was about 10 times bigger at a time. And they famously observed that the more control they gave users over who got to see what, the more users would share. Right. So MySpace is like, everything's public. People will just post GIFs or whatever. Facebook at a time when they had a newsfeed they let it gave you very fine grain grain control. So you can share just with specific groups of people or just your friends or the entire world, right? And the more fine grained that control is, what they found in their data is that more people share it. And so if you think about a spreadsheet, right, it's kind of all or nothing. Whether it's Google sheets or Excel file, you have to share it all because all calculations on both those products are done on the front end. And so what's really important is we'll have very fine grained controls, right? So for example, even if you wanted to share an entire spreadsheet, you can't because you can't just hide or encrypt a single column called salaries, right? So what we do is we give you very fine grained control whether it's column based or row based. So you can say I want people to see everything except for salaries. And what we found is that if we give people the control over how much of every anyone in the company gets a C, people end up sharing a lot more. So one interesting thing that we've seen is, for example kit.com used to be called ConvertKit. It's one of our customers. And what they did after they got started is they did it all hands or to give gave everyone access to Runway the entire instance because they were able high salaries. And they did a scavenger hunt because they implanted some kind of bug in the model and broke the company into teams to find the bug. And everyone was successful. And so having extremely fine grained controls is how you do it. Some companies might not want to share anything, some companies might be completely transparent and having the ability to control where you want to be is really important. All right, thanks for coming. Really appreciate you guys.

Sam, It, It's. It, It.