Thriving in Chaos with Greylock EIR Stephanie Hannon and General Partner John Lilly | Greymatter

Hi, welcome to the Gray Matter podcast. My name is John Lilly, I'm one of the partners at Greylock and this is a really special podcast for me today because I get to do it with Stephanie Hannon, who I've known for more than 20 years now. She's recently joined Greylock as an executive in residence and she's got a ton of interesting experiences that I'll let her talk about. So welcome to the pod, Steph.

Thank you. It's a huge honor to be here.

Just briefly, could you tell the listeners your two minute background from Stanford to here?

I'd be so happy to. My journey, I think has been about working with the best people to build really ambitious products all over the world. So I started my career at a company called Granite Systems right out of Stanford, a small startup founded by Andy Bechtelsheim, who founded sun, and David Sheridan, who was a Stanford professor. And the ambition there was to build a gigabit ethernet switch, which at the time was revolutionary, but a single Asics, so it was a sixth of the number of chips of any device before. And that was Andy's big dream. My background was in operating systems, so my first job was really building operating systems for that switch.

I remember you like you graduated a couple years after I did out of Stanford. You were the first person I think that went into a hot startup that got acquired almost immediately of our cohort.

Yeah, I think it happened only three months after I joined. So suddenly I went from being a 17 person startup to Cisco Systems which was a massive entity. And the profile of the engineering work changed because we suddenly wanted part of this ecosystem and have Cisco's operating system on our devices. So those were my early days developing network devices. I left Cisco in 2002 to go to HBS. And when I graduated from HBS, I was trying to look at the landscape of all the interesting entrepreneurial tech companies out there. And I joined Google. It was just a few months before its ipo. A lot of my great friends from Stanford had worked there. I thought they had big ambitious dreams. And so in my almost 10 years at Google, I worked on Gmail and the ambition to give everyone and unlimited storage for email was a significant challenge on infrastructure and how you reliably store email and handle spam and all those different things and then also scaling it internationally. So I think I took the first consumer product, which was an English only product, and made it work all over the world. And that's a lot more than just the interfaces. It has to do with the encoding and the dictionary and bidirectional languages, and even colors. Gmail had a lot of colors in it, and how can we be sensitive to that all over the world?

Why'd you make the mood of product management?

I think I was at Cisco with the very best hardware engineers from sun and the very best PhD students from Stanford, and I loved building and I saw a lot of positive impact. I think I built great things, but I didn't feel like I was uniquely contributing. I thought there was something about the intersection of talking to customers and clients and talking to engineers and interface designers and support and operations and sort of product management is a mini CEO of a product. And I thought I could have more influence and also uniquely take advantage of all my different skills.

I think the most interesting journeys that I've noticed are people who are trying to figure out what thing can they be the best in the world at?

Exactly. But it's a hard transition. I just always tell people, because when you're an engineer, it's measurable. You write code and then it doesn't compile. Then it compiles and it runs, but it doesn't work. And then you make it work and then you launch it. And you can sort of see this measurable progress. Once you move into product management, it's very ephemeral and you're sort of trying to influence people and move a project forward and make big decisions, but you're not actually hands on the keyboard anymore.

You've had plenty of experience on the business side of things doing product management at Cisco. Like, why did you decide to go to hbs? Why'd you do that?

I think a lot of it had to do with friends who had gone through the experience. And I felt my competence and I felt very solid in engineering and things that were happening on the technology side. But I had never written a business plan. I had never looked at a finance sheet. I had never tried to raise money. And my dream was to be an entrepreneur, start my own company. And I thought I could develop this set of schools at HBS that would be really valuable.

So you were chief Technology Officer for Hillary for America, Hillary Clinton's campaign for president. How did that happen? Like, how did you go from. By that time, you'd been back at Google for a little while, working on some civic initiatives. And how did you connect up? And how did you become the CTO for the campaign?

My journey to become the CTO for Hillary Clinton started in Chicago. So In January of 2015, I was at something called ORDCamp, which is a Meetup of creative people, technology people. It's a crazy event where you really just stay up all night. Like you don't really leave the space for, for 48 hours. And a lot of the Obama alumni live in Chicago. And so there are a lot of people from his technology team at this event. And I think around three in the morning, somebody approached me to talk about this opportunity. But at the time, I was running the product team for google.org, and we worked on a lot of civic stuff, civic engagement, elections, transparency in elections. And when they approached me, I really thought they were trying to make recommendations to me about Google and the work Google was doing and how Google should take more active role in the 2016 campaign. And it only over time became clear that they, they were actually trying to recruit me to interview for a job. But nobody would say that she was running early 2015. Everyone was speculating, but she certainly wasn't declared. And so they weren't like, hey, do you want to be Hillary Clinton cto? It was a very circuitous conversation. And actually it was only a few days later when they put me in touch with really the campaign manager and I started a series of interviews that it became clear who it was for and sort of the ambition of what the job was.

Why did you even take those interviews?

I think the best product people are curious. And I was really curious because the concept of engineering inside of a campaign is new and it felt like there was a lot of opportunity there, but it wasn't. You know, I don't. In my core, I'm not a person who's been really active in politics. It's not. I grew up in the D.C. area, but I didn't spend all my time reading about it or being immersed in it. So when I took the calls, I didn't actually think it was a likely outcome that I would take the job. But I thought you should always, you should always say yes to learning. You should always say yes to exploration. When they offered me the job, I was actually in Japan. I was there for the fourth anniversary of the Tohoku earthquake and tsunami, and I was speaking at a convention because I also worked on the disaster response product space at Google. And they called me and I was like, oh, my gosh, this is amazing. I'm so excited. But I want to meet her. That was my thing, is I had never met her. And I think you're going to give up so much to take this big risk and leave Silicon Valley and do a really, really hard job. I wanted to fundamentally Feel like it was for the right person. And so then I had to get all the way from Japan to New York. And then the whole time, you have 20 hours of flying to think about what you're gonna say the first time you meet Hillary Clinton.

Did you nail it? Got off the plane, went and talked to the secretary. Did you nail it?

Yes. 100%.

Yeah.

Yes. I made. You know, I was intimidated. I just made nonsense. I'm six feet tall. She's much shorter than me. But I felt like she was such a huge presence. But for me, being a woman who works in Silicon Valley, I think role models matter. And when they offered me the job, it was really easy to say yes, because I thought the journey to try to put the first woman in the White House would be epic and really important. And so it was very easy to take the job.

What did you think the job was when you took it, the CTO job?

My belief was I would be responsible for the technology strategy and implementation to support the mission of the campaign, which is make Hillary Clinton the 45th president of the United States. But I was totally clueless in what that meant because I had never even set foot in a campaign. And I think there's advantages and disadvantages to that that we could talk about. But it's pretty terrifying because I knew it. The very core. I needed to hire a team. And this was April of 2015. There wasn't a lot of energy around the campaign or the election. Nobody knew who was running, what kind of race it would be. And, you know, I walked in there, and I only had six days between being hired and her announcement. And if you remember, she was gonna go in a Scooby Van. So she basically drove from her house in upstate New York all the way to Iowa, and that's the way she launched her campaign.

Wait, what does that mean? She went in a Scuby van? She actually did that?

Yeah, she called it a Scubi van. She just got in her car and drove across the country and met people along the way, and it was her way of launching the campaign. But, like, I started on a Monday, and the Scooby Van was leaving at noon on Sunday. And we didn't have a website, we didn't have a payment processor. We didn't have a team. Like, I was there by myself. Yeah. So there was a team called the Groundwork, which is a fundamental technology platform partner that had already been selected. And they were heroic in that first week. And they are really the reason we launched a quality website that stood up to all the traffic that week. But we also had to pick a payment processor and we had to figure out, you know, in the early days of a campaign, you're really trying, trying to raise money so you can fund the campaign. That's very different than what I was doing in the last four days of the campaign, which is a lot of get out the vote, making sure people show up to vote, and anyone who's able to vote can. And so, yeah, so I just used the groundwork. We had volunteers, anyone we could get to help in those six days, but it was just really dramatic. And we did, and we launched a really beautiful website. But then my challenge is now I have to build a real team. And when you work at a place like Google or Facebook, you just have an infrastructure around you that makes recruiting happen. And when you're at a startup or when you're the CTO of the Hillary campaign on day one, you have nothing. You don't have sourcers, you don't have recruiters, you don't have interviewers, you don't have an interview plan, you don't have questions. And it was very scary. So I had to figure out how very quickly to start recruiting people. And I pretty much used in the early days, Google friends to recruit, Obama alumni to recruit, and the team at the groundwork to recruit. But the problem is sometimes they gave me all different answers on the same candidate. Like, we had the same quality bar, same culture, same intent. And so I think what made the process move forward was just the quality of some of the early people. I got to join me, the deputy cto, Kyle, and the chief product officer, osi, who's a longtime friend from Google and Stanford. And those two people are powerhouses. And they were two of my earliest hires. And between the three of us, we were able to start both defining our culture and our recruiting process, and then also kind of filling the pipeline and getting people in the door.

I teach a class at Stanford in the business school about HR and for startups. And one of the first things we talk about is how to recruit. I think the image in people's head is, well, recruiting is you sit back and you look at these resumes and you pick the best resumes. Oh, that person would be lovely. And this person would be wonderful. And you just take the best ones. I think they have this picture because, like you, they were at Google and Facebook, who are who become these megatrons of recruiting. But startups, nobody wants to work for you. Nobody believes it's going to work. And so you had to flip this bit.

It was clear early that it was going to be hard. I think a lot of it just because it felt so far away geographically. No, no, no distance from April of 2015 to the election.

Oh, time.

Time wise. But the problem with engineering as we both know is you have to front load development. And so for me it felt very urgent because if you're going to introduce a new field tool, like one of my obsessions was getting people off of paper canvassing to mobile canvassing or getting people off of canvassing at all to using text based activation. And if you're going to make significant investments in engineering, it has to be up front. And when you go after the primaries from 10,000 Vol volunteers to 300,000 volunteers, if you don't have the tools in place, June of 2016, you miss the window.

Wow, that's the literal scale from 10,000 volunteers before the primary to 300,000 after.

Yeah, I'm hand waving here in the studio, but yeah. And so like they're not going to introduce a new tool set in August of 2016. Sure. And it was hardest to recruit during the primaries, were much more competitive than expected and so we couldn't grow as fast as I wanted. So I guess time wise it felt very far away. So I knew that would be a challenge.

I guess we saw each other the day before the announcement and then I probably didn't see you in Brooklyn again for maybe six or eight months. Your technology team was 30. Maybe by the time got to the election it was 80 people.

Exactly.

So that's a lot of growth.

So much growth.

One to 30 in six months and then to 80 over the next year. So how did you get people to move to Brooklyn? Like everybody's in Brooklyn.

Exactly. Yes.

How'd you get people to leave? No offense, but your snacks were done. Terrible.

Yeah, it was really hard. And I just want to say, I always said in those early days I wanted a T shirt that said excuse my air of desperation because I would just go. You know, I also moved to New York for this job so I didn't have a strong network. You know, I wasn't immersed in the New York tech scene. So I'd just go to any event, any meetup. And I used my friends like Andrew McLaughlin who worked for Obama and he hosted a huge party for me. I got some of my first recruits from that. And in the beginning it was also a lot about the network because I got Kyle who came from Optimizely. Optimizely was really born of the 2008 campaign. Once Kyle came, a few people came. We got this amazing product leader, Deepa from Charity Water. When she came, a few people came. I had a bunch of my friends from Google. And so I'd say the vast majority of those 30 first hires were all people who knew one of us who had already kind of made the leap. But it was really hard because you're just asking them to give up so much. Not just a fraction of their compensation, but the intensity of the work is something it's almost impossible to describe. Having been at Facebook and Google, I thought I knew intensity. But the speed of things happening at the campaign and the drama of the moments and the big of things was huge. And you're doing it on this global stage where everyone is watching you, like any misstep will be exploited and magnified. So some of my other tactics were just having people come sit with us. I was big on like, oh, we just need two weeks of your time. Oh, you have some expertise in fraud and we have this donation system and please come sit with us. Or you know a lot about analytics infrastructure and we're trying to make some choices about which provider we're using. So come sit with us. And then most people who got into the energy and like felt it and sat there with us, kind of getting them to make the leap was easier once they felt it. The other thing is I just, you know, probably any entrepreneur does this. A sweet spot was people who had already left jobs. Like, I spent way more time trying to find people who had just left Google or Facebook because trying to get them to leave was hard. But if you were just searching and you're like, oh, I'm taking a year of travel, you know what, traveling in 2017 will be nice. How about in 2016 you come and help us? So those are some of the tactics.

Did the come spend a week or two with us thing work? Because it seems to me that that has a high potential to backfire. Because my observation was that your team worked more hours than anybody I've ever seen over a longer period of time. My observation is like you guys were working 16 or 20 hour days for a year and a half and startups don't do that.

Yeah, it might. I wouldn't use the word backfire because it's better that you know up front what you're getting into and you better viscerally experience it because it's hard. My hiring practices changed a lot over the year and a half because in the early days I didn't know how to check for ability to deal with chaos and change. Because you can imagine when I only had three or four engineers, I was making very different decisions about what we work on than when there was 80. And so some days we'd be like, oh, the most important things is commit to caucus. For iol, we're building a commit to a caucus experience because there's behavioral science that says if you say you'll show up, you're more likely to show up. But then all of a sudden we'd be working on that and an opportunity would come up to launch a new payment system or to launch a new partnership to raise more money. And suddenly money is the most important thing. So stock caucus, go donations and you just. The early days of churn, that was my hardest decisions as a leader. You don't want to churn people, you don't want to move people around. But sometimes we just had to make things work. And so testing for that and people feeling that when they were in the office sitting with us was better because then they knew what they were getting into.

Testing for ability to thrive in chaos. How did you do that? Did you just put people in chaos and see how they did for a week or two?

No, no, I don't think we could assess that on site with them. But during the interviews we would try to ask about experiences like when did your project change suddenly? How did you react to it? What's the most stressful environment you've been in? How do you deal with stress? Just having people talk about it. You sort of get a sense of who needs some continuity and consistency and who has been in environments where there was extreme deadlines and tight pressures or limited resources and how they reacted to it. So I think it was. I think we all know about how you test for technical expertise in interviews, and it was just trying to figure out how to assess some of those software skills.

So I always think about the question of whether you can get enough signal just directly interviewing people versus referencing and going around the side. Did you get enough signal on ability to deal with stress from the candidates or do you have to reference to figure that out?

We always did references, so we were very consistent about that. I always again, attribute it to the amazing leaders on the team who held the high quality bar in the face of the of absolute chaos. But we made some hiring mistakes and sometimes during the campaign we did performance plans and sometimes we let people go. And I think that's one thing I thought was really important because we didn't have a lot of time to mentor. We didn't have HR resources, we didn't have training modules, we didn't have time. So if somebody wasn't a fit, I tried to figure out really fast and deal with it fast.

Getting fired from a campaign must be a brutal blow because it's so emotional and you're so tied into the mission. That's gotta be even a tougher conversation than just a job.

Yeah, it's about the worst thing I've ever gone through as a people manager because the mission is so big and the campaign feels really special and it's kind of once in a lifetime. And so when it doesn't work, it feels really terrible.

Yeah, so one of the times I came to visit you, one of the things that was really obvious, they didn't really look like a Silicon Valley team. And what I mean by that, as I said, is that they were just more diverse. So more women, more people of color. Is that what you're trying to do? How did that happen?

I absolutely set out to build the most diverse engineering team possible because I felt like we were trying to elect her to be the president and we wanted to represent the constituency, the people we were trying to activate, to vote. And also I just think diverse engineering teams are better, more creative, better products. Everything's better. But it's a hard thing to do. And so I, in the initial days, just spent a ton of time going to events where women coders congregated and spent time together. And so it was intentional, but it wasn't easy. I think I told you this story where when I was about 10 or 12 people, a good friend of mine from Google came and sat with us for a week. And I was trying to recruit her. I was like, come be part of this early team. It's going to be awesome. And after she spent a week with us, she's like, steph, I'm sorry, I had a really bad experience here and I would never work with your team. And so I asked her to tell me more. And she had sat with somebody at lunch who had disparaged Googlers. She had been made fun of for not having a GitHub account. And she had done some code check ins that people just trivialized. They said, oh, why bother Code review this, it's so small. And those three things, to hear those, as a woman leader who's trying to build a diverse engineering team and to feel that this woman had a pretty negative experience sitting with us is pretty traumatic. Inclusion doesn't happen simply because you're trying to elect the first woman president. And because you're a woman leader yourself. And so hearing that, I immediately tried to take action, starting with open and transparent conversations with the team, asking everyone on the team to go to unconscious bias training, changing our interview processes so there was a woman on every interview in pair interview loops. And then we also just tried to be creative about the interview process itself, to be open to more diverse approaches. So not just whiteboard interviews, but we had homework so people could work on things at home. And then in our interview debriefs, we just focused a lot on soft skills and emotional intelligence and how people felt and experienced the candidates. And so, you know, in the chaos of the campaign, trying to be intentional about all those things was hard, but I think it changed our ship and changed the way people felt and the way the team operated. And so at our peak, we had about 35% women, which was sort of an epic accomplishment. It looked different and felt different than other engineering teams I've worked with.

You took the team to unconscious bias training. Did you get pushback on doing that from the team or from other people? Did you have to hide it from the people you worked for since it felt like you weren't gonna be working on things that directly contributed to the campaign?

Well, I'm proud to say unconscious bias training was available and an important part of the campaign for all teams. So I think as a campaign team in general, we were very diverse.

That was a top down priority for me. The whole campaign?

Yes. The unconscious bias training was offered to the whole campaign. So I would say there was no pushback from the wider campaign. I would say there was pushback from my team because a lot of people said, we don't need it. Like, we're sensitive to this. We don't need it. The people who said that were often the people that needed it the most. Like, there was one terrible era in the campaign where there was one walkway in our cubicle area and we were sending three people to a desk. So just imagine like, like squashed people everywhere.

Three people on a normal human sized desk.

Normal human desk?

Yeah.

Like, monitors everywhere, elbows everywhere. And also, we had no money for standing desks, so everyone built standing desks out of cardboard boxes. But it's just our office, and one aisle was just all men. And for some reason, I can't even remember why it started, people started calling it dude bro Lane.

Dude bro Lane? Yeah, that's not the place you want to work.

And it started to make people uncomfortable. And of course, I fixed that by breaking up the Lane and like making it more diverse and like having a talk about this isn't a language that that's okay.

Not letting it become a district.

Yeah, exactly. But in the end I made it required for everyone and it just also created a language that we could talk about. I think a lot of creating an inclusive environment is making it okay to call people on stuff, like you called me this word, or you said this in this way, or you talked over me in this meeting, or you didn't invite me to this meeting, making it okay to talk about those things. And that's what the training helped us with.

Let's switch topics a little bit and talk about products and how you thought about products. So over the course of a campaign, I imagine lots of projects. Did you think about projects separately from products and how do you think about all that stuff?

I think we built a lot of products, but there's also projects that we ran that weren't things we built inside the campaign. Whenever possible, I liked to leverage other technology systems or other tools. You know, it wasn't my dream to build everything from scratch, especially because we had limited resources and times. But often we had to build things simply because we wanted a certain customization or an integration with systems and tools like VAN as the Voter Action Network. It's an important database which has the repository of all voter information. Integration with VAN was really important for almost everything. So examples of products we built are a mobile app for every caucus captain in Iowa. Because historically there's no real time knowledge for a campaign manager about what's happening in each caucus site. And so if we put a mobile app in the hand of every caucus captain, we could have this real time dashboard of what was happening and also the caucus captain. There's a lot of complex math that goes into figuring out when people are standing in different corners, what number of delegates you're going to get and how it all works. And then the most important thing about putting a mobile app in the hand of every caucus captain is that we could look at our results versus the Iowa Democratic Party. The IDP had an API that we could read their results and if there was discrepancies between what our caucus captain reported and what the IDP reported, we could raise alerts and issues. So that's an example of a product and it felt like a product development and launch. Another example is Megaphone, which was a tool we built for the Iowa caucus, but we ended up using it and became a significant part of our GOTV effort all the way to the get out the vote, get out the vote. And that was really about, can we text message voters to remind them to vote, to tell them where their polling location is, or to recruit them to volunteer?

And.

And it's simple, but if you want to do it at scale and the text messages have to be one at a time, building an easy interface for that was a true product development. Another example was a voter protection platform. And so there's tens of thousands of people standing inside of polling locations on the day of the election, lawyers monitoring the polling locations, and there's a whole credentialing process to get them there. There's a whole ticketing and triage system for voter protection events, and then there's sort of an Insights dashboard. And so those are examples of, I think, true products that we built throughout this campaign. But an example of a project would be how can we work with Facebook and Snapchat to get more people to register to vote, to get millennials to register to vote, to get people of color to vote. I'm very happy if none of that engineering happens on the campaign side, but it's about kind of the partnership and the mission and trying to get other technology companies to get on board with it. And then another piece of the technology work was just activating people in Silicon Valley. I used every single human being I could to both recruit, but also to just come and visit the team and be a part of motivating the team and recognizing these amazing people who went on this journey.

Let's talk about the get out the vote text messaging thing. I know before you took the job, you probably talked with Harper Reid, the CTO for Obama for America in 2012. What's your sense of how the job in 2016 was different than the job in 2012 from a technology standpoint, I.

Do think the proliferation of mobile devices and the smartphone in everyone's hand made mobile at the forefront of this campaign. In a way, I don't think it was quite as big when I talked to Harper. And that's why I was so excited to not just think of text messaging activation, but also some of our biggest product plays where how can you get a supporter of Hillary to activate their friends in battleground State? And there were creative things we did with mobile apps and games to try to figure out if you knew people in our persuasion universe or activation universe. So I think that was a really, really big difference between 2012 and 2016. Another big difference was just analytics infrastructure. Harper would say this. Everything in 2012 was in a closet in Virginia.

You mean A machine closet. The servers were all in a closet.

Exactly. And so we use a system called Vertica, which is the core infrastructure for our analytics. And we made a big effort to move it into the cloud this cycle and to have it be distributed and on Amazon and reliable and scalable and cheaper and a lot of different things. And that was a significant. Probably one of our biggest engineering projects, probably took five months of eight engineers.

Just moving analytics into the cloud.

Yeah. And there's a whole tool set associated with it. And then there's all of the data, the historical data, and then you can just imagine the joining of data and all the jobs because you're getting constant. To successfully model voters and decide how to spend your resources, you want obvious data sets like the voter files. Voter files are public, but they're not in easy standard formats and they're hard to parse and they're different levels of frequency. But you want to join that data with lots of things that are happening real time in terms of donations or events or visits on the website or mobile app downloads. And you also want to join that with lots of data around education and spend and TV behavior and things like that. And so there's just a massive amount of infrastructure to manage all of the voter information.

And then I guess also the nature of the opponent was quite different. So things that are clear in retrospect now is like just the incredible level of spend and attention they had on Facebook in particular was quite different.

Our candidates were so different because I think between 2012 and 2016, because Obama people were very proudly and loudly for Obama. They were so excited to stand up and talk about him on social media and be active and vocal. And I always said in this campaign it was hard because people were more quietly for Hillary, like, you know, I'll vote, but I'm not going to activate and get involved in the same way. And so that was, I think, a unique challenge of thinking about how you spend your time and energy and how you increase engagement and things like that.

We could probably do a whole podcast on why that is.

Yes, a lot of the innovation of Obama was his adoption of social media, and I think he was really, really far ahead of of Romney in 2012, our competitor and his approach was different. And I think we came to know much after the campaign about the investment in technology and not just the spend on Facebook, but the creative use of micro targeting on Facebook. And I absolutely think that our competitor in this cycle used technology significantly to their advantage. And it was very different from 2012.

If you extrapolate forward to 2020, the next cycle. How do you think the job for the CTO is going to be different in 2020 than it was in 2016?

I think, first of all, the Democratic Party and people who care about the outcome of elections in 2017, 2018 and 2020 need to make significant investments in technology. Now. We can't have this boom, bust cycle where there's a huge spend around a presidential and then all the technology and people disappear, and then 18 months out we do it again and then that disappears.

And you think this is asymmetric between the Republicans and the Democrats? The Republicans have a more consistent spend, a consistent investment?

It would be impossible to know for sure, but it feels like there's a more consistent financial investment and maybe some of it happening outside of the party itself. It feels that way. And I think for us, for the Democrats, the exact moment when we need to be the strongest, when the election was over, there's not enough funds to build a significant engineering team at the DNC or any of the committees. And so I think, you know, the DNC just hired a new, excellent CTO named Rafi, and he wants to.

Rafi Krikorian from Twitter and Uber.

Exactly. And he wants to create a different, better future. And if he can be doing that in hiring great engineers in 2017, then I think we're on a much better path than we were for this election cycle. So that's really important. And then I also think there's just figuring out what are the right problems to solve. There are really brittle pieces of technology infrastructure that exist and are critical simply because a partnership was made with the DNC a long time ago. And if we want to extract those, if we want to improve them, if we want to build around them, those kinds of big, meaty projects have to be identified and started now because they won't be easy to displace. And so I think just a coordinated effort and investment now is going to be what's needed to create a different technology platform for all the candidates in 2018 and 2020.

And you think four years from now, the TV spend will be way down and all the spend will be around targeting and messaging people directly. Is that the way to win, you think?

I think you can be more effective in reaching people in the mediums they want, with the message they want and by people who are meaningful to them using technology and online spend. I think you'd see even in this cycle, the offline spend was very significant. And so I think another thing we have to figure out is how to shake up traditional means of democratic organizing and help people make or see the potential or do data and experiments, all the things we would normally do to kind of influence. So yes, I think the right thing in two or four years out is going to be much higher fraction, if not all spent online.

Would you do it again knowing what you do now?

I would absolutely have done this cycle again knowing what I did now.

You're not, you're not signing up for 2020.

Just telling the world.

Yeah.

Because I without a doubt learned more than I've ever learned. You know this because you're my friend. I gained an incredible amount of confidence because I just felt like thriving in this crazy environment with no resources and really no clue was hard. And then I have this amazing team of people. You know, I've worked on a lot of mission driven projects before, but the mission of this and the drama of this and what we went through together and just all the moments from, you know, different primary nights, to her first debate with Trump where we built a fact checker that sat on the website, to her becoming the nominee, to all of the, you know, being in the field and working with teams in Iowa, New Hampshire and Florida and Pennsylvania and, and you meet amazing people. People who work in politics are mission driven and they're very young and they're very amazing. And so I wouldn't really give up this journey for anything, but I don't think I could sign up to do it again because the toll on my physical health and just mental health was extreme. And I'm hoping to just use my learnings and scars to help create this better technology future for the whole party.

Yeah. Just as someone who's known you and paid attention to your work for a long time, it's amazing to look at you today and, and look at the work versus before. You've got all the same skills but a different level of confidence and intentionality, I would say.

Thank you.

You worked in Gmail very early. Gmail was kind of a revelation because it was this giant app that worked in the web browser, which seemed pretty weird. And there were a lot of things that you guys were figuring out with JavaScript and all this other stuff. But what are some of the things that you remember from your time in Gmail?

Well, I think one funny thing is people think the invite scheme, if people don't remember Gmail was an invite model and the invites, it was in Beta for 800 years and the invites were restricted and they were immediately selling on ebay. I Think in the early days of gmail, it cost $100 to buy an invite off of ebay. And what I think is interesting is people thought it was a clever viral marketing scheme. But the truth was we just couldn't afford to bring more people onto the system because it was really Paul Bucheit who had this innovation of putting sort of a group's interface on his inbox. And it was a 20% project and it was an experiment.

Paul, who's now a partner at Y Combinator.

Exactly.

And coined the phrase don't be evil.

Exactly. Amazing engineer and entrepreneur. And so that's what I think is interesting is we weren't using that as a way to grow, we were using it to restrict. And we couldn't get more machines at Google because when you're in 2004 competing with search for the machines or competing with ads for the machines, you have to be a very efficient, robust service. And in the early days, we weren't. I think that was an interesting story. I always am big on team culture and celebration. So we had Keith Coleman, who was another product manager on Gmail, bought Caribou reindeer dog toys, as one does. The code name for Gmail was Caribou. We had these really high ceilings in Building 42 at Google. We would raise one each time we hit a new million active users. And it was just a moment. I think I found this moose call online and I would do the moose call and I would come and we would like throw this thing up. And then one day Larry and Sergey were like, hey, stop measuring number of users. Measure active users. And so we had this dramatic cutting down of.

You pulled the moose down.

We pulled like 40 mooses down. And obviously we were monitoring active users as well. And you monitor a lot of important metrics as you're running a big service like that. But it was very like, hey, pay attention to the right metrics. And like we cut them down and we started re raising them and re raising them on active users tab. So that was really interesting.

We went through the same evolution working on Firefox at the same time. We focused on downloads and you focused on active users. And it just kind of evolved over time. What you think is important, but it's one of those things. Like you got whatever it takes to motivate the team at the beginning and just don't ever get too tied to the metrics.

Exactly. Another interesting thing in those early days was just how much email was getting forwarded to us from universities. We gave people a gig when the industry standard was 10 mega. And so people just lived their life managing their inbox and attachments, and suddenly they didn't have to anymore. And so people from all these different universities were forwarding their email into Gmail, and I thought, well, we could do it better. They don't have to have a Microsoft Exchange server at Stanford and then forward their email here. We should be able to host email for universities. And that became the gem of an idea that was eventually built into an administrative dashboard and the idea of hosting on other domains. And I went around to Stanford and Berkeley and all these universities and tried to convince them to let us host their email. They all said no. In 2004, that sounded crazy, but San Jose City College was the first to say yes. And so In April of 2005, we hosted Gmail with San Jose City College, and that became the gem of what's now known as Google Apps and this whole suite of business services that run for Google.

Yeah, I think in retrospect, that turns out to be a huge thing, and it's a play out of Apple's playbook, because Apple for the longest time had done discounts into the education market, discounts into universities and business. People thought it was kind of silly to use Macs and Apple, but people coming out of college were used to using it, so they kind of brought it with them. And I would say that Gmail and universities was the first seed of that for Google, which would lead to Google Apps and eventually sort of down the way to Chromebooks, eventually.

And I think it's a good lesson. You and I have talked a lot about what makes great product managers, and I think curiosity is the most important thing. And those early days, it was just sort of like, oh, where is this email coming from? Why is it forwarded? What can we learn about that? What's the opportunity? And I think digging into data and looking for insights is such a core part of being a great product manager.

So after Gmail, you went and worked on another thing that seemed impossible at the time, Google Maps. And you reinvented Maps in a lot of ways. And you got to work with a couple of special thinkers in that project, Lars and his brother Jens. Maps was a hotbed of talent because they had Lars, and Jens also had Brett Taylor working on it. So that's a pretty good crew. And then you signed up for another tour of duty with Lars and Jens on their secret communication project called Wave. But if you kind of play it forward a little bit, it looks a little bit like Slack. And some of the Things that exist now. Talk about what Wave was and what that experience was like.

Yeah, absolutely. So Lars and Jens invented Google Maps. And then a few years after it became part of core Google, they wanted to do a new entrepreneurial venture and their vision was really reinventing communications. And the way the world worked is you went to this tool for blogging and this for photo sharing and this for chat and this for document creation. Why do we have all these different tools when they're fundamentally similar paradigms? What I really love about the ambition of Lars and Jens is they wanted to do something that hadn't been done before, which was hundreds of people editing the same entity in a browser. And it was big ambition. And so Lars and Yen set it up like a startup inside of Google. And that was interesting. We could actually do a whole other podcast on that. Like, how do you be a big company and foster innovation and startups? Some special things about it. We were all based in Sydney, Australia. We had a different compensation plan and we were secret, so the rest of Google didn't know what we were doing. So we went on this journey and we had this big, dramatic launch in 2009. We got to be the keynote at Google. I o we have a YouTube video that 10 million people have watched, and it was awesome. And we, over the next year grew to over a million active users, but eventually we were canceled. And that was a pretty dramatic and hard journey for sort of a young startup. And there's a lot of interesting lessons there, but I think the themes of what we were trying to do with Wave live on in many different products like Slack and Quip.

Do you think it was a mistake for Google to cancel Wave ultimately? Do you think it was such a riot of functionality and a riot of user interface. There was a lot to deal with. Do you think they should have done something different than just canceling it? Should they have asked you to scope down something different?

I, of course, am biased. I will admit that. I 100% think we should have had a longer lead time. And I think the journey we went in those 18 months after launch, we thought we were building a replacement for email. I would say we were more building a replacement for document creation. And we had thought we were going to be a consumer product and we were really a small business tool. And so that journey took a while for us to get to. And I think once we pivoted to the more SMB, small group collaborators, screenwriters were using us, small technology teams were using us. We just didn't have enough time to Kind of reset, I think, the expectations of leadership. And also it was at a time when Facebook was going gangbusters. And I think maybe our leadership thought initially we were more of a social tool. And I think the energy was really around a new project that got created that year called Emerald Sea, which became Google and was Google's bet on social. And I really think we could have done something special. I would have been so happy if they held us. You know, we were about 60 people at the time we got canceled. I would have loved if they said you have to scale back to 20, and you have to reinvent these metrics around business opportunity and you have to work with Google Docs or you have to be an entity in Google Docs. I would have loved if we had a little bit more Runway. Sometimes when you're at a big company, you don't always know the reasons or rationale for what's happening. But I think we had the gem of something really special, and we could have kept developing it into something. I think a few of the most important lessons from Wave are good for product managers in general, which is we try to be everything to everyone. And we tried to tell this story about no matter what you're doing, whether you want to use it for group photo editing or you want to use it for document creation, you can just use this magical thing like, good luck. And we told this really big, amazing story, but then we only let 100,000 people onto the platform, and there wasn't enough density of people to really, like, gravitate towards certain use cases. And we didn't help them along by being very specific. Like, I always think if we had just done YouTube comments, which were such a hot mess in 2009, that would have been a good tactic. So that's one thing we learned. The other thing is we just wanted it to be an open standard. We didn't want to say Wave is this amazing thing, and you have to pick Google. So we did a protocol at the same time as building the product. So, like smtp, we built something called the Open Wave Protocol. And then we also wanted to be a platform because we learned a lot from Maps, and we knew a lot of the best innovation was going to happen outside of our product. And so we had this platform where people can build robots that sit on Waves and do cool things. And if you're trying to do something entrepreneurial that the world has never seen, and in parallel, you do product, protocol, and platform, it's too much. So when I look back, I wish we had Told the story of openness, but not tried to make it happen day one. And that was a hard learning.

It's not just that it's too much. I mean, this is another area, collaboration tools and how you build things that really work, that we could do for hours and hours. But it's not just that it's too much to do a protocol. It's actually just. But it changes the way you think about things because now you're trying to solve problems in generality instead of specific use cases. Back to your first point. So you would tend to try to describe narrower use cases for people to bring them. For users to bring them along better.

Exactly. We could have done the same functionality, but very specific about the user and I think that would have been very different. And then when you make big decisions on the product, like we at first could have email participants on Wave and then later we said, no, you can't do that. And then you're bringing all these developers along and they have certain expectations and then you take them away. So, yeah, a lot of things were hard about our approach.

Since you come to Greylock and started being executive in residence, you're thinking a lot about what to do next. What do you want to do next? You're thinking about civic technology, you're thinking about collaboration. How are you thinking about the next phase of your career and what's important to you?

Yeah, well, first I hope the audience of this podcast is going to call me and tell me, tell you what to do. Tell me what to. Let's see it in the comments. Yeah, that would be helpful. It's a magical and amazing thing when you work in mission driven organizations and mission can look like a lot of different things. I think bringing Google Maps to the world was mission driven. I think trying to do disaster response at Google was mission driven. Absolutely. The Clinton campaign was mission driven. So I'm trying to figure out what my next mission is and what would make me feel really proud and excited to go to work. You and I have talked about this a lot. There's been themes in my career about civic innovation. One of the early things I did when working on Google Maps in Europe was public transit. And I worked with an amazing engineer named Chris Harrelson because we wanted to bring all transit information into Google Maps. You might not remember, but back then it was just driving directions. And we created something called gtfs, which is a feed specification, which I think is one of the most successful open data standards for governments. And that was amazing. I left Google in 2011 to do a startup about residential energy and saving people energy in their homes. And it was really about building the first operating system for WI FI thermostats. And that was really mission driven. Later, at google.org, i got to work on a project we called Better Cities, which is, we have all this rich data about mobility from Google Maps in Waze. How can we give it back to cities in an anonymized way that can help them run their cities more efficiently and make decisions about how to use their resources? And maybe at a fraction of the cost of putting sensors in the roads or special cameras on the streets. That's been an awesome journey. I'm really proud to work with some people like Jen Palka and Code for America, which I think are pioneering a lot of really cool work in civic innovation and governments. And so I've been talking to a lot of companies since I came to Greylock that kind of operate in this space. And they, you know, they're all different themes of companies, from hardware companies and car companies to infrastructure companies. But there's amazing things happening in the Valley right now and all over the world, and I'm just really excited to kind of find a place where kind of a product person who's had my experience can have huge value.

Awesome. Well, thanks for spending time with us today, Steph.

Thanks, John. It was a huge honor.

That's it for Gray Matter podcast for this time. Thanks, everybody.