The Best of Kellblog: 10 Non-Obvious Things about Scaling SaaS with David Kellogg

Thank you. Thanks for joining our repeat session. So we did it at 9 o'. Clock, so I've done it once now, so you get to see the more rehearsed version. Let's jump in. So I'm Dave. I'm CEO of Host Analytics. If you happen to be in finance and work in planning or budgeting, give us a call. Prior to host, I was at Salesforce for a year, ran MarkLogic for six years from 0 to 80 million, and was CMO of business objects from 30 million to a billion. Quite a nice ride there, board wise. I'm on two boards right now, Elation, which is data catalogs, and Nuxeo, which is next generation Content management. Prior to that I was on the board of Granular and Astrodata. And in my copious spare time, I do write a blog that hopefully you'll want to see after the presentation. And a lot of the points, certainly not all, but a lot of the points in the slide, there is a blog post that really drills into the topic. So our intent today, and it was the guys at Saster who came up with the title, which I like because I like the non obvious. And they said, hey, cover 10 non obvious things about scaling a SaaS company. And I was like, well that'll be fun. And this guy yesterday tweeted like, how are we going to do that in 30 minutes? And the answer is with great difficulty and with a lot of speed. So my intent for the presentation, it's really kind of tops of waves. It's 10 things that hopefully make you think it's more thought provoking than question answering. And that's really my goal. If you walk away with a few things going, God, I didn't think about it that way before, maybe I'll think differently then. At least I'll be happy with what we've accomplished today. So the first one and you could accuse me of like, whoa, isn't this a presentation on non obvious things about SaaS companies? Why are you saying you should run the company about ARR? And the answer is, lots of people seem to think they do run their companies in ARR, but they don't. Just last night I met with a very smart entrepreneur, very cool company. He wanted to show me his financials and the first line was all about acv. That was mixing new acv, upsell ACV and renewals acv. I couldn't make heads or tails of it. Another time, people, they show me GAAP revenues, I'm like, you're a SaaS company. Why is your first slide gap revenue. Right. So if everybody thinks they want to be ARR first, but you only know you are based on the answers to these three questions. One, if you ask somebody how big a deal is, do they answer in ARR? If they answer in any other unit, your company's not ARR first. If I ask what the forecast for the quarter is, if you don't answer an ARR, it's about what I don't say. If I just say what's the forecast? Do you answer in bookings? What do you answer in? If you answer in ARR, that's correct. Right. Then your ARR first, what's the first line of the first slide of your investor deck or board deck. And if it isn't ARR, then you're not ARR first. So lots of people think they do this. It's actually a dangerous form of denial in my mind. But if you want to be ARR first, make this the first slide of your board deck or of your investor deck. And it's not only my opinion. I met a couple years back with Doug Leone, top guy at Sequoia. I showed him some financials in this format and he goes, Every SaaS company should present their financials in this format. And so it's not just me have a second opinion on it. And what this shows you is what I call leaky bucket analysis. Because a SaaS company is a leaky bucket full of ARR every quarter, sales pours more water in the bucket. Every quarter, customer success tries to keep water from leaking out of the bucket. Right. And therefore you can say we started a period with 100 units of ARR. We added 50 units, we churned away 10, therefore we ended with 140 units. And I can tell you that the value of your SaaS company is determined by the water level in the bucket and by how fast that's growing. So when you do this analysis, it's very important. You can see things like people who are say they sell 2 million of new ARR but churn 1 5. Say it's a big bucket and they're having trouble keeping it full. Well, that company's not worth very much, right? Because it's not growing. The water level in the bucket. It's got a big bucket, but it's not growing. So that's why we have the three growth rates down below new ARR growth, ending ARR growth, which really is the single most important thing for valuing your company and net new ARR growth. How much incremental did you do to the bucket. So everybody agrees we should be ARR first. The question is, are you? And if you don't, if your default unit isn't ARR and if you don't show your financials with an immediate focus on the ARR bucket and what's happening to it, I'm going to argue you're not ARR first. And Lord knows, please don't show up with a gap P and L. Right? These things will hurt you when you try to raise money. Right, because the VC wants to know the that you get ARR and you are an ARR first person. That's our first point. Second point. SAS metrics are a lot more complicated than meets the eye. A lot of us are kind of intellectually lazy and we want to assume these things are simple and you can get yourself in trouble if you don't take time to think about what goes into these calculations. So VCs are going to be looking at your company through a lens of these metrics. They're going to look at LTV to cac, they're going to look at cac, they're going to look at churn. And if you don't understand and kind of have a logical argument for how you calculate them, you could get yourself in trouble. Now, the purpose of this presentation isn't to try and do a two hour course in subtleties of SAS metrics, but I do want to just point out a few things that maybe you didn't think of that will hopefully make you realize I should dig into this. For example, look, the COC ratio seems simple. Prior quarter sales and marketing divided by new ARR in the current quarter. What could be easier? Well, does that new ARR include upsell or not? Most people would say yes. But some people may want to make a new business focused CAC that didn't Is sales and marketing. All of sales and marketing is just a variable like the sales commissions and the marketing demand gen. Or is it the entire base salary, the PR team, the whole banana? For most people it should be the whole banana, right? Because the question we're trying to answer is how much does your company spend to acquire a dollar of ARR? That's what the CAC ratio is, right? So typically it's all sales and marketing costs divided by all new ARR in the quarter. So should it include upsell? Yes, but who does upsell at your company? If upsell is done by sales, then wait a minute. If you ask finance, where do they put the cost of customer success? It's in sales, right? So you need to pull that out and say, wait a minute, my sales team does upsell. So therefore when I do my CAC ratio, I shouldn't include the cost of renewals because my customer success team just does renewals in that or vice versa. If your customer success team does do a lot of upsell and you're including upsell ARR in the denominator, then yes, you should include their cost right in the sales calculation because most finance departments will classify customer success as sales expense. The other nasty thing about sales expense, and this is probably for slightly bigger companies, but commissions get amortized. So there is a question of whether or not you want to restate this on a cash basis. Just say, wait a minute, if I've got screwed up comp plans that are over rewarding sales, I'm going to take those and divide them over the duration of the contract. And it's going to be very hard for me to see this with a broken CAC ratio. So if I reset and do this on a cash basis, maybe I'll see reality more clearly. So these are just the little kind of niggling subtleties that make something that seems so easy difficult. LTV to CAC. I mean, first, LTV to CAC is kind of the ultimate SaaS metric. It's what you paid for something on the bottom and on the top what it's worth, right? It's kind of the grand balance. And what you paid for it is about the CAC ratio or how much you pay for a dollar of arrangement. And on the top you're trying to say what's somebody's lifetime value in terms of subscription margin? So I have to run the service. So let's do this on a gross margin basis. Subscription, gross margin basis. Let's invert the CAC rate to get the lifetime and then divide it by the CAC ratio and that will get me a number. The bigger that number, the more it's worth relative to what I'm paying for it. It seems like a pretty simple metric, but the first question is which CAC? I just talked to you about five different CACs you could do. So which CAC ratio do you use? And I can make the same argument on churn. Which churn do you use? Because there's a lot of different ways to think about churn as well. And Lord knows, what do you do when churn becomes negative? Because the idea behind churn rates is kind of like you use it as a discount rate and a net present value formula, right? If I have this bucket of ARR. And imagine I have no upsell, and every year 10% of it leaks out. Well, it starts out with 100, then it's 90, then it's 81, then it's 72. Right. And I'm kind of using it as a discount rate on the value of my annuity. Well, that works really good except when CHURN is negative. And then all of a sudden your lifetime value is infinite. And you can go tell a VC that, but they're not going to believe you. So what do you do? And if you exclude upsell, it seems too pessimistic because wait a minute, we really do do upsell. I really do have negative churn, but it breaks your LTV to CAC math. So these are the things you need to think about. Here's an even more interesting case which is most of the thinking about land and expand in Silicon Valley is I sell somebody a $50 steak and then I sell them a $10 glass of wine, and if I'm lucky, another $10 glass of wine, right? That I have the big purchase up front and I'm getting these little things on the back end. Well, I met a guy last night who sells a $10 glass of wine and then sells a $50 steak, then another $50 steak, then another $50 steak, right? So he has a problem that looks like this. If he spends 100 units to get a customer worth 50 on that initial purchase. Well, it looks like his CAC ratio is 2.0, which is not good. A lot of VCs will run away if you tell him your CAC is 2.0. Or is his CAC 33? Because I can tell you reliably that two quarters later, these guys who I spent 100 to get are now worth 150. So what is my CAC ratio here? And the question is really, what do you mean by acquisition? Right. Because in this guy's case, if he can reliably prove that that pattern is going to happen, the then I think he should count the sales and marketing divided by, let's just say the first year customer value, which is an unusual way of looking at it. But if you have a small land, big expand model, maybe it's the right way. So here's another example. This is just a CHURN example, my last metrics example, to talk about how things can get confusing. But we'll take a simple case. Somebody buys 90 units of product A and 10 units of B. Six months into it, they buy 15 more units of A. They're happy and then the renewal comes up, well, what are we renewing? Are we renewing? What's the denominator in the renewal rate? Is it 100 or 115? Right now your inner fundraiser wants to say it's 100, right? Because you want to say, hey, I renewed 115 plus units off 100. That's what your fundraising side wants to say. Your operational manager side says it's 115. I'm handing a CSM a customer worth 115 units. I want to get 115 or more units out of that customer. So there's your first question. What do you do about kind of off cycle Upsell and how do you think about that? And the answer is, you certainly want to tell the CSM, go renew 115 or more. The 15 units of upsell. Finance typically takes it when it comes. So you kind of thus far into the discussion have lost credit for it. It kind of went away because it went to offset Churn in some other period. But more subtly, on day 365, is your company going to count that as 10 units of churn and 20 units of upsell or simply as 10 units of upsell? Right. And most sales ops people, if you talk to them about this, they look at customers on a contract basis and they're going to say the value of that contract went up by 10 units. So it's 10 units of upsell. So the underlying flows get lost. And most systems don't track it. Most companies don't track them. So I call it count level Churn. You're looking at it through an account lens, not through a product lens. And what about those 15 units of upsell? How do I get credit for them? And the answer is that's why they made retention rates, because retention rates are cohort based. And you say, take all the people who bought in 1Q17, what are they worth now in 1Q18 in aggregate? And that's my retention rate. And those are good. This is simple calculation and it actually shows what happens to groups of customers and how they expand or shrink over time. So you don't have to worry so much about the timing issues because you're taking snapshots. So all this stuff though, and this is just tops of waves, hopefully points out that, gosh, there's a lot more going on here than maybe I think. And it's worth drilling into these things because when you're raising money, VCs and private equity guys, they're very familiar with these metrics and they're going to ask those questions and if you don't have credible answers that are well thought out, you kind of lose a lot of credibility. On the flip side, take my friend who sells the $10 glass of wine. Most VCs would kind of run away from him going, wow, he spent a bunch of money to get just this $10 order. But if he comes in with a good story that says, wait a minute, 8 out of 10 times, they buy 50, then buy 50 and then buy 50 again within one year, his CAC ratio goes from 2 to 0.3 and they're like, holy cow. It goes from unacceptable to amazing just based on how he was able to frame it. So this stuff matters and they do react to the metrics. Third point, public CFOs may not get SaaS metrics. This was fascinating to me and I put it up here mostly as a bonus because one day I met with the CFO of a well known public SaaS company and he started asking me all these questions about CAC and LTV and renewal rates. And I'm sitting there thinking, how can this be? This person is a CFO of a public SaaS company and he doesn't understand a lot of these metrics. And then the kind of light went off and I'm like, wait a minute. He lives in a world where investors look through a veil at the company, right? They can't see all the metrics that you show VCs and private equity guys, they look at the company through a veil. So he only understands these outside in metrics like billings, right? So billings is revenue plus change in deferred revenue. It's kind of a way to try and hack bookings. It doesn't necessarily always work, but that's what they're used to. So even though he's the CFO by virtue of his job, he kind of places himself outside the company and looks in because that's what all his investors do. And therefore he doesn't actually know much about all these internal metrics that the VC world gets to see. So first, it was kind of amazing to see, and second, I would just say, look, how this might apply to you is boards love to hire the CFO you need three years from now. And that's great as long as that person can do the job you need done today and this guy can't. This guy would not be a good CFO at a private company that needs to raise two more rounds before an ipo. So my advice is, don't let your board push you too Far ahead and hire some CFO who's going to be wonderful at the ipo when you're three years from an ipo, need to raise two rounds because you need somebody who can do the job right now. And it's a fairly common pattern and this is just one example. So number four multi year deals make sense in certain situations. This is heretical. So I kind of put the heretical warning on it. Because if you say this in polite SaaS company you are likely to offend someone because hardcore SaaS purists think it's all annual multi year deals are evil. You're counting tcv and that scares me because TCV may or may not ever be realized. Are you taking your ARR and multiplying by five for a five year contract with annual payments? Who knows if you're going to get the money? Are you inflating the size of your company 5x in the process? Scary, scary, scary, right? That's the thought process that happens if you talk about multi year deals recklessly. Now the reality is these contracts are effectively rolling the dice. If you've got a 90% churn rate, I'm effectively exposing you to an annual dice roll where your contract comes up. We roll the dice nine times out of ten it says renew. One time it says not. Right. That's the way we should think about the renewal rate. So if I were to offer you a discount that was less than that churn rate, am I better off? I could expose you to a 10% decay from the churn rate. Or I could expose you to a 5% decay from my discount. Right. So my belief is whenever the discount I offer you for a multi year deal is less than the churn rate, that you win and I win. Right? It's actually a win win transaction. So at my company we offer these to customers. We do it at the very end of the sales cycle. We say you've done a full evaluation, you've chosen host analytics. Would you like to save a little extra money? Because you seem pretty sure it's going to work for you because you just spent six months evaluating it. If you'd like to do that, we could talk to you about a multi year deal. And here's the kind of discount we can do. And about half the time people say great. About half the time they say no thanks. And I don't mind because if I'm offering the discount at the right level, I'm actually fairly neutral. Like go ahead, I'll put you against the dice. Because that's roughly What? The discount I was offering anyways, kind of mathematically equivalent to me. Now the deal needs to be prepaid, right? Very important because I believe if you're salesforce and you sign a multi year deal that's not prepaid and a customer defaults on you, you're probably going to sue them and you can collect that money. If you're a $10 million SaaS company, you're not, right? You're not gonna have the money to sue them. So I believe that multi year non prepaid deals for small SaaS companies are a terrible idea. Why? Because they look a lot like renewals that are done by finance, not customer success. Right. Because if it's on a one year cycle, you're gonna tell customer success, go love this customer, make sure they're ready for renewal. Do all your normal love process. So when that contract comes up, they're ready to renew. If you hand it to finance and say hey, there's a collection due in 365 days, they're not going to call anybody, they're not going to be nice, they're not going to treat it like a renewal, they're just going to mail an Invoice at month 11. If the customer's unhappy, they're just not going to pay it. And all of a sudden you've got effectively a renewal and finance is going to say it's a contract, you owe us the money. If the customer is not using the software, they're going to say pound sand. And then where are you? You're nowhere. You gave away a price lock, you gave away a discount and, and you got nothing in return. So our argument is if you want to play this game, the deals have to be prepaid. Multi year, non prepaid deals to me don't make a lot of sense for small companies because your ability to collect those out year payments is virtually identical whether or not the contract's in place. Because if they're using the software and they're happy, they're going to pay. If they're not, they're not. It doesn't matter what the contract says. So they can make sense in certain situations. Now another heretical idea is that bookings is not a four letter word. Again, if you say the word bookings in polite SaaS company, you're likely to offend. So we must be careful when we say the B word. But if you define bookings sensibly then you can talk about it and define sensibly. Bookings are things that turn into cash in 90 days. Everybody cares about cash. So they're going to like talking about something that turns into cash. And as long as you don't mean TCV bookings, which is not prepaid. C Prior sermon. Right. If you're talking about something you're going to get collection on soon, then people are happy to talk about these. So at our company I said about half the time somebody will do a multi year prepaid deal. So we end up on average with two years prepaid. So if you looked at a build of our financials, I'd say in a period we did 100 units of ARR, we did 200 units of what we call new subscription bookings. That's prepaid subscription bookings. Our NSB to ARR ratio is the average, effectively average prepaid contract duration in effect. Then we add renewal bookings to new subscription bookings and then that gets total subscription bookings. Then you add services bookings on top to get total bookings. And this is a great way to build this because if you're smart, you put the focus on what I call tsb, total subscription bookings. Because if you care about cash, whether it's a renewal booking or a new booking, it's still going to generate, if you have subscription gross margins of 80%, it's going to generate 80% of that money in cash for you to go spend. But that's not true for services bookings. As we'll see in a minute. Most of you, 50k of services units generates 0k of cash. So let's get that out of the way and let's focus on subscription bookings. Because if you're offsetting deficits in subscription bookings with services bookings, you may feel great about hitting your bookings target, but you're going to find yourself at a real cash shortfall because they don't have the same gross margin. So bookings is not a four letter word. If you think about bookings as something that turns into cash renewals, a lot of us feel like renewals is a proxy for customer sat, and it's not. And it's a pretty simple point, right? We lose customers who are very happy because they got acquired by a PE firm and the PE firm says here's our standard tool set, or they hired a new cfo and the CFO says we use Hyperion because I've always used Hyperion. Right. That happens in our lives. And those were happy customers, Right? On the flip side, you often have unhappy customers renewing, typically because they didn't have their act together, like they missed an auto renew 60 day thing in a contract or they were trying to get a new system up and couldn't get it done right. So you can find both cases where unhappy customers renew and and happy customers don't renew. So let's not pretend that renewals are a perfect proxy for csat. Let's go measure NPS instead and then try and look at the relationship between the two. So pretty simple point, but the net of it is go measure nps. Do it ideally four times a year with a quarter of your base each time. This one's fun. You can't analyze churn by analyzing churn. This is a slide we show in our board meetings and this is what I call the churn taxonomy. It takes all the churn that happened in the period and talks about why. So 1000 units lost to lost accounts, 300 to shrinking accounts on the lost accounts, blah blah blah. 750 was business related, 250 was product related. It's a way to show your board and to show your team kind of churn by reason. Like what are the reasons people churn? Great slide, useful slide for any board deck. But the problem is it doesn't actually answer the question you really care about because. One sec. The question you really care about is what predicts someone churning. And I'm going to use an example here, fairly well known example about World War II. As bombers came back from missions, they would put little dots for where they'd been hit on these planes. At the first order, people said, gosh, these are where our planes are getting hit. So let's go reinforce them in those spots. Let's make them stronger in the places that they get hit. And one day a very smart person came along and asked a simple question. Which was what? What about the planes that don't come back? Wait a minute. Shouldn't we worry about what separates the planes that don't come back from the planes that do come back? And maybe the planes that don't come back are the ones that get hit in the complement of this pattern. Right? They get hit in an engine or they get hit where the pilot sits. Right. This is basically survivor bias, right? We were analyzing how to make the planes stronger by only looking at planes that survived the mission, when we should have been looking at what differentiated planes that didn't survive the mission from those that did. Right? And when you analyze a bunch of customers that churned, you're doing exactly that. If you just grab here are all the customers that churned last quarter. Let's go analyze the Heck out of them. You're making this mistake. And what you want to do instead is go get a table of all the customers, those that churned and those that didn't, with a bunch of input variables. Who did the services, how big is the company, how big is the ARR, what else, what industry are they in? Those sorts of examples that describe the customers, kind of things about the customers in the first columns and the other columns are did they churn or not? Were they happy or not? And then try and figure out the relationship between these inputs and those outputs, right? To see what differentiates people who churn from those that did not. So that's what I mean by saying you can't analyze churn. By analyzing churn, you need to analyze who churned and who didn't and try to compare them. Finding your own hunter farmer model is hard. People do hunter farmer because they want to focus their reps on new business and that's a very noble intent. And they want to pay lower rates on renewals and upsell because that's less hard work than winning new business. Now something to know if you don't know it already. Boards hate paying sales type commissions on renewals and they're right to. Sales people in SaaS make maybe 10 or 15% commissions and you don't want to pay that in a renewal. And boards hate double compensation. Another word not to say in a board meeting is double comp because if you say it, everyone's going to sigh and it's going to be a bad meeting for the next 10 or 15 minutes if you ever say those words. Never say double comp in a board meeting because boards hate it. And sometimes they're right and sometimes they're wrong. In my opinion, we need to let intelligence, not dogma, drive this. The first thing you need to be aware of the dogma. There's a lot of religion. But look, if you've got a standard kind of fries with your burger cross sell, it's not really hard to sell the fries if you're buying the burger. That's what a lot of hunter farmer is. So they say the sales guy goes sell the burgers and the CSM will go sell the fries. And it's not that hard. So I can get a relatively low skilled csm. We'll say it is. So that's hunters and farmers. We'll say the band is more salesy, that it requires more sales skills that require. This is what people often call account management. So there is some salespersonship required to do that. Well, that might require a farmers with shotguns model we call it. Right. So we still have hundreds of farmers, but the farmers are armed. Right. There are more sales and profile, they make more money. They come from a sales background. Say it's a really hard cross sell like we have. We sell planning and budgeting to the FP&A people. We sell consolidations to controllers. Different buyer, different product. It is definitely not fries with your burger. Right. In that case, we for that sale will use farmer's track for hunters. So when a customer success manager finds a consolidation opportunity, we send the salesperson right back in. Right, because if you don't, if you send a CSM to try and sell consolidation to that controller, you're going to end up with a dead farmer. Right, because you put your farmer against somebody else's hunter and that's something you don't want to do. So you need to figure out how your competitor is treating these models as well. And finally, to beat the metaphor to death, if you have the kind of small land and big expand like the guy with the wine steak steak, that person actually has an inside sales team that's pretty cheap selling the wine glasses. And all their enterprise reps are install based reps. So it's a crazy model. He calls it hunters of the zoo. Right, because they're in the zoo, they're in cages and he's got these enterprise reps shooting at them. And it's a great model for that sort of business. Number nine, you don't have to lose money on professional services. I was shocked when I found this chart in the KeyBank PAT CREST study. But only 20% of SaaS companies lose money on services today. I think it was a lot more in the past. So people have kind of figured this one out on their own. But to me the right model is you have a relatively small team of high skilled, high priced people. You shoot for single digit margins and things should go pretty well. You'll set a nice price point to lead your ecosystem of partners and you won't have the problems you have if you have heavy losses. Because heavy losses cause two problems. One, they hurt your blended gross margins, which isn't that bad because a lot of people will just look at subscription. But the real problem is, and I've seen this happen, a clever PE or VC guy will look at your services losses and add them back to your cac. And that can be a very bad moment for you because you're sitting there going, hey, I've got a pretty good CAC. But I'm losing 25% margins on services. And they go, well, if it's going to cost you a 25% loss on services to sell that software, that's actually acquisition cost. And then you're like, oh, damn, bad moment. So think hard about that. If you think you can fool somebody by subsidizing a lot of sales cost and negative professional services, there's a lot of smart people out there who's going to correct for that error. Finally, last point, look, you need not sacrifice long term team members and the alter of experience talent. I'm a big believer that Silicon Valley boards are a little too quick to kind of kick out the people who help build the company and they're a little bit heavy handed with them. I think we've been in a more relatively founder friendly era for the last five or ten years. But as you face these moments where your VP of sales, who's been so good for the past two or three years is starting to run out of gas, what do you do? And the board will very quickly say, let's bring in a new one and get rid of the old one. There's no way we can keep them. And I think if you're creative, you can find a way to keep them. And Salesforce, I wasn't there that long, but they were phenomenal at this. They valued people, they valued talent, and if you scaled out of your job, they'd try and find you a new one. They try a couple times, typically because they wanted to keep you with the company. And I think VC boards are a little too quick to kind of boot people out. So I don't think you need to make these sacrifices. These boards are correct in some ways, incorrect in others. I think the thing you can do to your company is create a culture of change, not a culture of stasis. Because if everything's the same every year, it's very hard to take one person out and move them down. If every year you're kind of reorganized and say you try this, you try that, you try this, then it's much, much easier to do that and you can get to a place where you can get everyone in your company to understand that they're going to have more fun and make more money. If they worry about being on the right bus, that is your company, let's worry about getting on the right bus. That's how you make money, that's how you have fun. It's not all about the seat, it's all about the bus. So if I want the VP of North American sales to run Central Region sales for a year, and he or she doesn't like that, Just do it. In a year from now. We'll see what happens. Who knows what happens? I don't know. But you'll still be on the right bus. You'll still be vesting your options. And if, if you do great in that role and love it, great. And if you don't, let's talk in a year, because we're going to change everything anyway. And it's a great way to retain your people, which is really the most important thing you can do as you build a startup. So thanks very much. Follow me at kelblog and check out kelblog.com thank you.