Edith Harbaugh

CEO of LaunchDarkly

Convincing Buyers to Buy Versus Building

Edith Harbaugh: It's great to be here. It's great to be here because I see so many familiar, friendly faces. Um, I actually started off as an engineer and then I was a product manager in marketing and sales. So I've pretty much seen every step of the life cycle life cycle of a software product. Um, today I'm gonna talk about buy versus build. And first we'll talk about why buy versus build is a good thing. Having been very early at several companies, the biggest challenge you have as a new company is that no one knows you exist. I think we all have this vision that we're gonna build this product. We'll launch on tech crunch. 7 billion people will show up and they'll all eagerly use this. That has happened to me. Never. Uh, the, the usual pattern is you have this big launch, nobody shows up. And even worse, if anybody shows up, no one cares.

Edith Harbaugh: So the very first part of any product is just trying to get awareness of it. And that's why I went into marketing. The really nice thing about if someone cares about buy versus build, you have a potential buyer. So that means that you have some things that you can start to work on in terms of converting this person from someone who's thinking about five or spilled into an actual buyer. So I'm going to walk through three examples of why you can win on buy versus build. So let's first consider a roast chicken. Uh, this is something that many of us eat quite often. It's delicious, it's nutritious, it's kind of a classic food. So there, there's this really easy thing that Costco has pumped out a ton of really discount chickens that costs a dollar 99. And this is a pretty easy trade off. You might think, you could say, Hey, I can get this chicken for a dollar 99, dinner's done. And why doesn't everybody do that all the time? So this is a kind of classic trade off of cost versus time.

Edith Harbaugh: There's all these reasons why people might want to cook a chicken at home. You know you might want to cook it with your special soy sauce with a little hint of Sriracha. You might want the potatoes the way your mom used to make them. You might want extra garlic and you might want to just really, you might have a lot of things that you want to customize and the chicken. You might also really content consider bill baking this chicken really core to your identity. I make the best chicken. Everybody comes to my house for chicken because they love my chicken and that gets a deprived. If you're having a bunch of people over at your house, it's not quite the same impact. If you unpack the $1.99 chicken from and say, here, I pick this up on the way home versus I made this for you. So that's, that's a reason why people might really want to build their own chicken.

Edith Harbaugh: Another example is Zuni Cafe, so I'm very fond of Zuni is chicken is not cheap. I think it costs about $35. Now that's not quick it an hour. So you literally have to go to Zuni, ordered the chicken, wait an hour for them to bake it in their wood fired oven and it comes out and it is delicious. It has this bread underneath it and it has this drippings and it is so good. So if you're visiting from out of town, please stop by Zuni, get a chicken budget an hour and it is really good. Do the, these are some examples of why somebody might say, Hey, this dollar 99 chicken I could get in five minutes, but I'm willing to invest much more money, much more time because I want kind of this delicious experience, a premium experience. So with those kind of five factors of why people buy versus build, I'm going to talk about how I applied them at two separate companies.

Edith Harbaugh: So this is a company I worked at at 2007, which was easy bloom. Uh, we did not have IoT back then, so we just literally called it an internet enabled garden sensor. Catchy. Huh. Um, the way it worked was you kind of jammed the business end into the soil and it would measure the soil temperature. So a witness and then the top would measure how much light it was getting. So the purpose of this was to tell you what plants you could grow. Uh, when we launched it in November of 2007, we expected to sell think about 5,000 units a month.

Edith Harbaugh: Our actual sales were a hundred units that month. As you can see, there's a pretty big delta between 5,100. Uh, the really nice thing about having a completely botched product launch at a really small company, we were eight people, is that the CEO said, Edith, whatever you want to do, just do it. It can't get worse. You know, it, you know, if she's, he's like, if you saw more that like we've, I think we sold around 70 units that first month. He's like some more than 70. Great. Cause we, by the way, we have this warehouse full of 20,000 units that are sitting around. So I looked at why people didn't want it to use it. There's a lot of resistance from different personas. What I found is that some people take guardian as something that's really, really core to their personality. Like they like being outside, they like feeling the soil with their fingers.

Edith Harbaugh: They like Kinda sensing the temperature. Uh, they just found that this was something that just was very threatening to them. Also, the price was an issue. Uh, it costs $69. If you remember December, 2007, this was also when there was kind of a global mortgage meltdown and recession and $69 seems like a lot of money to a lot of people. So what I did is I kind of broke down why people are resistant. There was the cost. Um, there was also that people felt that guardian was very core to their identity and they didn't want this getting in their way. So we dropped the price. We dropped it from 69 to 49. If you're ever considering price testing and there's some barriers. 49, at least back then was this real sweet spot where suddenly people said, oh, $49. Great. Uh, there are a ton of people who do not know what to get their parents for presents.

Edith Harbaugh: So we bought a lot of ads around, um, green, uh, gardening, mother gardening, father what you could buy them. And this was really good cause everybody's like, oh, 49 bucks and I have a present for my mom. Awesome. So our big peaks are actually mother's Day and father's day cause people are always looking for something to buy. Um, the other persona that we really succeeded with was we call them the gadget geeks and it was basically people in San Francisco as cities who had maybe one plant of their apartment and really wanted to kind of monitor that plant. And I was thrilled when I, um, I met the people from Heroku, uh, several years later and they sit in their office, they had about eight units of youth cause they wanted to be very scientific about measuring their plants. So this is one example of how we converted people who were very resistant to buy something and to people who wanted to buy a by attacking what their resistance was. So now I'm going to talk about something a lot more personal, which is LaunchDarkly. So LaunchDarkly as a company that, uh, I co founded and what we do is feature management. We allow you at scale to manage what features are on and off for different users.

Edith Harbaugh: When we started, nobody knew we existed. Uh, just as a fun fact for a long time. We got a customer a month and that customer was a lot of work. So the very first thing we did was we just did everything we could for awareness. Uh, we didn't show hacker news, we did blogs, we went on product hunt, PR, talked podcast articles, core message boards. And uh, a question I get asked now by startups, it's like, well, which one worked? And I was like, oh gosh, all of them. Like we did this huge product launch that we had huge expectations for and we got exactly one customer, but that one customer was envision and envision turned on to be a really good customer and that got us more customers. So my one bit of advice is the early days are super, super, super hard and it is a grind. Anybody who uh, knows better, please give them advice to me cause I need that help.

Speaker 3: Okay.

Edith Harbaugh: All right. So when we went on Show Hacker News (Show HN), the, the feedback was pretty violently opposed. Um, this was an actual quote. I can build this in two hours as booleans. Uh, and then there was like, you could go back and look at the hacker news thread. A lot of people be like, what the heck is this? This is stupid. And one of my favorite quotes is silicon valley has truly jumped the shark. And a lot of people I love Monty python said, no thanks. We've already got a feature flagging system. This was actually really helpful. So what anybody who had a feature flagging system that would let us look at it, I would go over and look at it. So I want to say, uh, there are some very good friends who I would go over and look at their system and I would see all the stuff that they had that we didn't have.

Edith Harbaugh: And I was like, Oh, you know, we really need to also build a user view. Or somebody else said, you know, we like our feature flagging system, but what it really lacks is a role based access. This is a real hurdle for us. So we did two things at the same time. We went away in all the features that people had that we didn't have, we weren't built and also all the hard stuff that nobody really wants to build. A third developer, um, SSO custom role based access. We went away and built that and then we did something that surprised some people. We published our blueprint. So we started writing articles about here's how to build a feature flagging system like we sell to developers. So what we did wasn't a secret. So what we're trying to do is we were trying to say, hey, here's how to build it and you have better things to do with your time.

Edith Harbaugh: So this is a picture of a hammer. Uh, I got an engineering degree from Harvey Mudd College. One of the projects we had to do was they gave us a hunk of wood, a chunk of plastic, and a chunk of metal and a blueprint. And from that they said, go build a hammer and not just a hammer but a hammer that matches this blueprint and you'll get graded. And they will come out with calipers and you know, measure your hammer. So it's not like an artistic project where you can interpret what a hammer means to you. It is your hammer, better match. And as engineering professors getting measure you, the hammer took endless amounts of late nights in the shop, particularly when you would try to a heat treat, the metal end and you had to anodize it. And then it went to form. After this I decided that I never wanted to see a hammer again, that I made myself and I was happy to go to home depot.

Edith Harbaugh: So there's two types of engineers. There's the ones who say, I want to build every hammer every time, and the ones who say, I have far better things to do with my time and I am happy to buy a hammer. You want to, if you're talking buy versus build, you want to find the people who say, I have better things to do with my time. Uh, by the way, this is not my hammer. This is my friend's hammer. I was so disgusted. I threw my own hammer away, but he kept us. So we published the blueprint. This is actually an article we wrote for AWS about how difficult it was to build what we had done. And this was really good because then we started getting people thinking, the original people who said, hey, I could build this at two hours. We're kind of like, Huh, to do this at scale.

Edith Harbaugh: This is actually a lot harder than I thought. Um, at this point when we published this article, we were doing about, uh, 3 million feature flags a day, which we thought was a lot. People sort of say, no, this is, this is pretty difficult. Um, so we really emphasize that you can build this. Why do you want to build this? And this gets into an antique tactic. A, I'm an engineer. You never want to challenge an engineer and say you can't do something. You know, cause if you tell someone you can't do this, they're kind of natural reactions where like you say, I can't do that. I can do that. Um, so this is a scene from Monty python where, uh, he's basically saying, uh, come back here. I'll bite you. I'm not dead yet. Um, so we never wanted to say you can't build the system. We always said you can build it, but do you want to, which is really subtle shift of, of course you could do this. That's not what you should be doing.

Edith Harbaugh: So here's the anti tactics and by versus builds avoid, um, if somebody is just not interested in what you're doing, don't get in a religious fight. So for us, uh, or in our early days feature flagging versus feature branching was still kind of a religious conversation where people just kind of said, I'm not interested in feature flagging and stop. Yeah, I was sometimes tempted to argue with them. I never won that argument. What you want to do is find somebody who's already interested in what you're doing and then have them get on board with using your solution. Another anti tactic is to say that you're not smart enough to do this yourself. Again, you're getting into an antagonistic relationship with your customer. Of course they're smart enough they can easily do it. You just don't think that they need to do to do that when they can just use your solution. And other tactically anti tactic we do, which didn't work, is if somebody has a really good homegrown solution and yours isn't up to snuff, they're not going to move. There's a really high switching costs. So every time we tried to say, hey, here's is great move to ours, it's just me missing this one feature that you really depend on. It just totally failed.

Edith Harbaugh: Things that worked for us. Um, every time somebody had an objection, we would go away and write a blog post. If you look at lunch directly, we have a ton of blog posts about how to scale, why real Tim is necessary, why things matter. Um, it to have an audit log. If one person is having an objection, it probably means that 10 other people in the company are feeding that objection to them. And what you want to do is have a really digestible bit of content that they can sport on. So if for example, if one person says, hey, we have concerns about scalability, you don't want to just answer unlimited email. You want to say, great, here's how we scaled to 10 billion flags a day. Please share this with whoever else in your company has concerns. We also kind of attacked head on this idea that feature flags cause technical debt. This came up over and over. So we actually wrote a blog post called uh, feature flags at attacking technical debt. So that way every time somebody had a concern about technical debt, we actually came up first. And SEO,

Edith Harbaugh: We did a ton of best practices and we also created a site called featureflags.io and all of this was again just to try to get everybody to look at us as the trusted resource for feature flags and in featureflags.io we also kind of told people how to build themselves. And this was actually a very, a, a sneaky tactic. Rowers or theory was that if you built it yourself, you would like feature flagging and you would eventually get tired of maintaining it yourself. So like our entire posture has been, if you want to build this yourself, great, we think in a year or two you will get tired of maintaining and come to us. So we never said don't do it. We just said, sure, here's how to do it.

Edith Harbaugh: So I hope some things you've learned from this is that buy versus build can be one. You just have to kind of unpack how people are thinking about your product. You always want it to be a friendly conversation. You never want to, you never win. If you out shout in a customer what you really want to be is seen as a trusted partner. So now we hear things from our customers like if feature toggling is at your core business, don't build it yourself. And I think this is a really good transition for the industry. Uh, I am old enough that, uh, there's people, the audience, remember that our issue tracking system back then was built in lotus notes like our QA, our QA QA manager had literally built it himself. Some people now are just like, what? Why would anybody do that? There's Jira, there's this, you're tracking systems you can buy off the shelf. Uh, I think there's a lot of stuff in the developer ecosystem that in 10 years we'll look back and just be amazed that people built it themselves. Not that they can't, but why should they? So thank you everybody. I hope you learned some lessons about buy versus build. I kept this deliberately a little bit short because I know I'm right before lunch. Um, so I will take a couple of questions and then let you have some delicious fruit.

Edith Harbaugh: Any questions?

Speaker 4: Hey, thank you for the talk. Eh, I'm, I'm running a product for the company and basically we have, we're seeing the same things, Eh, and I'm wondering what do you do with the, with customers or prospects that essentially is already built, have already built a system for themselves, but it's a obviously less mature or at least in a good case, it's less mature than, than what you have and eh, and you are seeing big companies. In our case, we saw someone like Uber that has built something and you know that they're heading to the long calls and, and you just want to help them kind of get convinced to a,

Edith Harbaugh: Yeah, that's great. I mean, so that gets into pride. So internally we kind of break down our personas and we actually have a distinct persona, which we call the, um, the Kombucha crowd. Uh, which is the Airbnbs, the Uber's the, that are just the googles that are just always going to build their own system for pride. You can waste a ton of time trying to talk to them and they're just not move. Where we found a lot of success was that people who, I'd say we're more resource constrained, like, but did it had an infinite money so that they would look at their roadmap for the year and somebody would say, why do we have this vipers of team over here working on infrastructure and when we could buy that off the shelf. So to answer your question, we found the most success with just people who did it were more resource constrained and we've tried to steer clear of just the huge giants who have so much embedded pride in money that it's really hard to move them.

Speaker 4: If I may, uh, kind of, uh, ask a follow up on that. In those cases at least, at least the, in a lot of times from personal experience, a very emotionally kind of invested B when it comes to like an engineering team. We were, we were walking usually with the marketing tech, Eh, so there is a part of engineering that is already dedicated. Like you have four or five engineers and you can replace them and let them do other things. But they're already working on, on marketing tech in our case. And their mindset is just, you know, we have to keep maintaining that.

Edith Harbaugh: Yeah. We have had the most success when um, somebody has built a system and is absolutely sick of it lately. Like they thought it was a weekend hobby and suddenly it's become their life and they're like, this is kind of dull or a when somebody has shifted jobs and like for a good example is somebody moves jobs. They say, hey, we don't have a feature flagging system. I want when I used to have, and that's a very quick sale. Um, if there is people who are embedded and who think of it as their job to do this, it's very hard to dislodge their [inaudible].

Speaker 4: Thank you. How can I be a more effective internal advocate for buying versus building like as a pm? I can't necessarily, I want to tell engineering like, this is not worth your time. I'm gonna run into many of the same arguments that you listed. I think the version that I heard was like, well, if slack wouldn't do this themselves, then we shouldn't do in ourselves. Um, just fun. Uh, so yeah. How can I, how can I better make that argument to my stake stakeholders?

Edith Harbaugh: Yeah. It really comes down to resourcing. Um, so you have to have been an engineer. I've been a product manager, uh, you have to really appeal to, hey, is this the best use of our time? Like, great, I know you can build this. Is this what we should be building? Like you, if you say stop doing that, it's comes off as a little Breesa. But if you say, hey, there's this really important initiative over here that we should be working on, and this is where all the, the, the company's future is, that's a much better persuasive argument. And I say the seven talked to many people who have given up their own feature flagging system and there's always this transition period where you're sad. I say as an engineer, like we had to shut down projects and you kind of this morning parade, we were like, you're a little sad. You've, you put a lot of work into this, but then you're like, okay, I'm working on this other new thing that's better. And then people are happy.

Speaker 5: Thanks a lot for the talk. I am a pm for very similar tools have been through this too. One thing I really like where you said it's about communicating, um, like, hey, are there reasons or here's where you go look at it. So we are an enterprise to be solid ab testing and personalization tool. So what, what has worked well, like, especially when, um, like there's a enterprise sales field or others. Um, is there a playbook that you have, uh, for communicating it properly? As in like you have a lot of belt of materials? Like what has worked well?

Edith Harbaugh: Um, just a, I'm sorry, does your question to selling enterprise buyers? Yeah. Yeah. Enterprise buyers. This is tricky. When you're a small startup, the first thing you have to realize is that your is not you. Your buyer has a lot of concerns about security, stability and reliability and they're sometimes putting their job on the line to use you. Um, so we always just said like, we take this very seriously. Like we, we got our SOC 2 compliance when we were a very small company, we didn't joke around with security. Uh, we emphasize scalability, uh, for, for context. Now we serve about 80 billion feature flags a day. So I think the number one thing you can do to sell it to enterprise buyers is recognize that their needs are different and recognize that you are there to help them and don't blow off when they say, Hey, I want you to fill out this 30 page security questionnaire because they actually need that. I think I have time for one more question and I know it's lunch time. So any final questions? Well, thank you everybody. It's great to see so many friendly faces in the audience.


Keep me posted on Empower 2019.