May 25, 2016
The Evolution of Dark Launching on Snap CI
Badri Janakiraman shares the story of the evolution of dark launching on Snap CI
Beyond true product-market fit, one of the few things you can do to grow revenue and enterprise traction is to hire strong sales pros. But most teams struggle to recruit and retain sales leaders with the right experience, drive, and talent to build the rocket ship mid-flight.
Pete Kazanjy: My name is Pete Kazanjy, I’m the founder of a company called Atrium. Why should you listen to me with respect to sales and go-to-market? I don’t have a background in sales, or at least I didn’t have a background in sales. My background is actually in product marketing and product management.
I worked at VMware working on the infrastructure team within the desktop virtualization team before starting a recruiting software company called TalentBin as a business generalist founder, at which point I had to learn how to sell software or else the company was going to die and go-to go out of business. It was a pretty good motivation to figure out how to do selling. That was about six years ago or so, I went from our first sales rep to sales manager, VP of sales.
Eventually the organization was bought by Monster Worldwide. Not the headphone company or the power drink company, but the old, dying recruiting job board where I lead new product sales for 600 really old school sales reps. That’s my background in going from being a founder, a business generalist, technical enough to talk at dinner parties about things, person to go-to-market leader.
And since then I run the nation’s largest sales operations and sales leadership community called Modern Sales, a sister community called Modern SaaS, and I wrote a book on sales for founders called Founding Sales. It’s essentially the book that I wish had already existed when I was doing a really bad job of figuring out how to do go-to-market myself. So, that’s why you should listen to me.
Dan Portillo: Dan Portillo, I’m a partner at Greylock and I’ve been there for seven years. I work very closely with our early stage companies, a lot of early team building and hiring the first of everything. Early engineers, early product managers, early salespeople. Started my career in ’99 working for a number of startups, and then I joined Mozilla when it was 20 people.
Part of the reason why I loved my time there, and I was there for five and a half years, was that we generated a lot of money and we had no salespeople. My friends were all salespeople, but I loved having a company that was predominantly all engineers. Then moving to Greylock and working with our companies and helping them to build a business, really learn the appreciation of go-to-market and how much that really matters.
It’s not just a wonderful product, but actually having a team that’s able to sell it and acquire customers is incredibly important. So, to just start off. Pete, will just hiring salespeople completely destroy the culture of your company?
Pete: No. Running out of money and not being able to raise your next round will definitely destroy the culture of your company. To the extent that you are able to hire and onboard people will that will help you scale your organization and continue to grow the organization is a good thing. Now, doing that in a way that is good for your organization, that’s a whole different topic.
A lot of founders have this misconception around go-to-market and salespeople in general that causes them to resist doing that, and that’s definitely an anti-pattern.
A lot of organizations like to beat their chest about the fact that they don’t have salespeople, like Atlassian or GitHub, or whatever. It’s total bullshit. If you do LinkedIn queries in their organizations they have tons of salespeople in those organizations.
Palantir has tons of salespeople, they just call them business development. And having that antipathy to somebody who is specialized and focused on revenue acquisition is something that is bad for your organization. That is an unequivocal truth with respect to SaaS companies.
Dan: I’ve seen this a lot with our early stage companies that have incredibly technical founders. There is a reluctance to bring in someone with a good market background. I’ve seen it repeatedly that a couple companies pay for it, in which it’s taken them much longer to get a product at market because they were building and taking in a lot of feedback and not necessarily understanding what people are willing to pay for and where the value lies, and the importance of getting something to market earlier.
When you’re really strong technically you tend to lead with that skill, and so sometimes you have to learn how to step back from that.
I remember in the early days of building Atrium you were doing a lot of early pre-customer work ahead of doing– and I’m wondering if you’d share some of that?
Pete: Totally. We’ll talk about two things. I can talk about some of the early customer development that we did at Atrium, and then maybe what I’ll talk about first is this customer interaction antipathy that you oftentimes see in organizations that think that you can just publish API documents and people will magically buy.
Has anyone heard of David Skok? OK, look him up. He is a partner at Matrix and he does a lot of really good writing on developer sales motions, specifically because he was an early investor in a company called JBoss which was an early application app server that beat ass on web logic and Oracle back in the day.
The way that they were able to do that is not because people would just download the software and use it and give them money, but it created this massive inbound funnel of people who were downloading the open source software and using it that was lead gen. That wasn’t the business itself, that was lead gen for the eventual business that was an enterprise business of higher ASP.
The thing to realize is that it is highly unlikely that you are going to get to enterprise scale on $10 or $50 a month.
Probably the only examples of that are that Dropbox’s average order value right now is $100 or maybe $150 dollars a year per customer as compared to Box. The thing to think about when you’re looking at that and saying, “Dropbox doesn’t have a really big sales organization and it just magically sold itself.” File sharing is a really huge market.
Everybody needs file sharing. Don’t accidentally pattern match that and think that you’re going to be able to get your business to that scale at $150 dollars per customer per year, because the economics don’t work. However, using that as a lead generation funnel like New Relic, for instance. They have a huge inbound lead generation funnel which then is used to engender sales conversations. That’s one thing to think about with respect to not mistaking or thinking that you don’t necessarily have to talk to customers because the product is going to be able to sell itself.
Yes, people will be able to download it themselves but it’s not going to be able to sell itself.
The second thing you wanted me to touch on was the customer development stuff. Early on, some of the things you can do early on that will set you up for success when you’re actually scaling your go-to market and as you’re doing your customer development part of your product development.
One of the things that we did, people are familiar with first round review and there is a really great article by a gentleman named Michael Sippey who’s the former VP of product from Twitter. It’s called Get in the Van, I highly recommend it to anybody who’s interested in customer development. It talks about doing customer interviews in a very structured fashion.
At Atrium for instance, we did 100+ or 150+ customer interviews when we were first starting out that covered the entire matrix of our user base. Sales operations people, VPs of sales, VPS of customer success, VPs of engineering etc. We were talking to them about how they use data to instrument the performance of their teams.
But in the process of doing that, in order to inform our product development in addition to recording all the insights, and recording everything and getting it transcribed, we’re also picking off people who were the most engaged and thoughtful.
Because very quickly in an interview you can figure out whether or not someone has their head up their ass, that they’re not interested in talking to you. Or if they are actually really engaged in the problem, and is thoughtful.
Those folks you can grab them and put them in your pocket for later and the traditional way to do that is with a customer advisory board. That’s really helpful because in addition to them being engaged and stoked on your success, there are also folks who can be your early customers and your early enterprise customers.
That’s actually been something that’s been very successful for us. It’s something that is a little bit more hands on than your traditional developer sales motion of, “OK. We’re just going to have a shitload of inbound and we’re going to dig through all the chaff and try to figure out where the needles are in there.”
But I don’t think that they’re mutually exclusive behaviors, you can have a big inbound funnel to figure out and that allows you to see, “OK. It looks like we’re getting a lot of people from Microsoft or LockKey. That’s interesting.” We had no idea.
Edith from LaunchDarkly and Dan and I were talking about this earlier, where she had an initial hypothesis that LaunchDarkly was going to be something that people will be able to slap down a credit card for, and then early on they started seeing that there were larger enterprise companies that were interested in talking to them.
So she got a piece of information from that and I don’t think they’re mutually exclusive. The notion of a large inbound funnel as compared to intentional customer development and customer advisory board go-to-market, they’re not necessarily mutually exclusive.
Dan: I see a lot of early stage startups, they’re looking to build their early go-to-market team and they’re looking to hire reps out of kind of name brand companies like Salesforce. Why is that potentially a mistake?
Pete: There’s two things to consider. First, when you hire your first sales professionals is really contingent on how developed your organization’s sales motion is. If you’re hiring salespeople with an expectation that they’re going to figure out whether or not the product fits the market, that’s a really bad approach. That’s a totally anti-pattern.
In a deck that I present pretty frequently I call that, “Sprinkling some sales on it,” like, “I’m going to hire a sales bro or a sales bro-ette in order to figure out and to do initial go-to-market for me. That’s a really great way of wasting a lot of time and energy. So that’s the first thing.
Don’t hire sales before you’ve proven that you can exchange money for value. That someone will give you money in exchange for using your software and they’ll get value out of it at a statistically significant amount.
What does statistically significant mean? It’s probably different depending on the size of the transaction. Not to speak on Edith, but we’re homies. In the case of LaunchDarkly after Edith sold the first half dozen or dozen $50,000 contracts, at that point you could probably get a salesperson in there to then go and replicate that.
Then once you do that and you say, “Great.” One of the things that is a helpful metaphor for go-to-market is actually software development, insofar as early on what you’re doing is you’re developing this process by which you sell the solution.
The term of art for that is a “sales motion.” The way to think about it is essentially just the software, like the sales software that you’re developing. Early on you’re iterating in that and you’re iterating and there’s bugs in it. And then when you get it to the point where it executes on your local, then what you want to do is clone that off and shove it into other reps.
Then after you do that or after you’ve gotten it to the point where it’s ready to go to another salesperson, the profile of a person that you’re looking for probably needs to be aligned with the software that you’re selling. A good example of this would be, we were talking about application performance monitoring earlier.
New Relic’s ASP for most of their existence up through their IPO was only around $5,000, $5-10 thousand, because they had this huge inbound funnel, installed the New Relic agent, get a funny T-shirt. “The data is flowing. OK. Do you actually want to see more insights from it? Great. Slap down your credit card,” etc. As compared to AppDynamics who also did application performance monitoring, but their ASPs were north of $100,000, etc.
Dan: And their churn was drastically less.
Pete: Yeah, exactly. Similar products but different approaches. In the case of AppDynamics, because they started out primarily focused on Java?
Pete: .NET. OK. So they started to focus on .NET which is very enterprise-y. You have a lot of organizations with developers in the middle of the country that need APM. But these are big ass organizations as opposed to sexy startups that are running on Ruby or Python, where the size of the organization is much smaller and the average sales price is going to be much smaller.
If you have those two different sales motions, the person you want to hire for the AppD sales motion is not a very transactional person out of, I don’t know. What would be a good example here? Like Xamarin, or something like very low ASP where there’s a lot of inbound.
This is going to be the app that AMX people came out of, computing associates. This is going to be an Oracle sales rep, or a high end database person who is used to long sales cycles, relationship-based selling, etc. Versus the New Relic salesperson is somebody who’s more transactional, is used to managing 50, 100, etc. concurrent transactions at once.
Understanding that the way to think about salespeople is not dissimilar from thinking about engineers, where you would never expect to hire a backend engineer who works in Java to work on all of your sexy front end features. It would be ridiculous. You should think about it the same way when hiring salespeople, you want them to have the skill profile that aligns with your sales motion.
Dan: The same thing is true for recruiters. I’ve hired a lot of early recruiters and companies, and interviewed a lot of recruiters from Google and other places. Usually when you have a product that everyone’s already heard of, it’s very different when you’re selling something that no one’s ever heard of before. You need people that can really articulate value and there’s a subset of people that help build a brand.
There’s people that work within that brand and I think have a very good friend that worked with me at a very early stage startup and he was awful. He couldn’t tell anything, and then went to the Salesforce and is now doing amazingly well there because there’s collateral and lead gen and there’s a very well understood go-to-market motion.
If you don’t have that, it’s a very different type of person that is comfortable walking a tightrope without a net.
Being able to understand that when you’re trying to build a brand, having someone that’s coming from a brand is used to having all of the tailwinds of being able to do that.
Pete: That’s really great. There’s the sales motion fit, which is what I was describing earlier, and then there’s the stage fit. A lot of times people mistake and they think about a brand as being monolithic. Like, “That’s a New Relic salesperson,” or, “That’s an AppDynamics salesperson,” or– I’m trying to think of another dev tools company but I work in sales ops and marketing ops so I’m having a hard time. As opposed to thinking or saying, “When were they a salesperson there? Were they one of the first 10 or 20 salespeople there, or were they the 500th salesperson?”
Dan: The markets changed quite a bit. There used to be, back in the day there was a company called PTC. PTC was known for having awesome salespeople, and a lot of them went to Bladelogic, and Bladelogic had a really great sales team. But now people change jobs so quickly that there’s not as much of an investment in training as there used to be.
There’s not a lot of good places where. “If someone worked there, they definitely are a good salesperson.” It takes a little bit more digging. What are some things to look for in someone that might be potentially outstanding?
Pete: The most important thing to do there is to do job simulation. The same way, it’s actually funny now that I’m thinking this through. A lot of these answers are not dissimilar from engineering answers, where when you’re doing engineering hiring you’re doing portfolio work, you’re doing code tests.
You’re doing whiteboard coding, things like that and essentially just simulating the way that the job would work. Otherwise whiteboard coding is not exactly how your job would work. You do the same thing with salespeople. Look at their portfolio as indicated by their historical performance, but also have them run through the actual processes themselves.
Dan: Sell me this pen.
Pete: Not the pen. That’s like, terrible. It’s more like, one of the things that I always like to do with salespeople is have them sell me their solution. I don’t really think it’s fair to have them try to sell you your solution, or at least it’s higher risk because them getting intimate with your solution is probably going to be a process over time.
Whereas they should be super intimate with their existing solution and if you have your act together as a hirer or as an entrepreneur, you should at least be able to consume their website and their marketing collateral in order to play a fairly reasonable customer. So have them run you through a sales cycle.
Dan: I also throughout the years kind of look for people that drastically outperform their peer group. If you get a sense of them within their peer group, are they 1.5 times better? I have a recruiter that I hired who used to work at Google and he was on the SRE recruiting team. The quota for SRE was 2.25, the average was 1, and he was doing 10 to 15. He was 10 to 15 times more productive than the average person on his team.
So, looking for people that are productive but that’s also not sufficient. You need to understand the circumstances of that productivity. Like, “Why were you so productive? Were you getting good leads? Was it one customer that drove your revenue number?”
Helping them understand why they’re that much more productive than others in the same cohort.
But my general pattern of looking for people that outperform people at their level, or have a job a level above where they should for their ease of experience has been a good indicator for me of figuring out what to look for. Given the network and the set of companies and the audience, what do you think? Does sales ops play a different role in companies that are more developer oriented, or that have a user adoption model?
Pete: Folks who are familiar with sales operations, sales operations is not dissimilar to dev ops insofar as it’s the function that makes sales run more smoothly. I like to think about it as the part of the organization that is the product management function for the sales org itself. It focuses on removing friction, analytics and instrumenting, the performance of the organization.
So when you need somebody fully full time focused on that, as opposed to the slice of the sales leader or founder that is focused on that, is oftentimes contingent on your go-to-market. If your ASPs average selling price is super high and you have a long sale cycle and you’re talking about– Use MemSQL as an example. Really big ASPs selling into large IT organizations and former Oracle sales reps that have $2 or $3 million quotas and they retire with 5 to 10 deals in a year.
You don’t really need a lot of CRM excellence in order to execute that because they’re only probably managing 20 or 30 deals concurrently. The level of sales operations excellence that is required for something like that is probably lower than a company like Xamarin that had a massive fire or like Snowflake. Snowflake is a better example because it’s in the same space as MemSQL. Because anybody can light up a Snowflake instance and start sending data to it and using it and their organization, they have this massive inbound funnel of users.
Differentiating between the ones that are students just goofing around with Snowflake versus a developer that works at Lockheed Martin is a really important thing. Also their ASPs are much lower than MemSQL, so if you have an inside sales team in San Mateo of like 15 mid-market salespeople each of whom are managing concurrently 50 or 100 deals, being more CRM excellent and making sure that the reps are on top of their deals. Not losing track of things becomes way more important. It’s really contingent on the sales motion.
So if you do have, let’s use Xamarin as an example. Xamarin was bought by Microsoft. What would we call them? I guess they’re like mobile development framework. Massive inbound funnel that was focused. They didn’t really have an outbound motion at all, they just had this fire hose of inbound and they had a really gnarly sales operations function that was focused on facilitating that inbound funnel.
Differentiating between the crap and people who were potentially legit, getting them access to the materials that they needed in order to have success, and then instrumenting their utilization in order to understand them.
“OK. Cool. In addition to this individual right here being highly successful with our product, also they happen to work for an organization that has 1,000 developers in that organization. Cool. We should attack that proactively.”
So they had a really heavy investment in sales operations because it was very important for their go-to-market as distinct from a MemSQL or, I’m trying to think of another. Maybe like a Rainforest QA where the ASPs are higher. It really depends.
So just being mindful of, “What’s the velocity, what’s the complexity of your sales motion?” That might pull forward the point at which you want to have more sales operations investment, as opposed to just running it out of a spreadsheet or out of your brain because you’re only managing 15 deals each of which could be worth $250,000.
Dan: Helping companies make the transition so that you have a really good funnel, you have good leads, you’re able to transition those leads and you’re getting the push to move upmarket. How do you think about the consideration of taking larger customers that might require different product features or require entirely different selling motion, and thinking about that? Having that calculus for founders?
Pete: Yeah. There’s very much a Goldilocks scenario where on the one hand you don’t want to tell yourself a story that you just publish documents or you publish your product. And people are going to buy it, and that’s going to be enough to build your business.
We already crapped on that idea earlier in this conversation. At the same time, closing really large enterprise deals can be dangerous for early stage organizations for a number of reasons.
One, they have outsized expectations for the delivery of product to them. Not from an SLA and uptime standpoint, but they’re very needy from an integration standpoint. Moreover, if 30% of your revenue is one customer, are you a product company or are you a professional services company for this one customer?
And if they know that and they can say, “Actually we want this feature over here,” that isn’t aligned with your roadmap, and by road map I mean the things that are going to create value for the thousands of other customers that you eventually want to sell to. You just have to be careful of being pulled in that direction.
I know this is a little bit of the product management thing I’m talking about right now, but you put yourself in that situation of having to make those product management tradeoffs if you’re going and prematurely selling to the enterprise.
Early on, I feel that for most organizations selling to the mid-market is probably the right solution.
There are exceptions to this, like there’s a Greylock company called Blend that sells mortgage application processing automation software. Super sexy. But I think the five or six biggest mortgage processing organizations in the United States use something like 30 or 40% of the market, and going and selling to them and building the product for them enabled them to have a solution that would handle the longtail mid-market.
So there definitely are exceptions there, but prematurely going up market, a lot of times what ends up happening is VCs will push on an organization to say, “I want to see your ASPs go up.” Or, sorry. “I want to see your revenue go up,” and there’s two ways you can make revenue go up. You can sell more deals, or you can sell bigger deals. If you’re not sales process excellent, then potentially an easier way of ramping your revenue is to just try to sell bigger and bigger and bigger deals, as opposed to selling more deals by scaling out the number of salespeople that you have.
That right there, scaling out the number of salespeople you have. Again, use the server metaphor. Just stamping out more job boxes and then putting the software on that such that you can sell more, is usually the safer, lower-risk repeatable approach. In general I recommend that people be skeptical of going up market until you actually have that mid-market motion going very well.
At which point, when you move into the enterprise you recognize that it actually is a different segment of buyer, who will have different product requirements and different purchasing requirements. Almost treat it as a new vertical as opposed to thinking, “The same way that I sold to SMB or mid-market is totally going to work here.” So, that’s my position on premature enterprise selling.
Or the metaphor I like to use is there’s rabbits, there’s deers, and elephants. Rabbits are fast. If you end up killing one there’s not really all that much to make a meal. Deer is kind of nice because you take one down you can eat for a few days on it. Elephant if you actually take it down it might fall on you and kill you, and then moreover if you do take it down now you’ve got to eat the goddamn thing and it’s so huge, and so on and so forth.
Dan: Some of the things that we, in sitting through the partner meetings, that some of the mid-sized companies are not that much harder to close than the enterprise customers.
You see that there’s a lot of effort that goes into understanding what makes a good customer, and the amount of effort it takes to close it, figuring out if the juice is worth the squeeze for that particular customer. Because if you’re going to have an intensive sales motion you have to make sure it’s worthwhile.
Pete: Yeah. That’s a counter argument to my argument. I was complaining to Dan earlier that I sold a $40,000 deal recently and it took like 10 or 15 meetings. And we were like, “That’s so much work to close a $40,000 deal.” But the point remains, if you’re going to have a sales process that takes 10 meetings to close, would you like $40,000 out of that, or would you like $200,000 out of it?
Now the question of course is, do you have the product that can service the $200,000 organization? As with most of these questions, it depends. I think first and foremost, the question was how do you approach a sales hire? Do you target somebody who is more junior, takes less capital? Do you target somebody who’s more senior who can put a program in place, etc?
The model and the design pattern that I like is assuming that the founder has, he or she has sold a sufficient amount of deals to start out with. I prefer to prove that you can bring on a more junior person to replicate that, to prove out that this can be executed on another human that is not the founder. And the reason why I like that design pattern, this again depends.
If you’re selling $500,000 ASP deals where the only way you can get in the door is if you have an amazing suit and you’ve sold to the buyers previously, then this recommendation is stupid. But if you’re like most of the organizations, like most SaaS organizations it’s selling into the mid-market with a new disruptive product, then usually this pattern works. First of all there’s more of those humans. Once you prove that with the first one, you can get a second one, a third one, a fourth one, etc.
The other thing too is hiring that more senior person, and by senior person I don’t think you mean a senior sales rep, I think you mean a VP of sales. The problem with that person frequently is that they’re not a do-er, they’re a manager and what you want right now is you want another seller, or there’s an intermediary there. This is a design pattern that I like for a first sales manager hire, or a first seller/sales manager hire is either a sales manager, or a high potential sales manager who hasn’t been out of the trenches for so long that he or she can’t do a demo.
Or it’s like, “I don’t want to do 15 meetings a week, that sounds like work.” That sort of thing. You don’t want to have that person, you want to have someone who’s like, “Yeah I’m used to being on calls with my reps all the time and being essentially an adjunct seller, or it hasn’t been that long since I’ve sold myself. But I have the upside optionality to manage people such that when we actually know that we can clone this out repeatedly,” he or she will be able to hire those incremental people and manage those.
I have a tendency to prefer towards the prior pattern, but it really is contingent on if the founder has sold those deals themselves previously, and has some sort of capacity to hire and onboard that individual. Again, David Skok and I were talking about this the other day. That’s a hard prescription to make if you have a solo founder who– I actually don’t believe that. I don’t have a background in sales, and I’m a pretty reasonable sales rep. I’m not a great sales rep, but if I can be a pretty reasonable sales rep then anybody can be a pretty reasonable sales rep.
It has nothing to do with innate talent. Maybe being crazy 10x SRE recruiter or a crazy 10x salesperson, which there aren’t really that many 10x salespeople. That might be innate, but getting to the point where you can just repeatedly sell yourself is actually not all that terribly hard, it’s just work and craft.
If you have two technical co-founders neither of which can even imagine the scenario in which they’re interfacing with humans on a frequent basis, like wild dogs couldn’t pull them to do that. In that scenario then you might try to go for that more senior person, but in a very specific profile where you’re not hiring–
Because sales leaders and salespeople are not product managers, and they’re not product marketers. They’re there to sell a solution that works, and they’re there to do it in parallel.
A sales leader is there to hire people to do that in parallel, and onboard them, and motivate them, and manage them. They’re not there to build the sales software, necessarily. In that scenario where you have the solo founder or the founder who just can’t even with the learning how to sell, then in that scenario you’re looking for a very specific unicorn-y, purple-squirrel-y type of early evangelical sales leader that is more like a weaponized product manager than actually a salesperson or as sales leader. That is not the common design pattern though, because essentially what you’re doing is you’re bringing on almost like a founder after the fact and that’s very difficult to find.
Dan: There’s not a lot of them. I’ve been lucky to work with a number of people that understand if you don’t know the value of your product or what you charge for a particular product, there’s a handful of people out there that fundamentally understand how to sell value. They go to a person and understand the pain. They’re able to do the math on how much it actually costs to get them to solve that pain, and they get you right up to the edge of it and then they figure out a way of doing that repeatably.
The way that I’ve seen founders handle this is they keep doubling their price with new customers until someone says no and then they go between the average between the last sales price. So they leave a little bit of money on the table in the beginning, but it gets them to that price to figure out where the value breaks.
Pete: That’s a totally great design pattern. We did that at TalentBin.
Dan: Just keep trying and increasing the price until someone says no.
Pete: Well, no. Keep increasing the price until your win rate gets between 20 or 30%. If you’re winning 100% of your deals you’re probably not charging enough. You want to have some people who are like, “Hm. No.” I apologize for the sales jargon. SMB depends on the market that you’re selling into, but SMB is usually considered organizations below 200 or 500.
Again, every organization defines this themselves. Mid-market is between 500 and 2,000 and enterprise is from there on up. Probably a better way to think about it is, how many people have to be involved in the transaction? Don’t think of it in terms of the size of the organization, but instead think about it in terms of how big of a pain in the ass the sales motion is.
Pete: The question was, “What are the perils of raising your average selling price?” There’s nothing wrong with selling more of your solution. The thing that can be potentially dangerous is if that requires you to start attacking customers that require a bunch of new functionality that is very costly from an engineering standpoint. I’ll give an example with Atrium.
We make pretty bad ass sales analytics software, and we hook into the e-mail, calendar, phone and CRM systems for a sales organization in order to instrument the level of sales performance that’s going on there. For SMB and mid-market organizations they’re like, “That sounds awesome. Let’s do it.” And they light up the data and it’s fantastic.
When you get to large organizations they’re like, “Whoa, whoa. Have you heard of the GDPR?” And we’re like, “Yeah we’ve heard of the GDPR.” So if we were to say, “I really want to chase this deal right here, but I have to become GDPR compliant, I have to go through SOC 2 and I have to go through ISO 9000 and 90210. I don’t know what it is.
If I have to take engineering hours and start going after that as opposed to building more user-facing functionality or actual value creating functionality, but instead I’m building software that is purely for the purpose of selling, that might be a bad tradeoff.
If I said, “But I want to sell more and I want to sell bigger deals, and in order to sell bigger deals I have to get SOC 2 compliance or ISO 9000 compliance,” and then I just blindly run after that. As opposed to saying, “Maybe instead I should use those engineering hours to build features that are going to lower my sales cycle or raise my win rate, as opposed to raise my ASP.
I can just attack the world of mid-market organizations, and then eventually I’ll get to that, but in the meantime, I have all these mid-market ducks on the pond. Why would I go after that when I could just more quickly go after– Does that make sense? There’s nothing wrong with higher ASP in general, the question is what’s the tradeoffs that you have to do in order to get–
Dan: What velocity are you sacrificing to be able to go? You’re adding customers at a certain rate, at a certain price, so if you’re going to the next segment how much velocity are you sacrificing? And can you grow into these accounts over time? The companies that we see that we really like are able to acquire companies or customers pretty quickly, but then have the ability to grow those accounts to a pretty sizable revenue number.
So you have this really beautiful curve of like, “OK. They’re adding X number of logos,” and those cohorts have negative churn and so they get it and they keep going up at the same time that you’re adding new logos. You get more of a predictable revenue stream that will grow and help you raise a lot of money.
Pete: Then the way you would test this is you start figuring out what the features are that you run up against, or where there becomes friction in the sales motion. I don’t know if anyone has ever read The Goal, it’s one of my favorite books. It’s an operations engineering book written as a love novel. It’s fantastic.
The way to think about your sales organization is just the machine, and you should try to sell into those larger organizations but then when you run into the fact that above this size of organization now they actually have a tools purchasing committee.
Instead of it being five meetings to get $40,000, it becomes 15 meetings to get $60,000. In that scenario that doesn’t seem like a great tradeoff. I should probably just run more deals that are going to give me $40,000 with five meetings, or whatever the number was. That’s instrumenting that in order to understand how efficient your sales motion is, and then also seeing where the actual closed loss reasons are. When you lose a deal it’s really important to characterize why the organization, why the prospect didn’t buy.
And you can implement this in your CRM with validation rules and checklists or multi-select in order to understand, “OK cool. They didn’t buy because there’s no SOC 2 compliance.” And then over time what you can see is how much revenue you’re leaving on the table, or what proportion of your deals that are coming through the pipe that you’re losing because of SOC 2 compliance, and then how much revenue was lost because of that.
At least you can make an intentional decision. Like, “Yeah. That’s OK for us to leave that on the table because there’s still plenty of these other mid-market organizations that we’re going after.” Or you get to the point where you’re like, “I don’t know. More and more we keep hearing about SOC 2 compliance and the deals are really big, so maybe we should prioritize engineering hours to attack that.”
Pete: If I’m understanding correctly, you’re talking about how do we evoke in the customer or in the prospect a comprehension of the pain that they’re encountering by not implementing our solution? And then the potential value associated with our solution. Is that right? OK.
The industry terminology for that is product marketing. Essentially it’s commercial storytelling, “This is the story of why this product is useful for you. This is the pain that you’re encountering otherwise.” I’m trying to think of the right way, or–
Dan: You see some of this with growth mechanics. Within that there’s a variety of different ways, like there’s some things that are discounting to say, “If you want to sign a long term deal we’ll give you half the price on an annual basis,” and there are some things to pull in revenue where you can do some discounted pieces.
There’s other things on the growth side that either incent people to sign up on the first day, which is you give them free storage or free consulting or a free consultation where there’s something that encourages them to start using the product which then allows you to demonstrate value, and then get the renewal.
And so in a lot of ways there’s the whole concept of, “How do you employ freemium?” I’ve worked with a number of companies that have been on the freemium side of, “What’s the minimum amount of things you need to give away in order to get someone to be a committed customer?” Usually that doesn’t happen from the sales side, usually that happens from growth leaders thinking about, “What are the sort of things that you can do to drive a certain customer behavior?”
If you understand what a customer journey looks like and what is the point that customer is now hooked? And what are the set of things that you need to do in order to get a hooked customer? That tends to be more on growth, demand gen marketing than it is traditionally on sales. I don’t know if you feel differently.
Pete: No, totally. In that case it’s like, “Here’s your pain that you have. Here is this thing that solves that pain. It’s low friction. You should install it and start getting value.” But in order to understand what that pain is, it’s kind of a continuum.
Product management is the elicitation of pain from users and in order to understand what’s the problem that you’re trying to solve here, and then we solve that by putting together the alchemy of software and all sorts of fun stuff in order to build something that solves that pain.
Then once we’ve done that we turn around and it’s kind of like the yin to the yang, we then articulate the stuff that we’ve built that fits to the problems that they articulated that they have. That we build software in order to solve, and now we’re essentially just re-articulating to them, “You know how you said that you had that pain over here, that when you hire salespeople it’s really hard to understand whether or not they’re ramping appropriately? We built this thing that instruments how well they’re ramping, and if you buy it it’s going to allow you to ramp them twice as fast.”
It’s a continuum where understanding and doing a really good job around customer interviews and understanding what the pain points are that you’re handling.
Again, I can’t recommend enough that Michael Sippey article from first round review and the structured customer development interviews associated with it. Because it’s not just, “What’s your problem?” But it’s also, “How big is the problem and how much would you pay to resolve this?”
Because this is the thing that Dan was talking about earlier, where good salespeople, really what they’re engaging in is understanding the magnitude of commercial pain that somebody has. Because if I use, again, LaunchDarkly as an example. If you have outages every so often because a bug goes onto production that fucks things up and you’re an e-commerce company, well that’s a pretty sizable magnitude of pain.
So understanding that and saying, “OK. Cool. You should buy LaunchDarkly. It only costs $100,000 but you have this many outages that cost you $5 million in lost revenue a year. You’re an idiot not to buy.” And they’re like, “You’re right. I’m an idiot not to buy. Here Edith, here’s $100,000.” That descends from understanding what the problem is, the magnitude of the pain associated with that, and then being able to articulate that.
And that gets packaged into promotions like, “E-commerce providers, do you have outages? It sucks, right? You should use–” And then you turn it into a Google spreadsheet which you plug in and you say, “How many outages do you have? How much revenue did you lose for each outage?”
And then of course on the back end you’re capturing that, that’s an ROI calculator which is essentially just a lead generation tool, and then you ship some lightweight outage preventer that is only a very small portion of the value of having full feature flagging, but now is a lead generation mechanism that drives inbound for your $100k ASPs.
Dan: Consumer companies have been good at this for an extended period of time. If you can get someone to engage in these three behaviors and this amount of time, that person then becomes an active user. In the early, even back in the Myspace days, if they invited three friends within this period of time they were going to come back, or if this interaction happened.
In a lot of the enterprise companies that have a consumer adoption model there’s a similar set of behaviors and activities that if you figure out that map, then you can then try and gauge your activities to drive that and get someone to do those behaviors in a specified amount of time, and then it becomes very repeatable.
Pete: Really great examples of that are Slack. Slack is the pinnacle of the inbound freemium sales motion where they haven’t had an allergy to actually selling as well, so they just have this amazing– Like, Slack is a virus within organizations. Somebody who has administrative access to a laptop installs it and invites a bunch of other people, and then it starts spreading out at which point the salespeople are like, “Got another one.” And then they hop on that.
That model can be very powerful if you understand those pieces, but then also understand that you actually at a certain point have to get involved. Forgive me for using a sales tooling example here, there’s a company called, anyone familiar with Yesware? It’s a sales e-mail software company that had a really great freemium adoption model, but they never got to the mid-market and enterprise features.
They never built the features for sales ops or for VPs of sales. They have a huge install base of onesie twosies all over the place, but then there’s two subsequent companies. One is called Outreach, and another one is called SalesLoft that have come in after the fact and really eaten their lunch by virtue of the fact that Yesware thought that was their business as opposed to it actually being a lead generation mechanism.
The question is around customer success when you hire for that, etc. So the design pattern that I like there is one, build the product and understand that the problem exists and validate that. That’s customer development. Build the product that fits that problem and then validate that people actually get value out of it. There’s a lot of organizations that can sell a solution where the customer doesn’t necessarily get value out of it, they don’t use it. The KPIs that you want to move don’t move even though they’ve bought the software.
Having a focus on customer success early on is important even if you’re still in a founder-led selling context. That’s the first and foremost. Customer success is actually very important because people have to get value out of your stuff or else they won’t renew, and that’s really bad. So then the question becomes, “Cool. Whose job is that? Is it the salesperson? Is it the founder’s job, is it the salesperson’s job?
In that scenario I’ll use Atrium as an example right now. I’m the only salesperson at Atrium, we’re still in the founder led selling motion right now, and I have two CS people. This is where I go like this. I have two CS people because we have so many customers that I have closed that if I were trying to service them it would be a problem for me. That’s not necessarily a common case, in my case because I’m very well networked in the sales operations and sales leadership community it’s pretty easy for me to get meetings with prospects and eventually sell them the software.
In my case, because we have so many customers right now it would overwhelm me in order to spend my time on that, so we hired more CS before we hired more salespeople. But you could imagine a scenario where it would only make sense to abstract CS off into its own function or its own human until we got to the point where we had three salespeople throwing off a lot of customers. It really depends. The way to think about it is, what are the tasks that are being done and what’s the opportunity cost of those tasks being done by the founder or the salesperson?
And when it gets to the point where, “We’re spending so much time doing support,” or alternatively, “We’re not doing an intentional job of implementing our customers,” and so we have dead on arrival deals where we’re like, “We sold them a deal,” and they’re not implemented three months later? That would be an indicator that you’re not spending enough time on customer success.
You want to instrument that proactively such that you don’t want to know that dead on arrival deal was dead on arrival 12 months later when they don’t renew, you want to know it upfront by understanding that they have no utilization and they’re not getting value out of your solution.
Having that instrumented and then seeing what the opportunity cost is of the people who are doing that, as compared to what else they could potentially be doing, like closing more business, is really the way that you’re going to know the answer to that question. Unfortunately there’s not a really easy prescription there, but instead it’s more like a framework to help you make the decision on your own.
Dan: Yeah. I think the velocity of the customers that you’re closing will dictate how big of a customer success organization you have. You don’t want it to scale infinitely, you actually want to figure out your engineering team to actually help in some of the customer success pieces as well. Because that will allow you to automate or solve the technical issues that are causing things to not renew or to block people from moving in. The shortcut is to put engineers on the customer success team until they can solve those issues, and those product issues get solved real fast.
Pete: Sounds like a VP of engineering would kick your ass for advocating that. Another way, we’re kind of talking about product management right now as related to customer success and go-to-market, but a really helpful way of doing product management or at least enterprise product management is to correlate the features to the KPI that they’re going to impact. In my case because I’m a glorified sales rep, the KPIs I care about is deal cycle, like how many meetings it takes me to close a deal.
Win rate, how stoked are people and how blown is their mind on the software? Because they’re going to have a propensity to buy more. The win rate, my batting average, is going to be better. And then ASP, like larger organizations that I can sell into. There’s another KPI which is essentially just cost of delivery, how many customers a customer success person can carry.
Right now we have a really hands-on customer success motion. All my customer success people are sales operations folks, because we’re sales operations software as a service a little bit, so we have sales ops people being our CS function as a way to overclock our CS function.
But right now they can only handle 30 customers or 40 customers each, which is not a scalable model, because if you have a customer success person they cost like $80 grand in San Francisco and they can only handle 40 customers, that’s like two Gs of customer success human per customer. That’s not super scalable, so eventually we will build features that reduce that cost by virtue of the actions of the customer success people do will ideally be automated. A very standard sales compensation structure for account executives is a 50/50 base on target. On target means the total compensation, so imagine that you have a sales rep.
The 50/50 base on target where the total compensation of that rep is around 20% of the total amount of money that they bring in. To make this more concrete say that you have a mid-market sales rep, there’s a little bit of circularity here where the cost of your salesperson should be correlated to how much can be sold and at what velocity of your solution. To use an example, say you have $150,000 mid-market sales rep and they’re making $150,000 a year.
Their base salary would be $75,000 and their variable would be $75,000, and then that variable generally speaking is going to be 10% of whatever revenue they bring in. They would have a quota of $750,000 and that would be a healthy compensation scenario.
If you have somebody, and a lot of this is based on the cost of living in a given environment. Imagine that you have software that you sell into SMB businesses, say you’re Groupon or say you’re Thanx or Fivestars and you’re selling loyalty software to 7-Elevens and bars and the ASP on those deals are $1,000 and it takes at least one call or maybe two meetings in order to close that $1,000 deal.
The reality is that person is probably going to, just because there’s only 24 hours in a day and eight hours in a workday, it’s probably going to be pretty tough for them to do more than $300,000 a year in bookings. So you probably can’t staff that person in San Francisco, because again, in order to make our unit economics good we probably wouldn’t want to pay that person who’s doing $300,000 a year in bookings, we probably wouldn’t want to pay them more than $60,000 a year.
And if we did, if we had to pay them more like $100,000 in order to get $300,000 in bookings, now we look like a shitty investment because in order to scale up our revenue numbers we’re going to have to hire tons and tons and tons of salespeople at not great margins. Instead, maybe it makes sense for that sales organization to be in Salt Lake City or Denver or fill in the blank. But generally speaking, the rule of thumb is a 50/50 base on target where you want your cost of sales to be lower than 25 or 30% of the total cost of acquiring that revenue.
Dan: Yeah. Part of it is that most of the comp survey data is pretty good for engineers or other people that aren’t in sales, part of it is that it’s a little bit more challenging to figure that out. Partly, who do we sell to and what kind of sales reps do we need? What is their opportunity costs for them going to work at our company?
We’re looking to hire these kind of people and they’re like, “OK. Their OT is $300k and unless our renewals are really high it doesn’t make sense to actually have customers like that.” And you don’t pay them on the renewal, so you’ll probably move it back and forth a little bit. You’ll figure out there’s the thing that you want, and the thing you can afford. Then it’ll just take a while playing with different types of people at different levels and companies to figure that out.
Pete: And sometimes, what Dan points to right there, is sometimes your pricing has to be predicated on what your sales motion is. The marginal cost of your software is free. So, it’s kind of funny. When MemSQL charges whatever the hell it is that they charge for their stuff, or Instabase charges whatever the hell it is for their stuff, a lot of the time it’s predicated on how hard it is to sell into the organization.
If you have to hire super slick Oracle sales reps or you’re having to hire people out of Cloudera or out of AppD and they’re making $400k a year, that’s their opportunity costs, as Dan says. So then we have to say, “OK cool. In order to pay a sales rep a $400k OTE in order to pull them out of AppD, how many deals do they have to sell?
Actually a good example would be OverOps, I suppose, they’re a Heavybit network company that does that kind of APM 2.0. In order to pull those folks out of there, how many deals can they sell and how many meetings does it take in order for them to do that? And what’s the win rate? “All right. It seems like they could maybe sell 25 deals a year.” “Cool. I guess our pricing has to be this in order for 25 deals to line up to $2.5 million.” So, there is a little bit. Sometimes your pricing isn’t based on the cost. Again, I was joking about the fact that the marginal cost software is largely free. It sometimes is based on the cost of distribution.
Dan: Asheem Chandna at Greylock is probably one of the most successful enterprise investors of all time. The guy has been investing 15 years and has lost money once. It’s unheard of. In the beginning he’s 100% about product, but then after he’s all sales productivity. He’s like, “All right. I know the product is good. Open up the spreadsheet, show me every rep and where they are at with regards to quota, how long does it take them to ramp,” and then once that’s predictable he’s like, “You hire as many reps as you can. Because that’s how revenue growth happens.” That’s why his companies consistently hit their revenue targets and why he doesn’t lose money.
Pete: Probably a thing that could be helpful for a lot of you folks is the book I wrote on Sales for Founders. Just ping me and I’ll flip you the friends and family copy, it’s a bunch of inner hyperlinked Google documents. I swear the content is good even though it doesn’t have a slick cover on it. But those are the things that I think could be helpful to folks as follow up on this.
Dan: Thank you very much.