Bret Taylor of Sierra: How to sell to Enterprise Companies as an AI Startup

🔔 Subscribe for more startup strategy and founder stories: YouTube

👉 Follow us on all social media platforms: LinkedIn, X, Instagram, and TikTok

🎧 Follow Frameworks for Growth on Spotify and Apple Podcasts

‍

‍

In this episode

Vanta CEO Christina Cacioppo sits down with Bret Taylor, Co-founder of Sierra, to discuss the evolution of technology, from the early days of cloud at Salesforce, to enterprise-ready AI companies—plus cultural lessons from companies like Google and Facebook, and what it takes to launch successful AI startups today.

‍

Bret shares insights from decades at the forefront of tech innovation—including his time as co-CEO of Salesforce and CTO of Facebook—offering advice for founders navigating today’s market and tomorrow’s challenges.

‍

🔑 Topics covered (timestamps):‍

00:11 - Introduction

01:00 - Building websites and the birth of a career

02:28 - Yahoo Directory and the shift to AI-driven experiences

03:40 - What Sierra does and how it powers brands with AI agents

04:34 - Comparing AI evolution to past tech waves

05:45 - Frameworks still missing for building AI agents

06:30 - Democratizing access to AI tools for all companies

07:36 - Competitive moats in the AI market

08:54 - AI value chain: from models to airport-priced lattes

11:00 - Innovation pace vs. shipping outcomes

13:23 - Business models built around outcomes, not effort

15:53 - Customer alignment and predictable pricing at Sierra

17:40 - What people may get wrong about AI and productivity

19:00 - Economic realities: intelligence isn't the only bottleneck

20:25 - The early days of Google Maps and why it went viral

23:00 - Integrating content into Maps—and being name-dropped on SNL

24:19 - What makes a great board member (and a cohesive board)

26:49 - Founders and board dynamics: trust, motivations, and self-awareness

28:10 - Operator vs. investor board members

30:03 - Leadership styles from Benioff, Zuckerberg, Marissa Mayer, and more

32:22 - Customer obsession at Sierra and real accountability

34:59 - How Sierra plans for rapid shifts in AI tech

37:23 - Why model maintenance is a massive challenge

38:45 - Bret’s worst tech predictions (no self-driving, too much tablet optimism)

41:00 - There’s no one-size-fits-all leadership style

43:41 - The startup machine that produces happy customers

45:11 - Founder focus: how to know where to spend your time

47:12 - Founding Sierra during the ChatGPT boom

49:00 - Writing a choose-your-own-adventure app for his daughter

50:54 - Why Sierra started with enterprise customers

53:15 - How AI changes the economics of customer support

54:07 - Frameworks for go-to-market: PLG, dev-first, direct sales

58:30 - Selling software is a spreadsheet, not just a steak dinner

59:08 - Enterprise-first strategy and choosing design partners

01:02:18 - What makes a good design partner

01:02:20 - Lightning round: Ben Franklin, VR skepticism, and Clubhouse fizzle

01:04:45 - Wrap-up and closing thoughts

‍

Read the full transcript‍

Bret Taylor (00:00:00):

I like empowering every company in the world to take advantage of technology.

‍

Christina Cacioppo (00:00:11):

Welcome to Frameworks for Growth. I'm Christina Caspo from Vanta and here with Brett Taylor. Thanks

‍

Bret Taylor (00:00:16):

For having me. Thanks

‍

Christina Cacioppo (00:00:17):

For being here. You're the co-founder and CEO of Sierra, chairman of the board at OpenAI. Previously, you've been a co CEO of Salesforce, CTO of Facebook, worked on the like button and Google Maps. Thanks so much for being here.

‍

Bret Taylor (00:00:31):

It's my pleasure.

‍

Christina Cacioppo (00:00:32):

So if my background research is correct, you grew up in Oakland. Did growing up in the Bay Area have anything to do with you getting involved in technology, getting interested early on?

‍

Bret Taylor (00:00:42):

Not really. Actually, maybe it was just Oakland, but I think when I was growing up Silicon Valley to the degree it existed was very much down on the peninsula and I did not have a lot of exposure to it. But the reason I got into technology is because I rear-ended someone's car in a parking lot.

‍

Christina Cacioppo (00:00:59):

Oh yeah, definitely. Tell that story.

‍

Bret Taylor (00:01:00):

You want to hear the story. Okay. So I was 16 years old. I had saved up money to buy a 1971 Volkswagen bus, which I bought literally on Haight Ashbury to be even more stereotypical. And I drove it the day I got my license, drove to a movie theater in Walnut Creek, California. And because I was an incompetent new driver, just nicked the bumper of a brand new Cadillac, didn't even have plates. So I left a note on the windshields because I'm an honest young man.

‍

Christina Cacioppo (00:01:28):

An upstanding 16-year old.

‍

Bret Taylor (00:01:28):

Yeah, a upstanding, 16-year-old. And the long story short is the owner of that Cadillac owned a gas station, called me up and said, you seem like an honest young man. Do you want a summer job? And I would every single day in the back of my Volkswagen bus be getting on my Dickies and my shirt that said my name embroidered on it. And there was a mechanic next door who was griping about how expensive it was to make a website and get it on Yahoo directory, which was all the rage at the time. And I asked him, how much will you pay me if I make you a website? He said, $400.

‍

Bret Taylor (00:02:00):

I was earning $4 and 25 cents at the time. So let's just say I was very motivational. Went to a local community college where they had a computer lab, learned how to make website, got paid $400, quit my gas station job, and went around to every local business and I had a saddle maker, two mechanics, a couple lawyers, and that's how I did it. So the secret to any successful careers to first rear in someone's car.

‍

Christina Cacioppo (00:02:26):

There you go. Little known.

‍

Bret Taylor (00:02:28):

Yeah.

‍

Christina Cacioppo (00:02:29):

What was Yahoo directory like at the time?

‍

Bret Taylor (00:02:32):

Well, I mean, I think it's relevant in the age of ai. Back in the dark ages of the internet, there was just relatively few websites. So there's literally a directory. It was like the yellow page. People probably don't even know what the Yellow Pages is anymore, but it is literally, you could go through category by category and find it. And once Alta Vista and Google Search came out, I think the web quickly got too big for a directory to make sense. But the reason I think it's relevant in the age of ai, if you look at most websites today, they are essentially a directory. There's a menu. You could probably visit every page on the Vanta website if you tried just clicking around.

‍

Bret Taylor (00:03:08):

I think in the age of AI agents where I think a lot of these digital experiences are going to be conversational, all of a sudden it goes from very narrowly defined by the company to something that's a lot more expensive. There's a lot more agency given to consumers, and I think we're kind of in this interesting transition of consumer experiences where everything you want to do sort of enumerated in a menu to be in this freeform conversational experience. I think we're sort of going through this broad transition digitally that reminds me of 1996 or whenever that was.

‍

Christina Cacioppo (00:03:37):

So speaking of ai, could you tell us a bit about what Sierra does?

‍

Bret Taylor (00:03:40):

Yeah, at Sierra we help primarily consumer brands. There's some large scale B2B companies as well build customer facing AI agents. So if you have SiriusXM in your car, you'll chat with Harmony, who's their AI agent. If you're maybe buy a new car and you're on a trial or if you have technical support issues, if you want to upgrade or downgrade your plan, if you have Sonos speakers in your house, which is probably more likely for you is my guess. Anytime you get the dreaded flashing orange light, you'll now chat with the Sonos ai. So we do a mix of, I would say customer service, just enabling people, not have to wait on hold and get answers quickly, but have really started to evolve into other parts of the customer's experience as well. So onboarding, sales, things like that. So pretty exciting.

‍

Christina Cacioppo (00:04:24):

That's very neat. And you've been working on the web for a bit involved in a lot of major web frameworks. My first project I built on Google App Engine, so use your framework.

‍

Bret Taylor (00:04:34):

That's cool.

‍

Christina Cacioppo (00:04:34):

If you think about all you've seen develop, what kind of stands out now in the age of AI versus some of the things that were new 10, 15 years ago

‍

Bret Taylor (00:04:44):

From a technology standpoint?

‍

Christina Cacioppo (00:04:45):

Yeah.

Bret Taylor (00:04:45):

Yeah, actually, if you look at the history of technology developing of development of frameworks is really tightly coupled with the emergence of these new technologies. So if you look at Xerox Park, they invented the graphical user interface. They also invented small talk, which was a new programming language to go along with it. Unix gave us c, I assume somewhere between Xerox Park and the Macintosh. We got ides and things like that. If you look at agents, which is all the rage here in Silicon Valley, which is just software that could be autonomous and make decisions, I'm not sure we've really invented the right frameworks yet.

‍

Bret Taylor (00:05:26):

I think right now a lot of people are doing prompt engineering and experimenting with this technology, but expressing a agentic workflow as a set of goals and guardrails, I'm not sure we have the right abstractions yet, and it's an area that we're exploring at Sierra, but there's also a lot of open source software projects exploring it as well. And I'm kind of looking forward to five years from now when we look back on today and go, oh, wasn't that quaint how we were approaching it? In part because I do think as people, you really need to explore technology to understand what the right abstractions are, and then once those abstractions exist, react, it just opens the door to more people taking advantage of the technology. So it feels early to me, it feels like the Yahoo directory age, not the Google Age, if that makes any sense. And I love the exploration of it, but I also don't think we've gotten it yet in terms of what the right frameworks and abstractions are.

‍

Christina Cacioppo (00:06:21):

Yep. When you think about the things you've worked on, do you think of yourselves as a tool builder, like building tools, building frameworks, abstractions for other engineers or builders?

‍

Bret Taylor (00:06:30):

Not just engineers and builders? Actually, one of the things I've thought a lot about, I've worked in consumer technology, I've worked in enterprise technology and B2B companies as well. I like empowering every company in the world to take advantage of technology. I don't love the future where only a handful of technology companies have the privilege of using AI agents or taking advantage of the cloud or whatever it might be. I want your local flower shop all the way through to 150 year old home security company like ADT, who's one of our customers. They should be able to take advantage of the technology in the same way. And I think that is a form of making tools, but there's a question of how do you take something that's incredibly advanced and fast changing and make it accessible? What are the jobs to be done for someone who wants to say build an agent or build a website? In the case of App Engine, I think that's almost an imperative for the technology industry. Otherwise we'll end up with an economy of haves and haves, nots, and I think we really want to realize the productivity gains from ai. Everyone, individual, every company needs to be able to use it.

‍

Christina Cacioppo (00:07:36):

That makes a lot of sense. And on the early frontier of ai, there's a lot of talk about AI and competitive moats, and if it is just a lot of prompting, what are the moats where you're on the edge of this? How do you think about that question?

‍

Bret Taylor (00:07:48):

Well, I'll talk about the way I see the market just in case it's useful. I think the moats might be different for all of it. So at the foundation, I think you have frontier models or foundation models. I think this will play out a little bit like the infrastructure as a service market. So there'll be a small handful of companies that have the CapEx capacity to train these models, and it just requires a ton of investment to do so. And infrastructure as a service. There's not 20 companies leasing data centers. There's like three or five depending on how you count. And then on top of that, I think there's tool making companies in ai. It's things like scale AI and things like that that are basically selling pick xs in the gold rush. What do you need if you're developing an ai? And then I think there's applications and if you look at the cloud market where you had infrastructure as a service, you also had software as a service. And actually I haven't actually added it up, and I think it would be interesting if someone did what is the percentage of spend on software as a service versus infrastructure as a service? And I think a lot of the value is created in the SaaS layer of the stack.

‍

Bret Taylor (00:08:54):

And I just think of it like a supply chain, like the margins of say raw coffee beans are lower than the margins of roasted coffee beans, which are significantly lower than a latte, which is significantly lower than a latte in the airport. And all of that is just basically adding value in the supply chain. And I think the same is true of ai, and I actually think probably the most exciting and undeveloped part of the market is the applications market. I think most people would prefer to buy a solution to their problem rather than buy a bag of floating point numbers, which is essentially what a model is. So that's really exciting. What are the moats? To be honest, I'm not sure anyone really knows. I think people think about this too much in some ways. I think I talked to Toby at Shopify about this a couple of times where it's like people are like, isn't that just a database in the cloud? Or I could build this on a weekend. I've heard that about every SaaS product in history,

‍

Bret Taylor (00:09:49):

And it turns out that, yeah, you could make a latte at your house, but if you're in the airport and you want some caffeine, you're going to buy it there. There's a lot of nuance. I think that perhaps the moat for AI is the same as the moat for software as a service, which is essentially better product, better technology, better go-to-market execution and the scale associated with all of the above. And it doesn't need to be related to ai. Maybe it's just the moat that's existed and software as a service for a long time. I do think there's technological moats certainly, but there is such creative destruction right now in AI with new techniques, whether it's chain of thought reasoning or distillation of models. It's not clear to me that any one technical moat is durable, but I do think pace of innovation might be durable and pace of innovation over a sustained period of time can be durable as well. So my sense is people are being too intellectual about it, and I think there's a lot of companies just building and making customers successful, and I think you'll end up really understanding the contours of the competitive landscape over time.

‍

Christina Cacioppo (00:11:00):

The best strategy is the one that works on the field, not the one you came up ahead of time. And one thing we talk a lot about at Vana is tempo and pace of shipping things. When you say pace of innovation, is that pace of shipping to customers or when you say pace, what do you mean?

‍

Bret Taylor (00:11:14):

Yeah, I roughly mean that the only thing I'd say is motion is not the same thing as results. And so I think the stereotype of enterprise software is like this bag of features and it looks like a rough Russian submarine of dials and yeah, it slices, it dices, it's a hand cleaner and a dessert topping. It's everything that model can work, but it can sometimes deceive you as a leader because you can mistake the pace of shipping features with actually delivering value to customers. And so especially in the age of agents where I think I believe that software is transitioning from selling productivity gains to selling outcomes. Similarly, I think that people will actually start buying a job well done rather than buying a suite of tools to get the job done. I think that it's very important to know is the momentum you're feeling actually directed towards the right outcomes.

‍

Bret Taylor (00:12:17):

And I think that's particularly challenging right now, but I do mean that I do just mean pace. And if you look at, I'm probably going to get these numbers wrong and I'm saying on camera, great idea, but call it two years ago, GPT-4 was the highest quality frontier model, and I think it costs roughly $60 per million output tokens on the human eval benchmark. It scored somewhere I believe in the sixties. GPT-4 oh mini today is around 60 cents per million output token, so two orders of magnitude cheaper and scored, I believe in the high eighties on that same benchmark and people online will probably correct every statistic. I just gave two orders of magnitude cheaper and almost 50% higher quality in two years. You have to have pace because essentially the assumptions of your cost basis quality, they're just changing so rapidly. I've never quite seen a market like this right now. If you don't have pace, particularly as a technology company, you have almost no chance to two years from now be a market leader in anything.

‍

Christina Cacioppo (00:13:23):

I actually want to go back to something you said about charging for outcomes versus charging for the process. That sounds like a business model shift. Is that what you mean?

‍

Bret Taylor (00:13:31):

It is, yeah. Actually at Sierra, we do charge for outcomes. We essentially charge for the AI agent making a sale or the AI agent resolving a technical support case. It's based on the premise of saying what is the correct secular business model for agents? And we really love the idea of paying for a job well done. I actually think if you look at the history of software, there's been an industry in evolution of technology and business model shifts. I think most famously in the early two thousands, software as a service and subscription-based software came together. They were connected in the same way I would say outcomes-based pricing and agents are connected. They're connected because if you're hosting a web service, it's both good for the customer and convenient to have one version for everybody.

‍

Bret Taylor (00:14:17):

If you have one version for everybody, what's your business model? It used to be you'd buy Windows 2000, then Windows xp, the naming conventions of Windows. Maybe it was a bad example here, but you'd buy a new version. And now because there's one version of the software that's continuously updating, it almost demands a new business model to fund that innovation. And what's interesting is then they become sort of decoupled too. You'd end up with Adobe selling creative cloud subscription-based pricing, but still packaged software. But the two sort of came hand in hand. And I actually view in this era of agents, I think software will start completing tasks and what a pure business model to pay for the task well done. It'll be an evolution. I mean, if I assume all the engineers at Vanta use coding co-pilots and agents, and I don't think the quality is there where you pay for writing a module yet because candidly they're not really good at that yet.

‍

Bret Taylor (00:15:16):

But you could imagine a world where you do get to that point and paying per token doesn't make any sense. It's like paying an engineer for how many lines of code that they wrote. So there's something to this. I think it feels really pure. I think we're, as I said, really trying to explore the contours of this new business because if you look at what Salesforce did relative to Siebel, which was the on-prem competitor they had in the early days, or that you can go to any cloud versus on-prem in that era, there's just an opportunity to not only deliver software in a new way, but actually align yourself with your customers in a new way. And that can be really powerful.

‍

Christina Cacioppo (00:15:53):

That sounds very powerful. Do you have trouble getting customers on board with that just because it's something different? Not that it's not better, but they're used to paying a monthly subscription or an annual subscription of however much, and now you're saying, Hey, it's going to be variable priced based on a job well done, but I can't quite tell you how much your monthly bill is going to be

‍

Bret Taylor (00:16:11):

Broadly at first. Customers love it. I mean, I think that if you go to any head of procurement or any CIO and you ask them for their list of their vendors and you ask them, are you getting the value that you're paying for?

‍

Bret Taylor (00:16:26):

You'll start to see Blend leave their face and often some four letter words with a few of those vendors. The enterprise software has sort of been this history. I mean, this is a reductive stereotype of the steak dinner and then a presentation on ROI, and It's like if you make everyone X percent more productive, blah, blah, blah, 

‍

Christina Cacioppo (16:40):

Our software is so cheap. 

‍

Bret Taylor (00:16:48):

Yeah, exactly. You should be paying. It's just hand wavy, let's put it that way. And I think, well, people I think broadly have loved this transition to software as a service. There's just not a lot of accountability. It's really usually up to a company to make themselves successful with the software that they've already purchased. I really believe that this changes the equation where it's up to Sierra as a partner to actually make you successful before we get any economic value from the relationship. So I think it's really positive. And the other thing is it is more predictable. I think that most of our customers have very large scale and it's fundamentally a spreadsheet at the end of the day, not hopes and dreams. And so I do think it's really predictable and I think customers love the accountability

‍

Christina Cacioppo (00:17:40):

Or they roughly know how many support inquiries or implementation projects or it is they have.

‍

Bret Taylor (00:17:45):

Exactly,

‍

Christina Cacioppo (00:17:45):

Well taking on this AI as helping jobs be well done, when we think about AI and the workforce, and let's go forward 10 years, so 2035, what do you think we'll have gotten wrong about how we think about all that now

‍

Bret Taylor (00:18:00):

About ai, broadly

‍

Christina Cacioppo (00:18:02):

AI and its impact on productivity jobs, the workforce? And again, maybe something you're getting right, but the conventional wisdom gets wrong.

‍

Bret Taylor (00:18:10):

Yeah, it's very hard to see more than a couple years into the future. 2035 sort of feels like science fiction novel territory to me. I'd say the thing I read or I guess listened to in this case most recently that changed my perspective was an interview with an economist named Tyler, I think Tyler Cohen Cowan, I'm not sure to say his last name. And he was just talking about different parts of the economy being impacted by AI differently and essentially making the argument that some parts of the economy, like our industry software can absorb intelligence with almost no limits because the main limiter to producing software is intelligence, but there are other parts of the economy where that is not true. So if you invented a new therapy, you'd still have to do a clinical trial, and that clinical trial would be the limiting factor to deploying these new therapies.

‍

Bret Taylor (00:19:00):

And his point was not pessimism around ai. My understanding is here is actually incredibly optimistic about these models, but more that are unbridled optimism about how the economy is going to transform is probably unrealistic, that there's actually many limiters to growth beyond intelligence. And that actually really, I've recalibrated my perspective based on reading more about it and in particular thinking about the next, call it five to 10 years where different parts of the economy will be impacted much more unequally, which will lead to a lot of good and bad things. You end up with a lot of just when productivity changes dramatically differently in different parts of the economy. So I've been thinking a lot about when you have something like general intelligence, how do you actually generalize it to other parts of the economy? That term is typically related to the intelligence, not actually the application of it. And I do think when I think about, say OpenAI, where our mission is to ensure that artificial general intelligence benefits all of humanity, that last part, all of humanity, I'm like, gosh, that could be the really hard part here. And so I've been thinking a lot about it just in the context of that mission.

‍

Christina Cacioppo (00:20:12):

Interesting, and I hear you on 2035 being kind of science fiction land maybe take you go the opposite direction. So going back 20 years, so again, if I did my research correctly, you worked on Google Maps in circa 2005.

‍

Bret Taylor (00:20:24):

Yeah.

‍

Christina Cacioppo (00:20:25):

Did you think Google Maps then would be what it is today?

‍

Bret Taylor (00:20:29):

Absolutely not.

‍

Bret Taylor (00:20:30):

So yeah, the story there, we all sort of compress it, but there was a group of, I always say five of us, but Lars and Yen Rassmuson, two Danish brothers who I'm still close friends with, me and a few other folks on that team were just still some of my closest friends. And it was at the end of 2004 and we made sort of the first version of Google Maps, and I remember sending it out to the engineering team right before we left for the holidays. So it was probably December 23rd just because we were trying to get up for the holidays and coming in hot and it got a lot of buzz internally. I think people at the time, probably most people listening don't even remember this, but webpages at the time were quite static, and this was, I think really the first web experience that was truly graphical and dynamic and in particular, you could drag the map and it would move, which sounds so trivial today, but it was mind blowing at the time.

‍

Bret Taylor (00:21:31):

We ended up launching on February 5th, I want to say it might've been the seventh, I should know this. It just passed the birthday and it really did go semi viral, which wasn't really a thing at the time, or at least that word wasn't, but it was like the top of dig or something. It was really popular in some of those corners. But the thing that really happened was when later that summer we integrated the satellite imagery from an acquisition we had done from a company called Keyhole, and we put it into Google Maps. You could switch between map and satellite, and literally the whole world was like, I want to look at the top of my house right now,

‍

Bret Taylor (00:22:04):

And I think it was like 90 million people showed up right? Then we ended up on SNL where Andy Sandberg did a wrap mentioning Google Maps, and I remember Larson, I'd be like, we made it, Andy Sandberg said Google Maps were good. So it was amazing. It was really fun. And in particular it was really fun. It reminds me a little bit of agents insofar as we were exploring the contours of a new technology, trying to always almost have a conversation with the technology about what's possible. Sometimes our ambitions would be here and the technology would be here. We'd have to redesign a product feature to facilitate what was possible at the time. I feel like that's a lot of what's working with AI is today your product design needs to take into account the limitations of the technology, which is evolving rapidly, but in particular, it just became a cultural moment. I joke and it's true, every time my father gets lost, he still calls me and yells at me. I'm like, dad, I haven't worked at Google for more than a decade, maybe two at this point. He's like, well, you must know somebody there. I just wanted to let you know I got lost. I'm like, thanks dad. We'll send it to the official feedback

‍

Christina Cacioppo (00:23:11):

Support@google.com.

‍

Bret Taylor (00:23:13):

Yeah, exactly. So it was fun and just mainly as a early young product developer, I think it just gave me a lot of confidence and to do all the things that I did after just because it's sort of like the feather in your cap kind of thing, just because it was so well-regarded.

‍

Christina Cacioppo (00:23:27):

Yep, so true. It's funny you talk about the panning and now that you say that the little MapQuest arrows.

‍

Bret Taylor (00:23:32):

Yeah, I mean it used to be you look at a map that was maybe 200 pixels and you click right and just like sit there and wait for it to render. It was truly, and the other thing by the way is it integrated a lot of product categories. So previously you'd have a Yellow Pages website and you'd have a mapping site, and the Yellow Pages website also didn't have any content from the web. It was just name, phone number, address.

‍

Bret Taylor (00:24:00):

We integrated web, so we had things like the reviews of a restaurant. We integrate the yellow peaches listings and we integrated a map all in one product. So it was really quite disruptive at the time.

‍

Christina Cacioppo (00:24:13):

I remember it being such a big deal, and it's funny because it's still one of my daily use apps 20 Years on

‍

Bret Taylor (00:24:17):

That makes me happy.

‍

Christina Cacioppo (00:24:19):

You've been on boards and been attached to company, I mean social media company, so you're clearly on the board of Sierra, the company you started, you're on the Open AI board, you're on the Twitter board across all of the, I guess, quip before Quip was acquired. When you think about really talented, helpful board members, you think about the canonical. If you could write the job spec of a board member, what's on that for you? You've seen so much.

‍

Bret Taylor (00:24:44):

Well, first I don't think all board members are the same. I do think a board is a collection of expertise, and I do think, well, a public board, it means a lot of specific things. You need a financial expert for the audit committee. Your compensation committee has a very specific purpose, and you probably want people who are familiar with compensation in your industry. As you and I know compensation in here in Silicon Valley is a little bit different than it is in other places. And so I do think thinking about a board, I like to think about it as the whole rather than the parts because you really do want these different areas of expertise represented. Maybe it's regulatory expertise, maybe you want some people who are operators recently. So as you're scaling a business, you can have a trusted advisor When you're thinking about growth or international expansion or a new executive on your team, you want financial experts so that as you're transitioning from a private to a public company or you are a public company, you have experts to partner with the CFO.

‍

Bret Taylor (00:25:44):

And I do think that finding eight to 11 unicorns is not really the way to construct a board. And you really want a group of people who are exceptional in their own way and you need to make sure they're cohesive. I'd say for, in my experience, there's a lot of nuance, but I think most CEOs and founders that I know, operators on their board because they are the most valuable in terms of going through getting practical advice on a day-to-day basis. Finding operators who know how to be an advisor rather than operator is probably the most nuanced thing, and that is a bit of an art form, but I would say it's something that's part of the reason I've continued to serve on boards is I do think it's a different muscle than the one I exercise every day at Sierra, and I think it's valuable. I've appreciated the great board members that I've had and I've tried to sort of model it and it's hard. You're used to running things day to day and you have that layer of indirection and adding values an advisor is really different. But I would say certainly for the founders, I know that's the board members that are probably the most valuable for them personally as executives.

‍

Christina Cacioppo (00:26:49):

I definitely agree with that. Maybe for a founder who's watching this, how do they try to tease that out? How do they find, or maybe are there questions, but just try to get a sense of that you understand operating, but you're not going to try to operate from the board at this company?

‍

Bret Taylor (00:27:04):

The best way definitely is references just because I think if someone has done it before, getting those practical experiences for someone who might be a first time board member, I think having that personal connection is probably the most important thing. I think you just don't want people living vicariously through your management team or a company. I think understanding the person's motivation, is it because they want to be a trusted advisor for you and the management team or is it something else? I think if you understand someone's motivation and you have that personal rapport, you can course correct on it. And I do think it's really important to understand people's motivations. Are they in it for the right reasons? And usually if they're not, that's where you can end up with some bad behaviors,

‍

Christina Cacioppo (00:27:53):

Different dynamics. Yeah. And point will taken on you and shouldn't try to find eight unicorns, but maybe for a founder who's just getting started and they're starting to think about a first independent board member, would you recommend an operator or is there any framework you'd use there?

‍

Bret Taylor (00:28:10):

Yeah, I do, and I'll caveat all with our experience is fairly limited. As business executives, you tend to extrapolate from a relatively modest number of data points. So the advantages that I think most investors bring is they have a lot more data points but have been depending on how long they've been an investor, a layer of indirection from the details. The advantage of an operators, they've actually probably encountered problems exactly like the ones that you're encountering, but they've only seen a sample size of one or two. And so one thing you as you're getting advice from other operators, depending on their level of self-awareness and humility in particular is sometimes you can end up copying the statistically insignificant details of certain decisions that were made. And people tend to confuse correlation and causation quite a bit in their life experience. This business succeeded, so therefore every decision I made was correct, is never true.

‍

Bret Taylor (00:29:11):

You can't itemize that receipt though. There's no way to know which decisions were the most significant. You have an intuition about it, and so sometimes you'll end up in a hard situation, you'll get some very strongly worded advice from an operator who's seen it once and really believes authentically and it could be the wrong advice for you. And so that's the hardest thing is finding people. So in general, I like to find folks who have, one of my main things I interview for both executives and board members is self-awareness. Do they have a self-awareness to understand where they might be extrapolating from just a single data point incorrectly? Can they contextualize their decision so that you can synthesize not just the decision, but all the context they had and try to figure out what might be different about your situation? And I find some folks who are very black and white about everything are rarely right in my experience.

‍

Christina Cacioppo (00:30:03):

Yep. That leadership, I dunno lesson, but the self-awareness being important to you, something you interview executive and board members makes a lot of sense. You've been at lots of the valley's biggest companies and peak moments and been around those folks that you've started your companies. What other sort of leadership lessons have you taken from those experiences and applying now at Sierra?

‍

Bret Taylor (00:30:27):

Yeah, it's a wonderful question. I've definitely learned different things from different folks I've worked with. One of the things that we try to do at Sierra's first principles thinking, I think it's particularly important right now with technology changing so rapidly. And I probably saw that most in Mark Zuckerberg though I've seen it from when I look at some of the larger firms I've worked with Google, Facebook, and Salesforce definitely saw it, but Mark was by far the most methodical leader I've worked with who started from first principles and then talked it over a very long timeframe, longer than I typically would to the point. It really changed the way I think. I try to think longer term, and I largely credit Mark's influence with that. And I always joke every time I was thinking five years out, he'd be thinking 10, and if I happened to be thinking 10, he'd be thinking 20 of like, oh gosh, I can't win with this guy.

‍

Bret Taylor (00:31:20):

But I think it is really important when you're in a period of rapid market shifting dynamics rapid if you're in a very competitive market as well, winning this year might not mean winning three years from now and things like that. So first principles thinking is definitely one that I care a lot about. And then I would say one of the things that I really felt I learned at Salesforce was just being obsessed with your customers. I think a lot of companies say that, but I really felt like I experienced that there and one of our values at Sierra is customer obsession, but a lot of people have that value, but I'm just talking about you actually mean most of our customers have my mobile number, and when something goes wrong, we drop everything, we fix it. And that's something I really thought I knew before then, but when you actually feel it, you're like, oh gosh, this is like, I was probably at a five when I should have been at a 10 here, and I've really tried to adopt that as well.

‍

Christina Cacioppo (00:32:22):

Okay. I was going to ask customer obsession and at Sierra, how does that manifest? So I hear you on if something comes in, drop it and fix the flip side of all these values of the flip side. The flip side there is someone being like, Hey, I was working on whatever, now I have to go triage this incident. But how does that actually work at Sierra? What I described is not at all what you described.

‍

Bret Taylor (00:32:43):

Yeah. Well first I'll start with I think it's really important to deeply understand what your customer is trying to achieve. One of the things that we talk about is we drive towards business outcomes, not technical milestones. Our job is not to launch an AI agent. Our job is to save you hundreds of millions of dollars in operating expenses for your contact center expenses. Our job is to drive top line sales growth. Our job is to improve revenue retention for a subscription business, whatever it is. And I think it's really important we understand that because I think often when you end up feeling reactive with customers, it's because your vantage point of the project is myopic and it's associated with

‍

Christina Cacioppo (00:33:25):

You're like, but I shipped all the things.

‍

Bret Taylor (00:33:27):

Exactly, exactly. So we try to hold ourselves accountable to truly understanding what our customers are trying to achieve, and we try to align the whole company towards that. And then in our experience that I'd say the negative case you imagine there, which is drop everything and it's not worth it, it rarely happens because typically when our customers see how much we are aligned with their business outcomes and we're accountable to delivering a business model all the way through to how we deliver it, they're respectful of that relationship too. We're typically in these together. And so if the proverbial bat phone rings, it rang for a reason. And so it doesn't feel random. There's not resentment. It's a true empathy, I think is the way I'd put it. And then to the point on interrupting people, and this is what good product and technology does. If you we're not perfect by any means, but you just want to make your platform as durable and reliable as possible. So if your PagerDuty or bat phone is going off too often it's a reflection of product issues more than anything else. And I think if you focus on doing good craftsmanship and giving the product, right, and then you focus on aligning yourself with business outcomes, I think you can really have the level of empathy that you predict the problem before it happens. And if the bat phone does ring, you're like, I want drop everything. This pains me, I want to fix it for you.

‍

Christina Cacioppo (00:34:50):

That makes a lot of sense. And then on time horizons, when y'all are at your best at Sierra, how long out are you thinking, what is your time horizon there?

‍

Bret Taylor (00:34:59):

It's the fog of wars thing. Exactly. Totally, totally. We technologically, we try to think pretty far ahead just because when you're building a platform like ours, you really want to predict where the technology is going so you can make good decisions. And we're always recalibrating actually as a good example of that, once very high quality distilled models. And for those people who aren't familiar with the term, it's sort of taking a very large high parameter count model, making a faster, cheaper, lower parameter count model that's within a reasonable margin almost as high quality. When those came out, it really changed our strategy towards fine tuning just because we had a really compelling set of high quality models on which we should fine tune. And it changed some of our assumptions around cost for some operations as well when chain of thought reasoning models became common similarly, it just opened new doors.

‍

Bret Taylor (00:35:53):

So you always have a view of the future and you're always recalibrating. And then with customers, we try not to be too exotic. We want to save, solve their problem today. And so it's kind of a shorter times horizon with the customer, which is like, let's just get some wins, let's go live. I'm really proud that the number of just live agents in the wild having millions and millions of conversations, and then you have a longer time horizon on the platform itself, and then you're always triangulating those two things. Maybe this customer problem could be solved very differently three months from now. How do we facilitate that? But I think it's really important to not be ivory tower when a customer has a real issue, you want to solve it with a technology that exists today, the last thing you want is to paint yourself into a corner technologically. I think that's actually one of the risks I see in the AI applications market is if a applied AI company is not agile enough, they could end up going from the best technology to the worst in 18 months if they're not careful. And so that's just a risk because the technology's changing so quickly. And so we really think a lot about agility and as long of a time horizon is one could have an AI right now, which is two years honestly.

‍

Christina Cacioppo (00:37:10):

One thing we've actually talked about with Vanta is we build with AI is the maintenance cost of some of these models in the evals and it just being very different than the workflow software. Is that something you've all seen or think through?

‍

Bret Taylor (00:37:23):

We have, and actually I would say for a very high percentage of our customers, one of the values they perceive in us is not simply building AI agents for customer experience, but abstracting their customer experience from the churn and chaos of the models underneath. Put simply, I think we want to enable our customers customer experience to be durable in the face of all the changes to the models. And it is a really rapidly changing new world. Even models released by the same research lab from version to version will rapidly change. And I think over time, the reason why I'm so bullish on applied AI companies and companies building agents is I'm not sure it's feasible for most companies to deal with the complexity of these models and evals. And even if you do it once, it's

‍

Christina Cacioppo (00:38:15):

Out of date,

‍

Bret Taylor (00:38:16):

Months later, out of date, six months later, it's like a lot it needs to be tended to.

‍

Bret Taylor (00:38:20):

And so I think the value proposition is even more acute and AI for solutions than it was in software as a service. And it was already, I think most companies should not be building their software from scratch. And I believe that a decade ago. I think it's even more acute in AI today.

‍

Christina Cacioppo (00:38:35):

So you clearly spend a lot of time and try to be on the cutting edge of these things, but on the cutting edge of technology for the last two decades. Have you missed anything?

‍

Bret Taylor (00:38:45):

I have, yeah. Actually it was funny, somewhere in the pandemic I tweeted something about where I was wrong and it was the best, my favorite response thread I've ever seen because people responding with every bad technology prediction they've ever done. But I've had two prominent ones. I've been wrong in a lot of little things, but when my daughter was born, one of my daughters was born 15 years ago, I bet my wife that she wouldn't get her license because we'd have self-driving on demand cars by now has not happened. And I lost that bet.

‍

Christina Cacioppo (00:39:19):

Did you have to buy her a Volkswagen?

‍

Bret Taylor (00:39:22):

$700 car of your choice! And it was interesting because I do think it reflects a common way folks like me will get predictions wrong because I'm a technologist is maybe be a bit too optimistic about the pace of technology change. And I think there's been a lot written about this, just about the sort of how quickly we got to very good autopilot but not how the gap between that and truly autonomous driving is just much longer than we thought. There probably would be some lessons in that for other forms of AI as well. And the other one, I thought the iPad would be more of a replacement for laptops than it turned out to be. I have an iPad, I love it, but it's not my laptop. It's just a different form factor device. But I don't know how many iPads or tablets there are in the world, but it's not as many as I would've expected by now.

‍

Bret Taylor (00:40:15):

And I remember when I was starting Quip, we were really thinking of the future as a phone and a tablet and really saying, what does the world look like? And certainly mobile phones have exceeded all expectations. I think it truly connected the world. These numbers are probably off, but I think there's a couple billion PCs in the world and there's as many smartphones as there are people. And so it fulfilled Microsoft's original vision of a PC on every desktop. Well, we don't need a piece now you have a supercomputer everyone's pocket. But tablets didn't really meaningfully disrupt the laptop industry as far as I know. And that was really surprising to me. I thought that form factor would take off in different ways. And I have my own opinions, but I'm not deep enough to know why.

‍

Christina Cacioppo (00:41:00):

Similar question, but more about people or leadership styles, someone you came across or a leadership style you came across. And at first were like, that's never going to work. And then turned out, actually.

‍

Bret Taylor (00:41:11):

Actually I'll broaden it a little bit related to that, which is I feel very grateful to have been able to directly learn from Marissa Meyer, who's my first boss at Google, Larry and Sergey, mark Zuckerberg. Mark Benioff. They could not be more different.

‍

Bret Taylor (00:41:27):

They're very, very different people. And so I've really come to appreciate that there is not a archetype for a good leader. You can have someone who's incredibly introverted and relatively soft-spoken and they can be an incredibly effective leader for a company or a team. You can have someone who's extremely gregarious, like Mark Benioff is an incredible executive and leader in the software industry. And so I would say my opinion of leadership is much more nuanced now, having not only seen them in action, but just seeing that just great companies have come from a lot of different people. It's changed the way I think about leadership because I think whatever the opposite of a formula is, I actually have really come to appreciate what those styles of leadership produce. And if you just look at say, a consumer technology company, they tend to be very product driven, rightfully.

‍

Bret Taylor (00:42:22):

And you can see a lot of product-led growth companies who have struggled going into direct sales. And there's a lot of nuanced reasons. But one of the things you need to have a great direct sales team is great salespeople need to want to work there. And if they're not respected by the executive team, that's going to be pretty hard no matter how much places that well, yeah, no matter how much they get paid on commission. And then similarly, if you have a great direct sales team, but your product is clunky, it's going to be hard pressed to get a great engineer there. It's not a comp issue. People want to be proud of what they're making. And so if you look at company cultures and leadership, the good and the bad strands of DNA are often intertwined. And I just don't believe in the caricature of a great leader or a bad leader.

‍

Bret Taylor (00:43:07):

I actually think you end up, you're sort of the product of your leadership and you need to accept the good with the bad. And talk about self-awareness again, you need to be self-aware that you are not infallible as a leader. And in fact, probably what makes you great has some side effects that aren't so great. And you can either just accept it and that's fine. If it's something that might be catastrophically bad for your company, you need to address it. So anyway, I have a very nuanced view of leadership and I would say more self-aware about my own leadership having seen such a wide variety of leaders.

‍

Christina Cacioppo (00:43:41):

That makes a lot of sense. What about for a founder starting out who wants to build a great company, but maybe the word leadership style or management style is a little like, but what does that mean? How could you distill some of that for them?

‍

Bret Taylor (00:43:53):

I like to keep it simple, which is your job as a founder is to make a machine that produces happy customers. And at any given point, you should decide what is the most important thing to focus on to make that machine work and produce the greatest output of happy customers. I think one of the mistakes that founders make is they're single issue voters. So if you're a great engineer, the answer to every problem is engineering. If you're a great partnerships person, the solution to every problem is some partnership.

‍

Bret Taylor (00:44:27):

And it turns out that companies are multifaceted and the great founders will not necessarily become someone they're not, but they'll have the awareness to know, this is the part of the machine that I need to focus on today. It might be product, it might be technology, it might be partnerships, it might be sales, it might be leadership. It might need to find a new person. And I think I really believe people can learn how to do almost anything. But as a founder, you need to know what to spend your time on. And I think that's the strength that I think people should focus on. And the other thing I'd say partly just as when there's these hype cycles in Silicon Valley, we're certainly in one with ai. We went through mobile a while back, we'd go through these ways

‍

Christina Cacioppo (00:45:11):

crypto every once in a while.

‍

Bret Taylor (00:45:12):

Yeah crypto, yeah, that'll come back every couple of years. Founding a company is not a popularity contest. No one cares. What really matters is are you building a sustainable business? And I think that it's very easy to confuse perception and reality when the level of excitement and hype is so great. And I think that's a common pitfall, but particularly now, it's a common pitfall. And I think that I really like to just think of this as, can you produce happy customers? What is the barrier to that? And just being self-aware about where the gaps are.

‍

Christina Cacioppo (00:45:47):

Okay, so maybe this is the answer, but I was wondering the skill of honing what you should focus on, how do you do that as a founder?

‍

Bret Taylor (00:45:54):

The technique I've found most effective is to imagine I'm giving myself advice, which is just a trick to, I think it's sort of seeing the forest for the trees. Great board members can often help here. And certainly I always like to end a board meeting, which is like, what did they say that surprised me? And I always, it's good to ask, what do you think we're missing inside of a boardroom? But on a continuous basis. I always think if I were giving myself advice right now, what would I say? And it's a nice technique that makes me lift my head up from the trees to look at the forest a little bit. And then similarly, I think it's really important to create a culture of trust in a company where people can say what they're really seeing and feeling because if there is an issue with a product that most companies, the product's great, We're better than everybody.

‍

Christina Cacioppo (00:46:46):

We're just not selling it.

‍

Bret Taylor (00:46:47):

Our competitors think whatcha talking about. And you don't want a negative culture where everyone's trash talking your product all the time. You need a culture of positivity. But having that culture where people can be like, gosh, I think this feature is worse than this competitor, let's go fix it. If you don't have that data or that feedback as a founder, you're not going to even know where the problems are. But I would say act like you're giving yourself advice and make sure you know what's going on.

‍

Christina Cacioppo (00:47:12):

Yep. Makes a ton of sense. I've heard the more extreme version of that, which is like assume you were fired and then you as someone else was brought in, what would you think?

‍

Bret Taylor (00:47:20):

That's hardcore, I might use that. It's a little depressing

‍

Christina Cacioppo (00:47:24):

Clearly. Then you're like, we'll do that

‍

Christina Cacioppo (00:47:27):

Sort of stuff. Well founding for you, so you've now founded three companies. Usually the second time folks will be like, oh, I just want to do X, Y, z different or do a set of things different. You're on go round three. Why'd you choose to do that?

‍

Bret Taylor (00:47:43):

I chose to start this company just by coincidence. When I left Salesforce chatGPT launched that month and kind of nerd sniped me a little bit. I just couldn't stop thinking about it. I had previously been really impressed with Dolly, but I think like everyone else, when GPT sort of three, five, and three had come out, you could really start to see the contours of something that felt like science fiction. And there's this Allen K quote that I really love, which is the best way to predict the future is to invent it. And I can't sit on the sidelines when there's something this exciting being impacted. So I'd either be doing open source software or starting a company. I'm a huge believer in capitalism and to rally people around a cause. And so that's why I did it. It was really passion for the space and just really wanted to explore where this technology is going to take the economy and society. And in particular with our business, trying to empower every company in the world to build their AI agent, the one that has their brand at the top feels really exciting to me, just feels really empowering. So that's why I did it. I'm really just passionate about ai.

‍

Christina Cacioppo (00:48:52):

What was it about GTP3 that you kind of kept you up at night?

‍

Bret Taylor (00:48:55):

It was sort of emotional for me where if you look at any science fiction movie, all the way back to 2001 is Space Odyssey or before, the way you engage with computers is by talking. There's no really great sci-fi movie where someone's got a mouse and a keyboard and it's clicking around. It really felt like computers could understand what I was saying with nuance. And even in earlier models like that that were not incredibly intelligent, you could see glimpses of true intelligence. There was a lot of very subtle points in there too. Using computers as a creative foil felt really different to me. I'd always thought of put computers in the box of really fast rules engines kind of thing. And then the idea that one of the things that I did when I had that period of, I was done working, but my co-founder was still winding down at Google.

‍

Bret Taylor (00:49:52):

I wrote an app for my daughter for you remember, choose your own adventure books, of course, where I turn to page 42.

‍

Christina Cacioppo (00:49:58):

Yeah,Totally.

‍

Bret Taylor (00:49:59):

So I wrote an app where she could choose at the end of every chapter, choose go through this door or walk around and it would write the next chapter based on what she chose. And it was not the highest quality novels in the world. It was really neat because what I liked about it is it was literally writing the book as she was making decisions, which I thought was really neat. Just the idea that I was writing technology that was writing a Choose your own adventure book, it violated the conventions of where I thought technology would be. So that was just really interesting to me. And so for me, it felt like a very material change in how we as people would engage with technology as significant as the network and the internet. I wouldn't have been so articulate then, but when the marginal cost of intelligence is going to zero, how much of society and software change is like everything. And that's just incredibly exciting. So I had that intuition and I just needed to explore it.

‍

Christina Cacioppo (00:50:54):

That's very neat. And then Sierra, if I'm correct, it's been launched publicly for about a year, but in that year you've got customers like ADT, weight Watchers, Sonos, these really big brands. That is not the path of most Silicon Valley startups of starting with the biggest brands. Why'd you go that way?

‍

Bret Taylor (00:51:13):

Yeah, we started really feeling that some of the larger consumer brands in the world could benefit most from this technology. Now, just to put some numbers on it, if you call your phone company on the phone, my guess is the price of that phone call would be $10, maybe 20, depending on how long the conversation is. It could be a little lower if it's really transactional. That's a lot of money. And so you joked about support@google.com or whatever, it's pretty hard to get Google on the phone. I'm not sure you could,

‍

Christina Cacioppo (00:51:43):

yeah, I don't that if that email exists at all

‍

Bret Taylor (00:51:44):

But if you just think about it economically, the average revenue per user just doesn't warrant any interaction costing that much. It just can't. And so you end up in this world where I think it's entire websites devoted to finding the phone numbers of companies because it's so hard to reach them. But it's not because these companies don't care about their customers. They all do. They just can't afford those interactions. They just can't. And so you've ended up in this world to date where most people are trying to build digital self-service capabilities so that you can, as a customer, get what you need without having to talk to somebody. And then when you do, it's very expensive and it's cost optimized to the point where many of those interactions, even if you do get on the phone, are

‍

Christina Cacioppo (00:52:29):

Tough to resolve.

‍

Bret Taylor (00:52:30):

Yeah, exactly. And so it's one of those areas that with ai, you can bring down the cost of those interactions by almost two orders of magnitude, and that completely changes the game. I think a lot of people think of it as a way to drive cost savings, and it absolutely does. We're saving some customers in the sort of nine figures of operating expense savings, which is amazing, but also just changes your options for customer experience. I always like to say if something goes down by tours of magnitude, a great founder won't say, I'll just recoup all those cost savings. They'll say, Hey, maybe I can have more conversations with my customer. I'm looking forward to a Super Bowl ad, which is like, call us. Here's the number. I dunno when they'll happen, but I think it will because it's actually sort of an economically viable thing to do.

‍

Bret Taylor (00:53:15):

So we started just because if you have millions of customers, this technology can really materially change your business now. And so we really said let's start really focusing on some of the world's largest brands to do so. Though we have small ones too, a lot of small retailers, we have a number of B2B customers, but I'm really proud that ADT home security, it's literally been around for 150 years. The T stands for Telegraph and now we're helping them make their first AI agent and doing millions and millions of interactions on this platform. That's just exciting. It's really fun to, it's also fun for me just because it's fun that so many brands that I'm a customer of, I can chat with this AI that we worked on together. It's just really satisfying.

‍

Christina Cacioppo (00:53:57):

At Sierra, you're working with enterprise customers at previous startups you've built for founders. You've done SMB mid-market. Any frameworks for choosing which customer to go after?

‍

Bret Taylor (00:54:07):

Yeah, if I have one piece of advice for founders, it's to be intentional. And I think I wasn't as intentional in the past and I've learned the hard way, the downsides of that. When I look at the different enterprise sales motions that have been proven to work, the first would be product-led growth. This is probably epitomized by people like Slack and Atlassian where the user of the product and the purchaser of the product are the same person and you the type of, you are empowered as that person to go sign up and use something and then typically the first purchase could even be on a credit card and then you can expand from there. And most product-led growth companies go through the evolution of first focus on the user, then hire sales team to sort of sell into enterprises that adopted your product. That only works if the people purchasing the product are the same as the people using it.

‍

Bret Taylor (00:55:00):

So it doesn't work, for example, with expense reporting software because it turns out that I file expenses, but my CFO is the one who chooses the expense reporting software. So first you really need to probably a growth is not a strategy, you can just apply to any category of software. It literally only works is by definition the user and the purchaser of the software, the same person. And I've seen a lot of entrepreneurs miss that. Key defining characteristic number two is developer. If you have a shell script or curl command on your homepage, you're that. I would say for me, Stripe and Twilio probably epitomize that business model. I always like to think of this as you're targeting a developer within the office of the CTO or the IT department. You're basically finding a developer who's empowered to find these new tools and then you end up going into a platform from an individual developer level.

‍

Bret Taylor (00:56:02):

It's a very successful business model when it works. And I'd say most of the developer tools fall in this category. You tend to be really oriented towards developer experience in that world because your go-to-market motion is effectively someone choosing to bring it in and it's sort of like a Trojan horse into a larger platform deal. And then there's direct sales and I think direct sales, it really matters what segment you're going after SMB versus enterprise. The reason I think it's really simple is how much quota you assign your salespeople and the average revenue per contract. And I think a lot of companies just don't do math very well and you can't afford to have the number of salespeople you want if your average contract size is small, which either means you need to sell something more valuable or raise your prices or something. I think it's just useful to pull out a spreadsheet and just do some math when you're trying to figure it out.

‍

Bret Taylor (00:56:55):

I think to some degree how valuable the thing you're selling determines your go-to-market model there. And I think people need to choose their segment appropriately. I think product-led growth tends to work really well in SMB because you can't really afford the right sales capacity for most SMB depending on, there's a handful of very high value products where maybe that would change. And then in the enterprise there's just limits to product-led growth. I can't think of one that really, really, really grew in the enterprise. There's definitely some exceptions and it's in part because you tend to just break that rule, which is the purchaser and the user are just not the same person in most large companies either because they're not empowered to or just candidly tool choice is much more nuanced at a larger firm. And so I just see a lot of entrepreneurs thinking not critically enough about their go-to-market motion and how that impacts what product they're building, how they price and package it, and all of those are very connected and I think is another way of putting it. And so I just think entrepreneurs should be very intentional about their go-to-market motion and very realistic about their average revenue per contract and things like that in sort of forming how they build their business.

‍

Christina Cacioppo (00:58:13):

I think one of the things building off of that, that was really surprising to me as I learned sales at Vanta, I thought sales was golf and steak dinners and chisel jawlines and it's actually just spreadsheet math. A lot of it you can do so much of, especially I think like down a spreadsheet

‍

Bret Taylor (00:58:30):

And a steak dinner

‍

Christina Cacioppo (00:58:31):

And a steak dinner yeah, but that works quite well

‍

Bret Taylor (00:58:36):

If you're doing your job as a software company, you're providing real value to your customers and you're just trying to figure out how they should consume it. And you can have a really productive conversation with procurement because you're providing something really valuable. And I really agree, and I think if you end up in that golf steak dinner, it's certainly, relationship building is a big part of sales, but you should really be talking about the value you're providing and it should be much more of a spreadsheet.

‍

Christina Cacioppo (00:59:04):

Any advice for founders that want to go enterprise first?

‍

Bret Taylor (00:59:08):

If you don't have a large network, it can be hard just because it's very noisy, particularly in AI right now to get people to return your calls. But the thing that I like to think about with true enterprise is don't go broad too quickly. Try to make two happy customers. And if you have two, you can make four. And if you have four, you can make 10. And I would say the model I've seen work most effectively is picking the right design partners for your platform and truly making them successful and not getting greedy until you have a representative happy customer. Your ivory tower view of what your product could be and should be is probably wrong. It's the opposite of the way you do say product like growth where you're really trying to get a lot of at bats and really try to look at the data, really have a first principles view of the product.

‍

Bret Taylor (00:59:59):

It's just hard to get a lot of at bats in the enterprise. So I think picking the right representative design partners you want for your platform and just making them successful, the way that fails is they weren't representative, right? But at least you have a happy customer. Yes. And I see a lot of people who say they want to do enterprise, who spend a year in an ivory tower making a product and not actually getting feedback from their customers. And the deal cycles for enterprise are long enough that if you imagine waiting two years before releasing your product, then waiting, I don't know, nine months or whatever to figure out it was wrong. You just wasted two years and nine months to find out you wasted two years and nine months. So that's the thing is I just say you have to be incredibly customer centric if you really want to go after the enterprise

‍

Christina Cacioppo (01:00:46):

And what are the right design partners?

‍

Bret Taylor (01:00:48):

This is where it's more nuanced. I think if there's technical risk in your product, you want to find ones who exercise that technical risk. If you are focused on an industry, which is a really great model for enterprise software companies, are they in that industry? Are they representative of other companies in that industry? Do you have the right use cases? A logo is a logo, but as you know, departments are far and wide within larger firms and you can end up with the right brand and the wrong use case. And again, if you're trying to learn from your product, it really needs to be that. And so I think that the art form is finding ones that sort of exercise all the risks you see in your hypothesis for your product, finding the right use case and then figuring out in say an industry or a market that you're interested in, are there specific customers in that segment who would just uniquely benefit from it?

‍

Bret Taylor (01:01:40):

Right now, maybe they're the number two player trying to overtake the number one player. Maybe they just went through a leadership transition and they're looking for a win in a certain area. Maybe just if you're a cybersecurity company, maybe they just had a cybersecurity crisis and you can go help them do that. So there's a lot that's probably very specific to what you're building, but that's the way I think about it. You want it to be representative. You want really exercise the product and you want both the quantity and substance of the design partners to represent kind of the universe of what you want to be when you grow up. So when you come out of it, you're not overfit to any one of those customers.

‍

Christina Cacioppo (01:02:18):

Great advice. Got some rapid fire questions for you.

‍

Bret Taylor (01:02:20):

Yeah, go ahead.

‍

Christina Cacioppo (01:02:21):

A book or article you've read recently that's changed how you think about things?

‍

Bret Taylor (01:02:25):

I mentioned the Tyler Cohen interview highly. It was a podcast, I can't remember which one, but it was really, really good. So

‍

Christina Cacioppo (01:02:34):

The Rakesh one I bet

‍

Bret Taylor (01:02:35):

Yes, that's right. Thank you. As I said, for me anyway, it changed my perspective on the future of ai.

‍

Christina Cacioppo (01:02:42):

One historical figure who would be a great startup founder, who would it be?

‍

Bret Taylor (01:02:47):

I got to go with Ben Franklin on this one. Founding father operated a printing press, kind of like a jack of all trades, but clearly deep on a lot of them. So I think he was a great founder in his day, but it'd be pretty cool to see what he'd do with all the toys we have nowadays,

‍

Christina Cacioppo (01:03:04):

Tech trend that everyone's really into but might be overblown the iPad of the moment one might call it.

‍

Bret Taylor (01:03:11):

That's interesting. I am somewhat skeptical of virtual reality, not categorically, but perhaps the size of the market. I just wonder if people want to have something over their eyes and their face for an extended period. So my hypothesis is that it looks a little bit more like say the video game market as opposed to something that's truly general purpose. I might be wrong actually on all these, I like being proven wrong. So the counterargument that I've thought about is when I saw Apple's approach here, the way they created an immersive set of screens for your computer, I was like, oh, that's really interesting. I hadn't thought of that. So there are worlds where there's sort of productivity gains from that that I haven't seen, but just in general, I think we're human people. We like to look at each other in the eyes. And so I am slightly skeptical that the size of the VR market is as big as some people have said.

‍

Christina Cacioppo (01:04:10):

What was something that seemed like a big deal at the time, but turned out not to be

‍

Bret Taylor (01:04:15):

During the pandemic? There was that period where Clubhouse and others had kind of, I don't know, describe them like live podcasts or I'm not sure what the category is. It sort of burned bright and fast, but didn't really sustain. And I wonder if it was a pandemic era trend as opposed to as an end enduring product.

‍

Christina Cacioppo (01:04:37):

Brett, it's been a pleasure to have you on Frameworks for Growth. Thank you so much for sharing some of what you've learned over your career and advice for today's founders.

‍

Bret Taylor (01:04:45):

Thanks for having me

‍

Christina Cacioppo (01:04:45):

And thank you so much for watching this episode of Frameworks for Growth. Hope you enjoyed it. If you're curious for more, you can check us out on Vanta.com. Thanks so much.

‍