The First Million is Always the Hardest
According to Accompli CEO and former Redpoint Ventures EIR Javier Soltero, no phrase has motivated him more than an investor saying aloud, “Your product sucks.” The entrepreneur took the core technology behind a company that had already deadpooled, and bootstrapped it as Hyperic to sell it to both Springsource and now, VMWare.
- My Background
- The 5 Stages of Hyperic
- The Takeaway
- Stage 1: Get the Machine Turning
- Stage 2: Adding New Gears
- Stage 3: Guilt by Association
- Stage 4: Tuning the Machine
- Stage 5: Scale
- Off the Record
So a story I've told before. I'm sharing it primarily because as Tom mentioned, I've been where most of you are at. I was an accidental entrepreneur and through that process I learned a lot and hopefully I'll be able to share in as few slides as possible, how we went from 5 engineers, who had never started a company before, to building a successful company that was able to raise venture money from top-tier investors, then ultimately went through one of the most interesting successions of a merger and then, immediately after that a very successful acquisition.
So I'll tell you some concrete lessons that you all will be able to apply to what you're doing. So background on me: I'm originally from San Juan, Puerto Rico. I've been in the Bay area for about 16 years. I worked at a variety of technology companies.
My first company out here was a company called Netscape. Anybody remember Netscape? Probably you all were in like middle school, or something like that, back when Netscape was in its formative days.
So, I've been, for a little bit more than half of my career a developer - a software engineer. I studied at Carnegie Mellon. The way I came to be an entrepreneur was very interesting. I, along with a group of very talented engineers, wound up working at this start-up company called Covalent back in the early 2000's. I joined in 2001, and at Covalent our mission was to basically build a commercial version of the Apache Web server, which at the time and still is, one of the most popular or the most popular Web server out there. So, the idea was at Covalent to turn the Apache server into a Red Hat, Linux kind of business- what do we do on top of Apache to build a successful enterprise software company around it.
Along the way the company managed to raisean enormous amount of moneyfrom top-tier investorswho I'll gratefully namedrop right here.So, Sequoia Capital and Menlo Ventures, thank you.Their money helped us build this awesome product.That company then folded.It folded in 2004 in part because it had beenover capitalized so there was a big lesson thereabout being careful about what you wish for.
The company raised something like 32 or 36 million dollars.The number varies depending on who you ask.I was the chief architect there.In the latter days of Covalent I helped developa new kind of management product that was builton a vision I had from prior experience of buildinglarge scale Web infrastructure environmentsfor which there was reallyno management technology to use.
So we built this website before Covalentcalled BackFlip, it was a bunch of ex Netscape people.The site really never went anywhere,but the one thing I learned was that we had these "ops" peopleand all they knew how to do was mess with Cisco routers and cables and racking stuffand bolting things onto pieces of metaland occasionally installing Linux. In most casesthey'd complain about that becausethey were FreeBSD people,or whatever the flavor was at the time.
So, what we learned from that experiencewas that we had to operate the sitebecause the op's people didn't know anything aboutthe actual application.
So at Covalentwe ended up building the product that we wishedwe'd had at BlackFlip which was a distributedmanagement product that allowed you to inventoryand autodiscover everything aboutyour environment and your applications.
It could collect every single performance counterthat was available in any part of the stackfrom hardware, to software, to middleware, to your app. It could control it so we could start stuff discreetlyfrom our console and doa whole range of other things.
We built that product and we're really fired up about itand we're like "Oh my God this is like the mostawesome management product ever."
But then Covalent actually goesand takes it to market usinga very traditional enterprise software model, after pouring $15 million dollarsof their money into developmentand marketing, and kills it right as it was going out into customer's hands.
So, right before the company ultimately folded,we came up with this idea and we asked"What do VC's do when companies foldwith the technology that their companies built?"
They don't repossess it because they don't knowwhat to do with that kind of stuff, right? So they basically, they let it die. Like that was the outcome that we all saw.At the time, Google was aboutto go public (for those softwareand Internet historians out there).So, the way venture investors tendedto think was, "We're about to make lots of money over here" or "This company is failing, let's wrap it up."
So we took advantage of that opportunityand came up with the idea of saying, "They [venture capitalists] don't want this software backand we've already sold it to 3 customers.They're gonna be pretty upset if theyget burned on buying theproducts from these start-ups. "
So, I and 4 other guys went up to the Covalentboard and investors and said "We'll make you a deal.We'll give you a dollar in exchange for assumingthe liabilities from these 3 existing customersand you'll give us every single bit of this IPand we'll call it a day."
For the venture capitalists, that was the way they gotto save face and say, "our company didn't go under" -- I guess today you'd call that a pivot, right,because there's a term for everything.
Now they're focused on managementand it's called Hyperic. Why?Well, because all other good names were taken, right?We were convinced, at the timethat we were going to ditch that name immediately.So we went off and acquired this piece of softwarealong with these obligations from verylarge customers that thankfully I had beenvery involved in setting up to begin with.So these are customers that I knewwe were expecting to get zero money from, by the way.That money had already been collected by Covalentand subsequently evaporated.
We said, okay, "let's go builda different business around this."We felt our motivation, we as technologists were like, "You know what? This is unfinished business."
The thing that was messed up about this companywas the go-to-market model and the way that the executive team made decisions.It wasn't the product.What did we hear from the company? This is a product in search for a market.
Now, there has been no phrase that hasmotivated me more than the idea of someinvestor saying, "Your product sucks".
It was about looking for opportunity and they were, if nothing, elated at the idea that we were taking this ridiculouspiece of software off their hands.We grabbed it and set offand started our own business.Part of setting up that businessand the story that I'm going to tell you todayis about the decisions that we made while not knowinga single thing about a go-to-market model, how to sell products, and how to doanything other than things that seemedlike the right idea at the time.
Just to finish on sort of my trajectorywe built Hyperic out,built it from a bootstraped company to a venture-backed companyto ultimately merge it with SpringSource in 90 days,selling it later to VMware.
I spent some time at VMware doing what occupied entrepreneurs do at large companies,which is try to be helpfuland productive and interestedand ultimately spent 3-1/2 years there.I learned a lot of great things whichI won't cover here today.Maybe that will be the subject of anotheroff the record Heavybit talk.
Now I am currently an EIR at Redpoint Ventures,although officially I am now CEO of a companythat is just getting started.So, enough about me.
Before I walk you through the 5stages of what we went through at Hypericand how they might be helpful to you, I'll tell you a very quick story.Like every good entrepreneurship storyI was tested very early on by the adverse forcesof the market and circumstance and what have you.
I had this very interesting conversation with my dadwhere I called him up and he's like,"Well, how's your thing going?" and I said, "Well, it's not going good.Things are just not happening.There's no money coming in and we don't haveendless amounts of money to keep going and I'm worried."
My dad, who has been an entrepreneur asked, "Do you believe in the product?Did you build a kick-ass product?Do you really think that your product is helpful?"
And I answered, "Well yeah, yeah, we built somethingreally special and I'm really convinced about it."
He's like, "Well if you really are convincedthat what you've built is awesomeand can help customers, then you should stick to it.The moment you stop believing in that is when you should give up."
It sounds a little corny and campy,but the reality was that I got similar feedbackfrom other people during that time.
I said to myself, "This is a product that solvesa real problem that people have. People have already paid money for this product,so it's not like you're trying to figure outif there's really a market here.You just have to hustle and figure outhow to get it into more people's hands. You need to understand theright route to making this business a success."
So that's what we did.
When all else fails, you believe in your productand its ability to help customersand that will get you through the day.
All right, so now onto the concrete parts of the lesson.The first thing we knew as 5 guyswho'd never run a company before: We had some idea of the problem that we solved,but we were convinced that whatever it was thatCovalent was actually doing to take thisproduct to market, wasn't going to work.
So we're saying, okay, "Let's go back to basics.What is it that we need to do to get thisinto customer's hands and get people paying?"
Well, the first thing is:Who's the right person to be talking to?Who's adopting this product?Who's influencing the decisionand who's ultimately buying?
Covalent, because it was playinga fundamentally new gamewith old rules, was going at this backwards.They were going first to the buyerand then working their way down from the buyer,to the influencer, to the ultimate adopter -- because that's how enterprise software companieswere built back in the day.
So, with them, I would parachute into an accountalong with a salesperson and say,"Hey, we have this awesome product, and can youplease let us install it so thatwe can show you how awesome it is? We can have a discussion about how it might helpyou, and then we can maybe talkabout you buying some of it."
But in the new world at Hyperic we didn't have this luxury.We didn't have any money, zero.No seed money, nothing, there was no money.
So we said, "Well okay, we can't do that.We can't fly anywhere,we have no money to go anywhere.So how do we do this?"
We thought about it, "Well, who are we trying to help?Is it the CIO of Morgan Stanley or whatever?No, it's the sysadmin.That's the person who touches and interactswith this product and who we help, right? That's the person who isthe first person we should talk to."
Okay, so it's the sysadmin. Great. Check. Next question: What are they looking for?Are they looking for a next generation Webinfrastructure systems management orwhatever they were saying when we were selling Covalent's version of the product? No, they're looking for Tomcat Monitoringor Apache Management or MySQL Monitoringor EROS or whatever.
They're looking for a productto solve a specific problem.They're not looking for a general vision.They're not looking fora systems management platform.Nothing like that.They're looking because they have a problem.
They run a bunch of Apache, it's breaking.They need to monitor it. Boom. Okay great.
Where are they looking?
Keep in mind this is 2004,but where are they looking?They're going to Google.If I had a problem and I'm like a sysadmin,and I manage a Web infrastructure, I'd go to Google.
What would I do?
I'd type Tomcat Monitoring into Google. Based on the results I would make a decisionabout what I'm actually going to try.It gets even more nuanced than that, because we said,"Okay, they're going to go to Google, but see our buyer - this sysadmin - is the most jaded person on the entire planet.They don't trust anythingthey hear from software companies."
So, they don't believe anything you say.So on Google it turns outyou can buy your way into placements.Our view on our target adopter was so nuancedthat we knew that at least our guy or galis smart enough to not necessarily click on adsthat they believe can be gamedby buying and so forth.
So, what does that mean?
Well, it means thatwe need to be the #1 search result fororganic search results forTomcat Monitoring, et cetera.So for those of you with computers in front of you,you can go look up VMware Management,for example, and I believe Hypericto this day is like the #2 search resulton Google for that.
The idea was they're gonnalook for something specific.They want to land on genuinely good contentthat speaks to how you solve that problem. It's laid out specifically with pictures and in our case, a list of every single metric eventand control action that we gotout of that specific product was listed.Yes, by the way, we monitor MySQL, Linux,Solaris, Oracle, all this other shit, too. But for you my friend, you're looking for Tomcat and this is what we do for Tomcat.
The next thing we want you to do is click onthis bright red download buttonthat happens to be,next to this stock photography that apparentlyevery single company uses or used at the time, that had the most multicultural crew ever assembledinside of a data center includinga woman in a dark red sweater standing nextto an aging guy, and a white guy, and a skinny guy,and a fat guy. It was hilarious.
I used it, you know, in at least 7 or 8 occurrences.We recently dug up that picture for posteritybecause it's a classic.
The idea was, there's a button to press where you get to a download,and the shortest path between [a customer] pressing thatdownload button and gettingtheir hands on the software, is what we [the company] were looking for.
So guess what? We did that.
Then next thing you know, in late summer of 2004,we start getting registrations. All of a suddenthere's people filling in the formand telling us who they are,and giving genuine phone numbers, and all this stuff.
I'm like "Sweet, and so what do I do?"
Well as the designated sales guy in the companyI start calling people up and then I get stuck.
Basically I was calling people and realizingthat they had downloaded our software,but they had not installed it, and that was a problem. Part of what we were trying to designin our sort of poor-man's go-to-market strategy was to get everybody to install the softwareand by the time we talked to them, they'd be delightedand ready to pay, right? Not quite.
So, I'd call these people up and I'd say"Hi, I'm Javier from Hyperic, blah blah blah"have you installed it?""No, I haven't had a chance, or"Hey I tried it, but I couldn't get it to work."
Then I'd ask "Hey, can you please give me some time towalk you through it?"Some people did, most people didn't.Somehow it just never added up.Things got really ugly.We were basically at the end of our ropein terms of our tolerance to not make moneyand to endure the hardshipsof bootstrapping a company.
So, I started thinking"We're in really, really deep trouble."Because I had no other recourse, I got together withmy other co-founders and I said"Hey, we're in really deep trouble. People are just not able to install our productso I think the only thing we can do is basicallystart applying our product and engineeringbrain to this to see if we can help solve it."
So, what do we do?
Okay, well let's start with common sense. So you're selling to the busiest,most jaded person in the world, all right?
So given that they're busy and jadedwe have to basically line up and write out all the stepsfrom the point at which they're looking for us,to the point in whichour product is actually showing value.And that's an exercise, by the way,that I recommend every single personin this room actually go and do.
I mean be specific and detailed about the user steps.In our case it was [the user] goes to Google,enters Tomcat Monitoring,goes to landing page, clicks button, fills form,picks package -- I mean every single step.In our case it turned out it was 39 stepswhich is a shitload of steps.So, we said, "look, this has to be 3 steps."
Of course, my co-founders who had known meby then for a while were like, "Dude, you are crazy. Get out of here with this, there's no way. We can't do it."
We gave ourselves an hour.We said, "We're talking to the busiest personin the world and that person is going to give usan hour of their time and that's it. " It turns out today, sadly for all of you,that person is going to give youno more than 20 minutes.
Why? They're busy and frankly, they've heard it all.So, show don't tell, right.So out of a very, very significant hardshipthat we faced as a team, we didn't get on ourcreative marketing horse and start talking differentlyabout what we were doing. And we didn't juice up our Google AdWord spend.
We basically went back to the product and said"Well the only knob we can turn here is to make our product easier to install,"and arrived at what I wound up callingthe "theory of reasonable defaults".
It's basically, Don't ask the user questions. If they find Tomcat, fucking configure it,and do whatever magic.Don't ask them what metrics they want out of this, just enable which onesyou think are the right onesand they'll go and change them later.
Like there was just some fairly obvious thingsthat we were thinking as engineers neededto be decided by every person,but the most important thing was thatduring that exercise cementedour view on how this process was supposed to gofor our customers. And I should give due credit to my co-founders,because I wrote none of this code at the time.
Inventing things like this are part of the reason whyHyperic today is a) still called Hyperic,and b) still sold as a stand-alone product by VMware.
And that is basically the value of time. In a category of products where people toldbig stories, and people hated the category ,and every productthat was in that category -- Hyperic, for all its warts, was still the only productin its class that would be up and runningin less than 20 minutes.
The skill-set required to install Hypericwas no more advanced than the skill-setrequired to install Microsoft Word.I think that was a very,very important benchmark for us.
We did that and next thing you knowthings start actually happeningand we start getting more interest from customers.We said, "Well why are they going to pay?What is it that we charge for?"
We had a number of offerings where you could run it on 3 computers and it's freeor, you know, anything above thatand you have to pay.I don't have the right answer for each one of youin terms of your business, but you have to havea very, very clear viewpoint on where the value is.
I'll talk a little bit about that further onin the presentation, but it's better for youto not think of the paid component like you're offering a strict list of features.In our case we wound up arrivingat a list of principlesthat when applied to anything about how webuilt product or offered services,meant that you had to pay, right.I think it was performance, reliability, and scale,that were the 3 axioms uponwhich we built our paid offering,but I'll talk about that a little bit more later.
The last part and this is for all of you who havethe unfortunate task of having to sell,even though you don't like. But as entrepreneurseventually you will have to selland you will have to do itespecially if you're selling to IT. You'll have to get on the phone with somebodyand convince of the orderand at times, if you're a kind-heartedsort, it's the hardest thing ever.
People will string you along endlesslyuntil you say, "You know what,time to pay."
And they'll answer, "Oh, I was gonna use the free version. Good bye."
And if you don't have the balls to do that, well then you shouldn't be in this gamebecause frankly the buyer, that same jadedand busy customer is also the cheapest personyou will ever meet in your life.They love free.They'll take you for everything you've got.So keep that in mind.
Okay, the next thing that happenedis we started getting calls from companieswho wanted to OEM our product and OEMfor those of you who don't know is basicallypeople wanting to license our productand redistribute it as part of their own offering.
Now initially when I was first approached by JBossat the time, who was just taking off as the sort of open sourceJava application server of choice, I was really uncomfortable with thisbecause we were a 5 person bootstrapped operationthat was just getting theirdirect sales act going. The prospect of licensing our technologyand having somebody else sell itit was very uncomfortable.
It felt like I was giving away the keys to the store.There was really no way that I ended upbeing able to get comfortable with the idea thatno matter how much money we were going to makepotentially that it was the right deal to make.
Frankly, I didn't have the blue shirted, khaki wearing,beady guy on our team to go and putz aroundwith these guys and arm wrestle over term sheetsand all this other bullshit.
It was just me, and frankly doing the wrong dealat that stage for us would have meant basicallyselling the company without knowing it.
So I was very confused about thisand it really tormented me for a long time.So I'll spare you the torment and I'll give yousome clues as to why or why not you should do this.
So, why should you do an OEM?
Well, you're a small company.You can't get everywhere that you want to get toso if you can, use OEM distributionand this is different than resellers.No one in here should be contemplating resellers,in my opinion, because no one can sell your shituntil you figure out how to sellyour own shit basically.
So in terms of an OEM if you have a potential partneror licensor who can get your product to placesthat you know you can't get to,that's a good sign.
The second thing isyou're going to get paid for the privilege. So as a company that's bootstrapping, it turns out that when we ultimately did doour deal with JBoss after much harassmentand negotiation. And we were paid very well for itand ultimately it got usa higher profile as a company.
Now Hyperic was a package software product.It's different and a lot of you guys are doingdeveloper services.We can talk a little about that in the Q&A.I'm curious if any of you have entertainedthe possibility of OEM'ing or doing licensing deals?
Regardless, the idea of doingoutsourced distribution or licensingis something powerful.
Why shouldn't you do an OEM?
Well if your product isn't suitable for it. Namely, if you can't divide your product upin such a way that you can license part of itto someone and still retain enough valueon your own to proceed with your own plan, then you shouldn't do it.
That's when you're giving away the store, right.
Then secondly as I already mentioned,if you don't know how to sell the product, thenyou don't want to have somebody elsetry to sell it for you. The chancesof them being successful are slim to none. For OEM's in this case, you should assumeby all accounts, that the answer is no.
So, good guideline: If you can't put yourselfin the shoes of the potential licensorand see with very, very strict clarityhow they would package and sell that productsuccessfully to their customer, then don't work with them.
You need to understand both sides of that deal. You can't just think, "Okay, I'm gonna get paidand the absolute sunny day scenario of this dealis going to be worth to me 2 million dollars." You can't just let them do whatever they're gonna do, wish them good luck and expect a check every quarter.
That's a sure fire way of getting royally screwed in this business.
Another good guideline: This isn't worth doing unless you expectto see 15 to 30% of your revenue or your bookingsor whatever from this source.
If it's going to be incrementaland you can't see a way for anOEM deal to turn into a significantcontributor of revenue in cashto your business, then don't bother.They're hard. They're not autopilot-able things.They're certainly not thingsthat are implicitly successful, they're hard.If you do them right like we did, and it ended up even years later, our OEMand indirect distributionturned out to be about 30% of our business.
It was really, really helpful.
So anyways, some good observations on howto add new gears in the distribution leverage of OEM.The third thing.Okay, so the wheels are turning.Maybe you have some OEM.So if your product is getting out thereone way to overcome the jaded and sort ofskeptical nature of your buyer, is association marketing.
One of the observations that I came up withwas that every small business wants to pretendlike they're big in some way.Every mom and pop IT shop thinks thatthey roll like Google does.So the idea is that you land a customerthat is a name brand customer,and you use thatas leverage. You help customers see the way they operateand the way that you would help thembe more like the name brand customer.
That's a very powerfulstory to tell, and we did that very successfullywith a number of high profile clients.It's tricky because they're clients that are oftenvery expensive and they take more out of youthan you get in terms of your revenue. But if you manage them correctlyand if you pick very few of them (ideally one), then at least for every customeryou get after that, you can say"Yes, you could see how we did this for Google".
The last bit on this isthe land and expanse element of this. The people talk about starting with a small toeholdand growing your presence in an account. But don't assume that happens on autopilot.It takes just as much, if not more effort,than it does to land a new customer. Part of the reason is that unless your product is bloody miraculous, it's gonna take effort for the customer to seeenough success in order for other peopleto really be wired up to advocate.
So you need to think very carefully abouthow you achieve that and how you buildyour internal champions and so forth.
It's a really important componentand later on when we grew up as a business,we had two vectors of new business.
- We had new-new businesswhich was new business from new accounts,and;
- New business from existing customersand we had different strategies to deal with both.
Step 4, this is probably the most interesting bitthat I'm gonna share with you today - Tuning the machine.I looked at everything we did.It was like we were assembling an engineand it was the most fucked up engineyou can imagine, honestly.
It was an engine that we were assemblingfrom all kinds of scavenged partsand guesswork and whatever,but it ultimately did start turning like an engineand it was moving, and moneywas coming in and it was working.
So the next question became pricing.
So on pricing.Pricing is a really, really complicatedand often troubling thing.There's 2 approaches you can take.
- What I call scattered plot, which is basically when you don't know whatthe right price for your product is,so you just do your best tokeep it within some boundariesand see where things land.There's some strong advantages to thisbecause in an early stage companyyou just don't know, right.You'll know later because as people becomemore familiar with the productand as you know a little bit more aboutwhat it is that's worth sellinginside of your offering today,you'll be able to calibrate that over timeand do some intelligence on that.
- The second approach is discipline,which means you're gonna be absolutehardliners on this, there's only one price.That is very courageous.Anybody here doing like sort of discipline pricing?You have one price and it's non-negotiable.
So for scattered plot, acknowledge your pricing strategy and guide yourself and whoever is doing your sellingto say here's the boundaries.We're gonna spend our firstx-number of months or year,getting deals within this bandand then we'll figure it out.
For discipline pricingmyou're gonna have to have the courageto hang up the phone and be like"The price is x," and our friend,the jaded, cheap, disbelieving, busy,friendly IT buyer is gonna say,"Well you know, there's always a deal,and why don't you have published pricing?"
IT wants to have it both ways.They want published pricing becausethey demand that you tell them upfront.Some buyers, a lot of them, are going to say"I don't trust any vendor that doesn't put the priceson their Website, but then when you say"Well that's the price. You saw it on our Website."they're gonna say, "No, that's not the price.The price is what is today the 29th?What is the price on the 31st?"
Right,and they're gonna do all that kind of BS whichin a disciplined pricing model you have to becourageous enough to say, "the price is the sameon the first of the monthas it is on the 31st of the month."
I commend you if you're able to do that.That is very, very, very, very hard.It's hard for VMware and Ciscoas it is hard for, you know, all of thesmall companies that are gathered here today.
More importantly, you need to understandthis machine.I'm like, "Okay so what are the pieces of this engineand how do I effect change?What are the knobs?"
You're first spending a lot of timejust assembling an engine that runsand then next thing you know, you'rein a position to say "I'm gonna tune certain parts. I might upgrade certain parts."
That's sort of how we looked at it.So in order to do that I'll share with youthe construct that I arrived at in orderto explain to the company and our investorshow money actually arrived at Hyperic.
I called it "the diamond" and there'svariations of this. I mean you hear a lot of peopletalking about the funnel purely froma sales lead perspective,but that is a very myopic viewon how someone goes from being a prospectto being a customer.
This is a lot more about being in an open source companyas we were, or in a company where there isa free product and then a paid product. It's about how the dynamics work between how people moveand self-qualify themselves through that process.
So how does this work?Okay, well at Hyperic the first thingthat we wanted people to decide whenthey showed up at our Website, waswhether they'd have an open sourcerelationship with us or are you going to havea commercial relationship with us?
Unlike some other companies like MySQLwho had had a dual identity.There was a dot-org sort of open source siteand a dot-com commercial site.Hyperic had really only one site.So the expectation was that somebody wasgoing to show up at our storefrontand make a decision.
So the important thing there is, you wantto make sure they make the right decision.You also want to make sure they make a decision.
So, there's a decision that says, "I'm an open source user, I have no intentionof talking to a salesperson.I am kicking tires. I don't know if I like you yet." So those people are goingto go down the left hand sideand the action, the measurable event of that,is a download of our open source product.
On the other hand, we had a downloadof the commercial product.Now the important thing about the right sideof this diamond was because we were a commercial open source product inan established, predominantly proprietaryor traditional enterprise softwood category -- there were plenty of people who showed up atour doorstep evaluating us againstcommercial proprietary vendorswho charged a shitload of money.
We didn't want to indulge the cheapnessof that prospective buyer,nor did we want to miss out on the opportunityof showing them what the full weight and power of our sales team,our sales engineers, our product organization,and all that were.
We wanted to make surethat if you're showing up at our doorstep with cash in hand,that your ass is going down the right sideof that diamond for sure.
That would result in a commercial downloadand so off we go.
The next step in an open source download eventwas the actual installationof our open source productand this we knew because our product phoned home.
This again, this is biased I suspect a lot of you havemodels that don't necessarily apply directly to this,but this is kind of how you should think about it.
They [customers] start using your product and thenwe have some sense that they registered and have actually activated it.They've gone through an event that tells usthat they're actually using it in some way.
On the commercial side it was a little bit differentbecause right after the download and the registration,you'd get a phone call and that would qualify youand find out if you were actually serious. It found out if there was an active project, who you were,who were the other competitors, et cetera, et cetera, et cetera.
So actually, in fact, I mislabeled that.
There's sales lead and then there's sort ofthe notion of a qualified opportunitywhich is really the next bit.The difference between a lead and an opportunityis basically not only do we know thatyou're actively looking,but we feel like we have a chancein actually winning the businessand we're on your shortlist of potential purposes.
In every part of this there are numbers.So for those that are sort of disconnectedfrom the notion of bullshit go-to-market planningand metrics, it's very easy.
Up here 100 people showed up on our Website.33 of those people would make a decisionto either go one way or the other.That's pretty good.That means 1 out of every 3 peoplewould take an action on our site.
From a Website conversion perspectivefor this kind of product, we feltpretty good about that.And then you'd have a certain number of downloadsand those downloads would result in installs.Now the degradation of this and the delta between the numberof downloads and the number of heartbeatsthat we got on a given day, told us somethingabout whether our product truly wasas easy-to-install as we thought it was.
It turns out, as we learned immediatelyafter we put this process in place, thata huge percentageof these downloads were Windows. But then, a very small percentage of theseinstalls were Windows. That told usthere were some real issues withour Windows installer.We fixed that.
This conversion or this sort of percentageactually increased tremendously.
Similarly, the delta between commercialdownloads and sales leads tells you when you are actually luring the right people.Maybe they're looking for a product, they think your product is the right one,but it turns out it's not.In the sales lead to qualified opportunitiesthen, you're more into measuring whether yoursalespeople are actually navigatingthe account correctly and are taking customers through a process that gets themmore excited about buying stuff.
In the last vector, we came to the conclusionthat it was very difficult, if not impossible, to directly monetize our open source product.
We played with it and realized that the cheap, jaded buyerneed to be on theright side of this diamond.Anything over here and they're sort ofpredisposed to avoid paying for anything.But what we did notice was there wasa significant number of things that becamequalified opportunities instantly.
More than 3/4 of our qualified opportunitieswere actual open source users.Why is that important?Well, the whole reason we did thiswas because we wanted to createa self assisted POC, an evaluation processthat involved no interaction betweenHyperic and the prospect until it was prudentto actually talk about a commercial engagementand money.
It costs you money to have peoplebasically bullshitting with thosethat are not going to buy.So if you have sales people, marketing people,any professional other than product peopleengaging with prospects that are of the wrong kind,you're wasting money.
These designs, these kinds of businesses,and I think this translates very wellto the sort of SaaS and cloud-basedmodels of today, are fail fast businesses. You should spend no time directly engagingand courting those early-stage free offering customers.
You should build a business model, a company, and a productthat guides people through that processno matter how long it takesbecause you can't force this to happen.
So ultimately as we looked at this diamond, we saw 3 very important functions of the companythat could be held accountableto each part of this recipe.
- Up here this was all about marketing.How are you bringing people to the site?Is it SEO? Is it PR? Is it events? Is it?Whatever it is, whatever program spend you're doing,comes in up here. The decision to go one way or the otherthat's the decision around copy.It's a decision around Website structure,landing pages, A/B testing, all that fun stuff.Marketing could be held accountable.If we ran a program that saidthis month or this quarter we wantway more downloads of our new open source product, marketing has the knobsto actually turn further in that direction.To put it in context of a smaller companyall of these things are things you caneffect on your own. You don't needa full marketing team like what we hadwhen we figured this all out. By understanding the mechanicsof the machine, you will do a lot betterat trying to effect change and it willfeel a lot less like you're just tryingto throw darts to get more money and more leads. It's just much more concrete.
- Product: Similarly, any drop off betweenthe download event and either an installor a sales lead, was a function of whetherthe product was actually doingwhat it was supposed to do.So if it was an installer issue,or there were quality issues in a release,you'd see those manifestthemselves very quickly in that process.In fact, when we looked at the timewhen we put this together, we had an initiative.So this was called "Project Diamond"and we had this initiative about tryingto effect the supportabilityof certain sized environments.So there's a separate graphthat I'm not showing, where we distributethe number of nodes of our software with self-supported customersthat require no interaction from us on the one side and basically occasionally file a support ticket,but they don't need hand holding pre or post sales.And to the right of that lineI expect there are high-touch customers.Being able to move that line and makingour product better and more self sufficientin higher scaled customers, ultimatelywas a function that was expressed in theway these numbers actually worked. These are the kinds of initiativesthat as you grow up and becomea bigger company,you can give people a rallying cry,a measurable rallying crythat allows them to get better.
- Lastly, with sales: I think with the various conversion events in between here, turns out were actually good at Hyperic.So for things that wound up herem I thinkwe won something in the order of over half of any opportunity thatwe were actually actively engaged in.Now if you trace that number all the way backto here that doesn't mean that we weresuper crushing it. But at least we knewthat by the timethat somebody got here, the chances of usactually winning were very good. I encourage you to use some kind of constructlike this to basically understand your engineand understand how it's all put together, in order to then be able to effect change on it.
And then the last bit is the subject of scaling.And this I have the least amount of content onfrom a bullet perspective.People think the next step after you've raisedseed capital and after you've kind ofestablished yourself as a business, is to"scale the business". A lot of that involvesthe scaling of your sales modeland scaling of your ability to make revenue.
The debate about when you should hirea sales leader is one that is kind of hardto answer and depends on a case by case basis.
My personal advice: The reason why I waited to hire a sales leader,is the same reason why I'm writing codefor our new project.I want to be able to know how hard something is,before someone can come and bullshit me about it.And it's the honest truth.
Sales people, sales leaders in particular,will show up and dazzle you. You might think, "Okay, here's the keys. You have a number, here's your bag, I'll check in with you a couple of times from hereuntil the end of the quarter."
And that's the wrong move.You need to feel as the entrepreneur, direct ownershipof making that number.
At Hyperic it was such a big dealthat every engineer, every support rep,every person who worked at our companyknew what our revenue target was,our bookings target was for that quarter.
Most companies actually don't do it this way,but because I came into the whole processsort of obsessed about it, and becauseit was such an integral part of our successas a bootstrap venture, I remember distinctlybeing in all-hands meetings and havingQA people come up to me and say"Jav, how's the quarter looking?Are we going to make the number?"
That is a good sign because when you lookat things in terms of that diamond people knew what they could effect and feel ownership over the financial successof the company.
You don't want to ever create a situationwhere there's a separation and the product guy says, "I'm just a product guy building product,and the sales people well if they didn'tmake the number well then they suck. "
I worked at plenty of start-ups that operatedthat way, they all failed, all of them.In terms of how you go and make this case to VC's, well the best way having sat inside ofa venture firm for the last 4-5 monthsis to be able to speak from experience.
So unless you know and unless you havesome opinions and grounded facts aboutwhat it takes for people to actuallyspend money and pay you for stuff,then there's really no point in going toa VC and saying "I want you to give me moneyso that I can go and then hire somebodyto go help me answer that question."
They're not going to go for it.I mean they might, but they probably suck as VC's.
And then the last thing I encourage you allto think about in the event that you arethe founder or CEO, is that your role will change,so in the best case scenario you willevolve a lot in terms of your responsibilitiesand the things that you're most focused on.
You should be conscious of thatand not be sort of a passenger on this journey. You should decide who you're going to be, how you're going to stay involved, what you'll needto compliment your skill-set in orderto make this company successful. You should as, "What do I really want to stay involved with?What do I not give a shit about?"
A good example of this is with my good friend Martin Mickos, the CEO of MySQLand his thoughts on facilities.
He didn't give a shit about facilities,so much so, that one day I went to visit himin his office and I noticed that he wasin a different office, and I said"Oh, you moved."
And he said, "Really? I don't know I just came here."
I asked, "You're not in charge of deciding the office is on the 7th floorinstead of the 6th floor."
And he answered, "I have other people worry about that.I don't worry about that at all."
So knowing what you care aboutand what you don't care about, is really important.
Some people care about facilities.I didn't care about facilities.I blame Martin for that.And that's it. Thank you for listening.