a16z Podcast - Growth vs Efficiency: Can You Have Both?

Episode Date: July 21, 2023

Despite the ease of product building, sustainable growth has become increasingly challenging as many traditional channels no longer deliver the same results.In these challenging times, we explore the remaining growth opportunities. How can we achieve a balance between efficiency, profitability, and growth? Which channels are still relevant and how can they be effectively mastered in 2023?Join us as we discuss these questions with three seasoned experts who have successfully navigated similar confusing times in the past: Gina Gotthilf, leading Latitud and renowned for her impressive tenure as the VP of Growth at Duolingo; Kieran Flanagan, a long-time SVP of Marketing at HubSpot; and Bryan Kim, Consumer Partner at a16z, who held various leadership roles during Snap's hypergrowth phase up to its IPO. Topics Covered:00:00 – What growth opportunities exist in 2023?03:00 – What still works in growth?05:02 – The fundamentals and framework of growth07:02 – Assessing growth potential and product lead growth09:33 – Distribution11:15 – Your moat and growth14:00 – Integrating AI for product lead growth16:23 –Trust as a moat17:23 – B2C growth - discovery and breaking through18:51 – Can you get ahead of the incumbents?22:03 – Examples of companies and products24:59 – How do you foster a growth culture?28:38 – Experimentation  34:20 – How companies using AI in distribution44:43 – Growth challenge: What would you do with $10K? Resources:Check out Latitud: https://go.latitud.com/Check out Zapier: https://zapier.com/Find Bryan on Twitter: https://twitter.com/kirbyman01Find Kieran on Twitter :https://twitter.com/searchbratFind Gina on Twitter: https://twitter.com/ginag Stay Updated: Find a16z on Twitter: https://twitter.com/a16zFind a16z on LinkedIn: https://www.linkedin.com/company/a16zSubscribe on your favorite podcast app: https://a16z.simplecast.com/Follow our host: https://twitter.com/stephsmithioPlease note that the content here is for informational purposes only; should NOT be taken as legal, business, tax, or investment advice or be used to evaluate any investment or security; and is not directed at any investors or potential investors in any a16z fund. a16z and its affiliates may maintain investments in the companies discussed. For more details please see a16z.com/disclosures.

We need to try to get back to like fundamentals and how do we actually iterate and ship fast. We have to build products that are constantly evolving and that are constantly learning and evolving from those learnings. The only way really to stay ahead of that curve on that consumer side is get the lightnings keep hitting. I mean, that sounds hard. Be real, that's incredibly hard. Second product hitting so rare. Third product hitting, you're a generational company. It's like there's this big wave coming and it might hit your ship.

Is there a way to avoid the wave? I don't think so. I think it's kind of a tsunami. If you sit there, it's just going to wash you away. People will want to talk to people so, like, community becomes much more important, referral becomes much more important, where the mouth becomes much more important. The new limiting factor is possibly creativity. That's a very, very fun place to be.

Despite it being easier than ever to build a product, it may actually be harder than ever to grow one, especially sustainably, with many tried and true channels no longer working as they once did. Paid platforms were hit with app transparency tracking, SEO is now threatened by AI chat, social graphs have moved to interest graphs, product-led growth is easier than ever to copy, and virality is just as transient, if not more transient than before. And those are just a few of the examples that founders are facing. So what growth green space still exists, especially in an era of austerity?

How do you balance efficiency and profitability while still growing? and what channels matter and how do you master them in 2023? These are just a few of the questions we addressed in this episode, together with three people that have successfully grown companies through similarly confusing waves of the past. First off, we have Gina Gotthelph, now leading latitude, but perhaps most well known for her impressive reign as the VP of Growth at Duolingo,

growing them from 3 to 300 million users with no paid budget. We also have Kieran Flanagan joined the conversation. long-time SVP of marketing at HubSpot, where he was a founding member of their international business all the way up to HubSpot hitting over $1 billion in ARR. And finally, we have our very own Brian Kim, partner at A16C, who also held several leadership roles at SNAP, from their hypergrowth days all the way through IPO. All right, I won't keep you waiting any longer. Let's dive in.

As a reminder, the content here is for informational purposes only. should not be taken as legal, business, tax, or investment advice, or be used to evaluate any investment or security, and is not directed at any investors or potential investors in any A16Z fund. Please note that A16Z and its affiliates may also maintain investments in the companies discussed in this podcast. For more details, including a link to our investments, please see A16Z.com slash disclosures. What are you seeing in terms of growth that still works in 2023? Are there white spaces that companies can tackle? What's happening is kind of really interesting because there may be, like, AI is somewhat of a tailwind for like maybe 12 months or 18 months for the best people.

And then who knows after that, maybe a headwind across everything. So I think certainly in search, I really dug in there. And there are lots of examples of the click-through rate. decrease in and the amount of traffic you get from Google search pages decrease in. But I still think there's like a lot of value in that. Paid ads very similar. I still think there's a lot of value in that. What really happened with the popularity and growth of the internet is like marketing

growth became much more measurable. And they could have a seat of the revenue table and they could say, I put a dollar in here and I get $3 not here. And the growth that's starting to happen today or where things are starting to shift is a little bit back to like this indirect causation. These channels, like if you look at the platforms that grow and it's like short from video, TikTok, YouTube, like these kind of media. channels. If you look at where I think things will go with the evolution of AI, people will

want to talk to people so like community becomes much more important, referral becomes much more important, where the mouth becomes much more important. All of these things are going to shift us back into like the place where we have to be okay doing things where you can't measure everything directly all the time. And I actually don't think that's a bad thing. I think the over index of measurements forces everyone into the middle and everything becomes like mundane and lacks creativity. And so I think that there's like some pros and cons to actually this happening, but there's certainly a change in terms of how you are able to do your work and actually relay the value of that work to like the founder, who the founder is for the most part still going to be stuck in the binary, give me $3 for the dollar that I give you. We're almost like going back to the origins of advertising. Like there's that saying that says you need to hit someone seven times before they convert.

And Facebook maybe changed that. Maybe it was like two times. and then it feels like we're sliding back in the other direction. I think it has always been the case that people need repetition to really notice something. And yes, more and more, there's more media, there's more for us to watch and listen to, and we're more distracted. So I think that that becomes even more the case. But I may be simplistic, but I don't think that much is changing because to me,

thinking about growth is about the framework for understanding what makes people tick, what people care about, how to reach one person or one, media outlets or one channel, whatever it is, that will reach a lot more people in the most efficient way possible. What are the moments in your product that people really, really care about? Meaning, like, a number of people who actually encounter that moment. And if you actually make some sort of like AB test or change, you'll actually see significance as a school results on the other end, because people really care about that one moment. It could be a notification. It could be a screen. You know, we keep seeing these trends and things grow and

ebb SEO rules are constantly changing because we're at the whim of algorithms and so we're just constantly trying to adapt. Places become saturated. Like at one point it was really trendy and cool to do ads inside the subway in New York City. Like someone discovered that in 2012, which is funny because it feels pretty obvious. Then suddenly everyone was doing it. Everyone's testing with notifications and everyone's saying there's too many and it doesn't matter. I think all of those channels are still very relevant. It's about finding how to actually impact how people feel and finding, you know, new ways of impacting people that are cheaper because less people are using it over and over. So I think that that will continue to be the

case and it just depends on what's the trendy channel and what are people paying attention to. But if you think a priori like that, like how do I reach one that can then reach a lot or what are the things that I can modify that actually make a difference, then I don't think that much has changed. I have like a slightly heretical take, so this is going to be fun. But, you know, when we talk about gross, especially, I know ACKD, we're all in Silicon Valley and the stalwart, what have you. But when I think about gross, this like gross tactic, gross, you know, channels and even the word gross hack became very popular in that, you know, there's this prevalent theory and sort of Kieran touched on it where you can put in a dollar and get back why, put a next dollar, get Y back. And I think that certainly became true. And there became so many channels and there became so many wisdom and try it and true tactics, whether.

whether it's SEO or ambassador program or community building where everyone's trying to learn from each other. I'm like, how do I hack this thing? Gina said something interesting is sort of how I think about it, where like moments in the product when people go, aha, like this is great. At the end of the day, I think Gross, we're sort of conflating two things. One is the actual product, is this sticky. Does it enable you to continue to come back? Because Gross at the end of the day is like, how many people continue to use your product?

It's like not a singular metric of like on day X, how many people showed up? Well, that's useful for today. But actually, for a company that is an ongoing concern, what you want to know is, like, how many people consistently show up? And how does that grow? It's a cumulative concept, which means that, you know, the fact that people come back day and day out is probably one of the most important factor in assessing the growth potential and growth of it, which to me goes back to a little bit of product

where when people first come on, like we all know this, but, you know, the biggest drop off typically day one or day zero when, you know, people come on, They do onboarding, and maybe they didn't get to the aha moment quickly. So why would they come back? If they don't come back, then your game as a gross is like, how do I get more acquisition, right? Because gross to me is also sort of a, you know, numerically almost an equation. There's like, you know, people who showed up yesterday plus new people that you acquired minus the term people, plus the resurrected people.

That's sort of like an equation. And when you think about how to get the retention up, I think it all goes back to product. Why, as founders and operators, why is there gross in your product? Like, why is it useful for more people? And is your product constructed in a way, as Gina said, like, you know, helps people understand what your product is about when they first use it? Can you deliver that product value again and again consistently? I think now we call it product like gross.

But it's honestly like a core product value and how you actually define that is one. And a lot of things that we're also talking about here is like distribution. So there's like a product and the distribution side of it. And if we think of like distribution, I think, you know, the panel is exactly right. There are areas where it's less good now, and there are areas that are continuing to be good. You know, you talked about advertising. There's that saying that, you know, you never know where half of it goes. And, you know, maybe we're going back to that.

But, you know, I think there are many other areas that continue to work well, whether a subway ad or remnant TV as or radio. It depends what your product is. And when your intended audiences, if you're, you know, marketing or appealing to young generation of, you know, gamers or communities, are niche areas and Reddit and Discord are actually pretty good places. I also think, you know, ATT did kneecap a lot of the sort of paid advertising side of things. But, you know, the AI, like here and I also talked about, can also be used for that purpose, right? Facebook or meta is investing heavily on, you know, AI side of it so that sometimes all you need to do is just say, hey, like, I have this idea. Here's like the base creative. And now they're able to like change how it looks,

to add different, you know, copy, like change the thing to see what works automatically at scale with AI and deliver higher value. And that's also, you know, very promising, I think, direction that we're seeing where the efficacy of the paid side, it may also be restored. Yeah, I mean, I like the breakdown between there is product and there is distribution. Those latter up to growth. But maybe if we just attack this idea of building a good product, that's not a new idea, but it does feel like, at least to me, maybe we are in a different era where it's easier to copy products. And so is it just about being a first mover in an era where things are easier than ever to build?

And maybe Gina, you can speak to that because Duolingo kind of famously grew with no marketing budget. Is that correct? Or a very little marketing budget. And it was exactly product-led growth that resulted in that company becoming so big. Yes. So what I think has really changed with tech is that people could say, oh, my product is really good before. It wasn't as easy to know whether a product is good or not good as it is now. You can actually see that in metrics in terms of, as Ryan was saying, just in terms of retention

and people coming back again. So now that defines what a good product is. But it's not just about building a product that's good, but also the fact that with tech, we have to build products that are constantly evolving and that are constantly learning and evolving from those learnings. But the main thing, I think that like when you talk about duolingo, what we're trying to do at latitude now, which is my new company, the important. of not just building a great product, but understanding really clearly what your moat is.

And your moat needs to be something that makes it really hard to replicate your product. And it can be anything, right? So for example, if Kim Kardashian launches a shoe that's exactly just like every other shoe that exists out there, the moat is Kim Kardashian because her brand and her audience and the fact that she is an influencer, she's not telling people to buy the shoe, that's a moat. It means that like it's the thing that makes it really hard for you to go and just do that again. With Duolingo, the product is the moat. And that's the most incredible thing.

Because of what I just said before about the iteration and the iterative practice, anyone at any particular moment in time, and many did try, could actually go and copy exactly what Duolingo was doing. They, in fact, use the same font. In many cases, they use the same design colors, like the same format. You can do that. But you don't have the data and the team behind it that knows how to use that data that is constantly learning from every single day of all of the usage that we're getting

and understanding what's leading to every single decision and why we're modifying things in a particular way, that's impossible to catch up to. So at every moment in time, even if you mimic exactly what Duolingo is, there's no way you're going to get to the next day of Duolingo because you don't have all of that history or the knowledge of how to use that history.

So you could just spend all your time trying to copy Duolingo every single day, but that doesn't actually get you to solve a problem for a consumer and understanding how what you're doing is actually solving a real problem, which is at the core of, as we were saying, a retentive products that people actually like. And that defines a good product. So it's not as easy as it sounds. It's definitely not as easy as it sounds. Kieran, I want you to tap in here. You now are the CMO at Zapier, which I feel like actually is a great parallel to what

Gina just said. I mean, Zapier has so many, you could say clones or competitors, right? Like, I've used many of them. Many of them are cheaper. How do you think about that differentiation and building a product and to Gina's point, like getting the right data and involving it in a way that you can stay ahead. You can have a moat. Yeah. I think first remover advantage is useful if you can build a healthy retention curve and that retention curve helps you make that product much better over time. And so you're just like making it much harder to compete with you. And that can be in all manner of things. Like Zapier's moat for some time was just they had much more expansive coverage of apps and the amount of things that you can automate with. And then more recently,

it's like how do we integrate AI? How do we make automation much, much easier for a company? how can they actually do natural language within their apps. One of the things that's super interesting about the AI part, just going back the distribution and then the product part and the product-led growth. And AI is a phenomenal unlock for product-led growth because product-like growth will get naturally built into every single product because every single product will have a natural language interface. And so you will actually be able to onboard products and use products

without actually any human intervention. And I think that's an incredible augmentation of how apps work in general. But I think the moat part is going to be much more complex. Like I'm talking through the lens of B2B, not through consumer apps. But certainly for B2B, it's going to be somewhat more complicated to figure out over time because AI does commoditize things and make things much easier to replicate. And we've seen like AI maybe one of the few platform shifts that actually skews towards the monopolies and conglomerates versus like these individual companies who just get sucked up

and there are features of these existing platforms. Like it kind of skews towards if you have distribution, you might wing. I think AI features might be more of a retention play than an actually new acquisition play, which is like if you build the AI features into a platform, no one will go leave you to actually use all these different like point solutions. So I think modes will exist in similar formats. There maybe just companies with less of them. And I do think distribution is a moat. Retention is a moat. Network effects will still be a moat. Community is a moat. And so I think of all of those things in relation to Zaptur, like how are we touching in all of those things as we continue

to go to that company? One thing I want to respond to what Kieran said, at latitude, I would say the community is our moat, but the reason why the community is remote, in addition to the distribution aspect of it, is the trust building. So trust can be a moat because that takes a while to build. So for B to B product, I think that's also super relevant because someone can come in to the market and just like offer the same product, but you've built a brand on the Zapier side that is super trustworthy, you know, and so people will continue to trust new features or new solutions that you put out there instead of trying to, you know, maybe try a new product. I think that that's another aspect worth considering. By the way, Kieran, how many times a day do you get asked if it's

Zapier or Zapier? I don't even know if I say it correctly. I say Zapier, but it may be Zapier. I should probably ask. I heard one of the founders say it's Zapier. It rhymes with happier because it makes you happy. Oh, there you go. Okay, great. I learned something about, I learned something about it That's how I thought about it. It's zaps. Yeah, yeah. The main reason I join is just I like to say the word zaps. Like, it's just a cool world.

I know we talked a little bit of a B2B. Just talking about more of the consumer side, gross and thinking about distribution, it has changed. I think, you know, a long time ago already, Mark Zuckerberg, Meta already said, you know, it's a battle for the home screen. And more and more, even recently, people are downloading less and less apps over time, part one. Part two is, it's harder and harder to get discovered in the App Store. You know, App Store ranking is hard to sort of get discovery now. So now you have this issue of, one,

people are less willing to download brand new things. And two, even if they sort of kind heard about you, it's like harder to get discovered. So I think part one of the consumer side, the platforms have, for lack of a better word, maybe I can call it ossified, where there are two distribution platform, Google Play, and Apple Store, great, and they just distribute apps. And people redownload a ton of apps. They retry a lot of things. But brand new things, not so much. Or certainly less than before. So now you have this conundrum of like, okay, like it's just harder to break through the noises. And we talked a little bit about, especially Kieran talked about, you know, maybe AI is an accelerant for retention, maybe less of an acquisition.

I think that's correct. And that's interesting where also we have a lot of these like incumbent right? Like, especially in the consumer side, incumbents come with incredible modes. They already have the network effect. They already are synced to everything. They already have login for everything. They already have all your data. They know what you bought. They know you. So you talked about how do you stay ahead? It's very, very, very difficult, actually. Even if you have a clever hag, all these companies come up with incredible new consumer product. And they enter the zeitgeist. And through entering the zeitgeist, they're able to solve the first order question. How do I break through the noise and get distributed initially

and have more people to try it out? That alone is a feat. That's incredible. If you have highly retentive product on top of that, legendary, that's amazing. Fantastic. And all these that product that we talked about did have that. Now the question is, okay, copying. Gina, you talked about people copying down to fonts. I understand that so deeply. I have that Snapchat for four or four or five years. And, you know, down to the font and every single like UI level, yes, they copy everything. And how do you actually fight that as a new entrant, especially when incumbents come bearing a lot of these already like weaponaries, if you will, that have retention and data and everything? It's hard. And I think actually, and this is like a sort of difficult answer, the only way

really to stay ahead of that curve on that consumer side is get the lightnings keep hitting, which is so, so, so hard. I think getting, you know, the first hit is so hard. Second product hitting, so rare. Third product hitting, you're a generational company. So it's sort of that innovation curve that I think is like very interesting. So for the companies that are newly entering and building new product, and we're on the side of the builders, right?

incumbents may have like all the advantage, but we want to see new products. We want to see new, you know, hungry founder that are building for the next generation. And they're not happy with the current set of solution. So the only way potentially is, you know, there's series of things. But like step one boss game is like, how do you get through the distribution noise? Step two boss game is, you know, how do you get retentive product? Great. Step three, and this is like a holy grail in my mind is like, how do you get that lightning to hit?

Continue to enter Zykeyes. And for that, you know, I think it's. thoughtfulness. I think it's the structure of the company, I think the culture of the founders and the team of wanting to continue to put out the best and brightest work that they have continuously versus potentially resting on laurels or optimizing for things that I've worked in the past and thinking that their first iteration product is what is going to bring them to the next, next, next level because they may not be because the distribution advantage of the incumbents as well as the difficulty of getting further discovery. What's an example of that?

That sounds really hard. Is there an example of a company that you've seen execute on that hard mode really well today? I mean, you know, one of the last examples I've seen in, you know, one of our portfolio companies, Be Real, has done that too with certain features. But if you think about, I'll talk about snaps as that's the name I know well and, you know, it's been copied so many times, what have you. You know, the first hit, one may think it's a disappearing image. That was revolutionary. That people thought of images as something to keep on your phone. It's like for recording and for keepsakes.

You turned that into disappearing thing, which made the medium itself a message, communication method. That's revolutionary, very interesting, you know, sexting was the first use case, but it, like, catapulted the company in new heights. Great. So that was going really well. Now then the kicker is they then started to add stories, right? Stories of, you know, these disappearing photos like linked together that shows you a day and time in your life and a passive, way but also goes away revolutionary in that sense then that to me is a second zeitgeist hit which by the way what a great example of something being copied to right everyone has so many

different everybody has it like LinkedIn has stories like no dig on LinkedIn but like everyone has stories now so you know it's a medium and like a it's a thing that worked really really well and potentially even burst like the short video type thing that's trend so to think about the second innovation that's like a huge innovation third all also is, you know, they didn't rest there. It was like, okay, we are seeing more and more people abandon the front result of the front facing camera. What is the problem?

Well, they don't like what they look like. Well, let's add dog ears on people's face. Let's add like rainbow pukes on people's face. And again, that's another thing, Steph, where everyone copied. Like, everyone has lens now. Like, it's very, very good at us in TikTok, you know, puking rainbows, blue eyes, what have you. But these augmented reality effects are now prevalent in consumer products. So this is one example where I think each subsequent hits have essentially catapulted the company in the new height.

And that, I know from inside, was very difficult, very long term with several bets going at the same time and being very thoughtful of what to kill and what to put up there. So it's that system of innovation, it's that hunger and continuous paths that I think is very important. that speed and hunger and being able to implement all that at a fairly large and established company, I think is a very difficult thing. So I think there's a cultural aspect. I think they're the founders mentality of always wanting to be ahead of the curve, always implementing and sort of looping in the new products that the users are looking for. How do you foster that culture? I think especially in 2023 where maybe budgets aren't as big, maybe some of the talent you had before is not used to that austerity or that idea of having to ship really quickly,

maybe spend less money and just figure it out. I've talked to a lot of companies this year that were CD public. I feel like a lot of founders are actually excited to get back to shipping. Like I think we went through this ballooning of companies during COVID. And everyone's like, wow, like I don't really know how this company even works anymore. Yes. And we need to try to get back to like fund a. fundamentals and how do we actually iterate and ship fast. But I think what really is excited in tech is like getting back to the reason that people were excited to get into tech. Like certainly I was as a failed software engineer is like to do work. Like not to like figure out how to like coordinate hundreds of people, but I actually like to get back and do work and ship things. And I think the faster you ship, the more you learn. So I think it comes from the founder. I guess it's a long winded way to say like I think you'll have some founders that embrace us. And actually at their heart, they actually thrive and they kind of chaos.

They actually enjoy this period of time, especially like a lot of founders who are building with AI. I've met lots who are like, this is the best time to be a founder because we have no idea how this is going to go. And they love that. They love like not knowing what the end result is rather than just tweaking the dials and trying to eke at little bits of growth each and every day. And that trickles down from them. They set the tempo and it's up to the rest of us to up level to meet that bar. This comes from two places. One, as Kieran said from the founder, so from the top.

And I think that what that the founder can do is come up with artificial deadlines. For example, we have to launch in September because we're doing an event for 4,000 people, and that's when we need to launch. And that's it. Or like a duolingo would be like, oh, you know, we told Apple that we're actually going to make this sprint, and now that's there. So we just have to do it. And so suddenly it's not like something that, like, can be pushed because there's this external deadline.

A lot of times it was artificial because you kind of create it. But now it's set in stones. I think you can help from that perspective. And the other one, I think, is actually creating a culture from the very practical point of view of, like a process. Like, for example, sprints, you know, so you have like two weeks sprints. And that's sort of the cadence at which things happen, and maybe that's the cadence at which you're doing A-B testing. Making sure that you are prioritizing as part of that sprint and that you're not just shipping

to ship because then you can just like spray and pray, but you need to like really be learning from everything that you're shipping and iterating. So that has to be part of the process, whether it be retrospectives or whether it be having accountability. Like our product team at Lattitude has a show and tell every two weeks and where and sort of the whole company is there expecting to see what's evolved. at Duolingo we had a practice of making sure that every experiment had a proper write-up in the beginning with all of the hypotheses and what we actually expected.

And then like two weeks later to be able to come back and then look at all of those because otherwise we'd be running around thinking about the next test and we'd forget to go analyze, think about what we learned and apply. So I think that all of those, you can sort of have little levers that create the culture from a practical standpoint so that you can set and forget and not have to, as a founder, keep watching and seeing if people are moving fast and iterating. It's so funny you mentioned the artificial deadlines because I was thinking about this recently, about how, you know, when you go through college, you have exams, you have projects that have due dates.

And then when you enter the workforce, especially over the last few years, you have those, but they feel very, very gray. And so I think it's just a funny acknowledgement about how sometimes that works in the workforce. But, you know, you spoke to experimentation. How do you think about that in 2023? Should companies, especially when things are tighter, should they be leaning more into experimentation? because it's like, okay, we got to figure this out. Like Kieran said, we don't know where AI is going. Or should they be really drilling down on, you know, maybe the one or two channels that they know work? I always think of, like, your growth initiative as a portfolio of bets. And your portfolio

is, like, very similar to, like, how you think about investing. You want to have a spread across things that you have a high degree of confidence in. And the thing that you have high degree of confidence in is where you have historical context is say, we've done something similar before and it will likely work. And I can measure that against, like, what's a potential upside? But you definitely want to take some things that are like more my speculative guests. Now, all my speculative guests in terms of investing from 2020 to 2022 are all down by 80%. So they definitely do lose. I've lost a lot of money. But in terms of your growth portfolio, you want to take those big swings. And I think one of the things that happened to your point, Steph, that's phenomenon

of like, this is kind of a deadline, but it's not really a deadline. Like, I've been in tech for a while. That definitely was not always the case and became the case when tech was ballooning. And I think it's good to get back to actually like this stuff truly matters and hidden these dates truly matter. But I think the other thing that we've got away from is like having people in the company who will take risks, who have deep convection about things, who will take risks, who will like say, hey, we think this is the right thing to do. Give me a couple of people and I'll either prove this at weekends or nights and I'll figure out if this is like worth doing or not. And I don't have any historical context to say this is a good idea, but like I've

talked to customers and I think this is the thing that's going to work. And so I do think that we need to have experimentation across the entire spectrum, like from like high degree of confidence, high impact to like really game changer thing, not a lot of context to say this works, but like is a real thing we're doing. And if you are at the point where you can have statistical significance because you have a lot of users, then I think, you know, you have to, everything and it needs to be an experiment, even the things that you already know, like the channels that are already working. Because if it's not an experiment, then you're not looking at it from a very scientific

perspective and gaining data that will actually tell you whether you can replicate this when you launch it to a lot of users. But I agree 100% with Kieran in the portfolio approach. It's not either or it has to be and, but I do believe in being as data-driven as possible. Of course, there are exceptions and there's an art to things, but to be as data-driven as possible, and that's where you find experimentation to be super important. And if you are an early-stage startup, you need to take big swings, because if you get incremental improvements in things, you're just not going to get anywhere.

You can, like, improve things 2% over and over, and no one will notice you. And the way that I would approach it from a very tactical perspective is I would list all of the ideas of things that I would like to work on, whether it be like tried and true channels that I want to iterate on or big swings, and I would just try to rank them based on ROI. Like, how hard is it going to be to do this? What do I think is the return going to be? And then just try it. And that's the information I would use to make a decision about the portfolio of things would be working on as a team. Yeah, just to give one example on the big bets, because I think sometimes we think of these things as like the kind of AB test.

For HubSpop, like acquiring the hustle for us was a bet, right? We had deep convection about the future of B2B content needed to be much more engaging, and need to be part of the everyday presence of how people interact with content, wanted to engage with content. But we had never bought a media company before. You can't A, B, test that. Yeah, we had, like, never had, like, media talent

that we had to, like, figure out how to integrate to the company. And we used that talent to, like, build a podcast network. You built the creator program for us, Steph. We built all of this stuff, a really incredible presence on YouTube. And we got incredible talent that I think is a real moat against most B2B companies because they do not have this talent within that company, that's like an example of like a sizable bet

mapped to something that we had deep conviction about. So I think those kind of things do matter. There's like a fun matrix of like important and urgent and important non-urgent, like the two-by-two type thing. And, you know, some of the bets are, you know, both important and non-urgent, important and urgent. But then you have these other, you know, ideas where you have this inkling, especially in the consumer side,

it's nearly impossible to predict how something would do, especially if it's like a fairly large bet. So there's a little bit of an intuition where, you know, you sort of rank a couple of options and like, look, I think there's a big product bet and that could be one of the experiments or sort of projects that you work on. Kieran, you said something earlier, I really loved, which is at the end of the day, the founders and operators do what they do because they want to ship things and, you know, ship things that people use. Back to the fundamentals. I keep thinking that's one line where like the ComC never made good sailors. And so, you know, we're not in a like a common SC. And then what are good sailors made of?

And they're probably made of like a lot of experience, but also knowing how to do the basics really well, really, really, really well. And so going back to that fundamentals, to me, is really interesting because there are two different ways to deal with this, which is like, how do I get out of this and what do I do? And there's all these like tactics and hacks that you might be able to do. But there's another point of like, no, actually the way we become like an important generational endearing company is to get back to the foundation and fundamentals and build stuff

that people want to use and genuinely as helpful to them. And I think that constitutes some of the ways as how people can think about making big bets. Totally. Maybe one topic we could just address directly is AI, because on one hand, you're all kind of speaking to this idea of the fundamentals, focusing on what really matters. But then you have this big wave, like, you know, to use your sailor analogy, Brian. It's like there's this big wave coming and it might hit your ship and you got to figure out how to integrate it, avoid it. How are you seeing companies really kind of intelligently, creatively on more so the distribution side or thinking about how companies can grow? You know, look, this is very new and I'm very new to understanding AI. So I just put that

there. But I think that from a distribution perspective, there's two ways to approach AI. One, what are the ways in which we can use AI to reduce the work that we're putting into things. So, like, when we're talking about ROI, like, that's the investment. It's just, like, reduce amount of time spent on things, like, for example, copywriting or image generation. I am a purist when it comes to copywriting. So, like, I think that everything needs to be, like, Pulitzer Prize-winning copy. So I don't actually like copywritten by chat GPT, but it's actually, like, 75 to 80%

of the way there. And so I think that you can probably use that, you know, to benefit your company in many ways either by helping organize ideas or coming up with a V1 of copy. And I think you may be able to do that with image and potentially video as well. So I would think about how to leverage AI to reduce work, to automate things and just make things cheaper from a time and spend perspective. Gina, by the way, you got to incorporate that into your prompt. You got to say YouTube title written by Pulitzer Prize winning writer. Yes, I'm sure. I think maybe we'll get there. But here's why I don't think it will get there because I think that human emotion and the ability

that humans have to understand each other and how to make each other feel is going to be really hard for AI to pick up on. And so we are able to come up with just that one word or just that one punctuation that really drives a message and that's going to be hard to replicate unless it becomes a little bit of a recipe, at which point it stops impacting people because people become a little bit numb to that. But then that drives me to the second point in which I think we can use AI interestingly, which is people are very impressed by AI right now, means they're more likely to pay attention, which means they're likely to think it's funny, which means they're likely to think it's moving or shocking. And those are all things that

help your brand become remembered. And ultimately, that's what marketing and brand work is. And so I would leverage AI in as much as I can to accomplish that. What I think of AI for distribution, I kind of split it into two categories. And Gina covered one of them, which is there's a lot of people that are just using AI to replicate the things they do. And so that really is just like a cost saving. You know, marketing I think might move from like more of a, channel domain expertise to more a generalist. Like we have generalists in early startups. I think that will actually be true of like stage companies

because these generalists will be able to stitch together these kind of micro apps to create customer experiences by just using data sets and using AI modeling. And that will suit like operators and technologists. That's great news, by the way, because like so many of us are generalists and I'm like, oh, great, I'll have a job. Yeah, yeah. And I think people are like enjoy data, enjoy learning how to use apps

and are able to think like product builders. To think marketers and growth people need to think like product builders. Like, how can I use this to sit together some sort of build an experience for someone? And I actually, how do I creatively do that with these apps and data? So I think there's like to replicate what we already do. And you see a lot of this with like outbound sales, right? Like that's the number one use case you see right now. Hey, I can drag in all this content and use chat to BT to create very, very cool like,

outbound sales. The stuff that I'm interested in is like maybe that's 90%. The 10% is like, what are things you can do with AI that is not possible today that helps in distribution? I'll give you like one quick example. So like if you come into Zapier because it's like 6,000 hours, and there's, I think, if you actually, all the different, like, triggers and things, like millions of different variations of what you could actually automate.

So how would I ever know to figure out, like, what I can onboard you to or, like, what you need to do when you actually stand up's user? And the best way I could do that is, like, build an automation guide just for you, right? And today, the way I would do that is I would get someone to reach out, talk to you about your tech stack, figure out what role you are, and then contextualize workflows in a language you understand that are, like, contextual to your job, right? Hey, Steph, you're a marketer. You want to generate more leads, you want to send leads to the sales team, you want to enrich them, you want to make sure they can close in the customers more.

Like, you understand that language and you understand why that's valuable. Again, today, human would need to reach out, write the automation guide for you, so not economical, not scalable. One thing I've been testing is, like, we've been testing, like, chat CBT being able to do this, like, grab data from like something like Clearbit or something like that, tech stack, be able to feed in your role, be able to feed in contextual information about your tech stack company, and then have that write an automation guide for you, and then actually be able to surface that automation guide, whether through sales, whether through chat, whether through all these different mediums. And it's really good, like, surprisingly good. Like the first time I did, I was like, boy, like, this is, like, wow, like this is something we cannot do at scale today. That's the 10% that I think if you do as a fast mover, you get this like little window

of time before everything gets commoditized and everyone just does the same stuff, that again, if you have good retention, you actually get meaningful growth there because you have that retention curve, you can get those users to stick, upsell, cross-sell, and actually that's really, really valuable. So there's the 90% that I think most people are trying to replicate the things that we already do. But that 10% is super interesting, which is what are the things we can do at scale today that we could not do previous to actually having this technology? I'm smiling because I heard this quote once, which is that the best growth hacks are never shared. So you've got to be careful, Kieran, sharing that 10% on the podcast. I need to

I'll keep the rest of the ones to ourselves. I'm smiling because this is such a fun conversation and Gina and you both hit on like sort of the non-replicability of like human and I'm thinking about this Nikon ad that just like came out a few days ago which is like don't give up on the real world and like it has these unbelievable image of real places and like a fake prompt that like would generate such real image and it was a very powerful message of like look like yes you know AI can do a lot of things but don't give up on the real world and it was very powerful and it's creative and you know can AI come up with that I don't know but. So it's very interesting. On the consumer side, a couple things. Going back to your question, Steph, it was like, what can the companies do to use it to your advantage? One is the willingness to try, as Gina said, as it's quite high, meaning at least for the consumers, for the first time in a while, they're willing to give it a shot. They're willing to actually download and use the product and try it out and go through the app download and all the things, which they weren't before. So for the first time, actually, you have this window of opportunity where people are willing to.

to give it a shot. So now, acquisition side, you have this sort of potential window where you can appeal as like either first mover or a great product to get a lot of sort of usage. You know, Kieran shared a Zapier example where you could also use it very strategically to deepen your core product value. Like, why are people using the product that you have out there? And how do you actually use this new capability, new wave, if you will, to further strengthen and deepen the reason and why people use your product. And sure, you can brand it AI and make it sexy and more people will be excited, what have you.

But I learned one thing from Snap where the luxury, the sort of lens product was a state-of-the-art AR product, like augmented reality product. We did not call it augmented reality for four years. And we actually implicitly banned it inside a company to call it an augmented reality because that's technology.

That's not the point. The point is for users to giggle and use it and have that emotion. So as you think about these new technology that are being enabled, and Kieran talked about the 90% and 10% net new things, but there are these like ways to deepen the core product value of why your product exists that I think, you know, you can call it AI, you don't have to call it AI, it could be abstracted away, it doesn't matter if it makes the job that you do much easier in two ways, one, either take care of the rot things that you're doing in a mechanical turkey way, great, or, you know, all of a sudden deliver value that wasn't possible before because, or cost prohibitive or net new, I think that's incredible. So, you know, people talked about, you know, is there a way to avoid the wave? I don't think so. I think it's kind of a tsunami.

If you sit there, it's just going to wash you away. So you do need to respond. And how you respond, I think, is important where rather than saying, oh, there's like a new shiny thing, we're going to bolt it on. I actually think it's going back to the fundamentals to say, hey, like, why do our product exist and why do customers use our product every day in and day out? how do I use this new capability to actually augment that I make that even easier? So I get very, very excited about existing companies as well as net new companies because

they both now have a new thing, right? Like for existing companies, there are ways to actually deepen your mouth, actually, use this product, abstracted a way to make it so much more useful than before. New companies, all of a sudden, you have this moment in time where people are willing to give it a shot. And in the consumer side, what I also get really excited about is there are these net new behaviors. I know we talked about how we'll never sort of fully rely on computers for creativity. But, you know, the cost of creation has nearly gone to zero now.

The post-production of this podcast, the actual creation, the ums and ahs and the eye movement, everything. You can actually pretty easily do that through AI potentially with a lot of new products out there. And all of a sudden, you know, maybe you don't need a huge. maybe you don't need all these amazing fancy gadgets to do what we do and the bar is lower the entry point is lower meaning that you know all of a sudden the marginal cost of creation overall whether it's image or video or prompt or pie what have you maybe quite low which then means the new limiting factor is possibly creativity and that's a very very fun place to be speaking of creativity i wanted to end this with a little bit of a challenge a prompt for the

three of you. You know, we talked about a lot of different things. We talked about product and distribution and how those come together. And, you know, one product that I work on is the A16C podcast. And if we think about austerity and trying to grow in turbulent times when, you know, maybe budgets aren't as high, if we had $10,000 to grow the A16Z podcast, given all the channels that you could invest in, how would you think about that? And please don't just tell me build a better product because we're working on that. But on the distribution side, how would you use that money? Well, from someone who has tried to build two podcasts, my favorite word for podcast growth is like grind. It is like grind. Because there's not a lot of meaningful things

that work for podcast promo. So I'll just give you a wacky idea instead. That may or may not be usable. So one of the things I would do in this has kind of been done, I guess, before, but it might be a more novel way to do it, is like get your audience. Like pick 10 entrepreneurs from way back in the past that are no longer with us all the way back to Henry Ford and all these incredible entrepreneurs get your audience to vote

and like what's the number one person you would like to hear it interviewed in this pod? Get them to vote. Whatever the top one is, use a tankade give to someone to create an AI version of that person. It's pretty easy to create a deep fake version

and then you should just interview them for half an hour and put that out into the interweb and that's probably a good way to like get interest in the podcast through something that will hopefully get a lot of social traction. I have an equally wacky idea.

Being of consumer mindset, I think one of the distribution channel that works these days is TikTok. So 10K, not a lot, but maybe the way to do it is find a intentionally quite controversial topic. We talk about it. Use AI to splice it up so that it's like entertaining and like very easy like to the point. And essentially the call out is fight us. Stitching coming, fight us. And like a hot take, fight us. And they'll actually use that remix culture.

of the internet to let that sort of do what it does. And the reason why the dollar would be interesting is sometimes it's interesting to seed these things. So have like one or two, you know, influence or whatever, like over founders who like we can, you know, invite over and like, here's a piece of content now go debate that and actually be pretty brutal because like people love a good fight and put it out there and see how that does. I actually kind of love this idea. Even just the tag Fight Us is so strong. That's cool. about it and I have a very boring idea, but this is what I would do. I think that your bet here would be to try to increase your audience by looking at places where your audience isn't,

but you have a huge potential audience. And for me, that would be people in older demographics that are very business savvy and have been very active their whole lives and kind of get technology but are a little bit out of the loop and don't know what A16Z is and I've never heard of anything related to it. And I have had conversations with brilliant people in finance or maybe like in their 60s, let's say, maybe even 50s, who have no idea what A16 Z is, which in my world is crazy. And those people read the Wall Street Journal and those people read the New York Times. So I would go and do something with one of those massive publications to reach those audiences.

The thing about growth is that it's not just about growing, but you have to grow in terms of like the right people. So one way to game your question, which is something that growth seems to do all the time, is I'm just going to go to a country where buying a user or buying a listener is super cheap and I'm going to go and like, you know, do some sort of massive campaign because $10,000 goes so far in like Southeast Asia or like Latin American countries and then just get a bunch of people to listen to OA16D, but it won't actually matter because they're not your core audience. But that would be a gameable response, which I just want to leave here because I think it's

really core to growth. In order for you to actually come up with a good growth strategy, you have to think about how can you game this problem. And if you can easily game it, then you're looking at the wrong thing. Totally. Thank you all. This was so fun. Thanks for listening to the A16Z podcast. If you like this episode, don't forget to subscribe, leave a review, or tell a friend. We also recently launched on YouTube at YouTube.com slash A16Z underscore video, where you'll find exclusive video content.

We'll see you next time.