YC SUS: Gustaf Alströmer and Eric Migicovsky discuss growth tactics

Exciting! Welcome to another week of Startup School. I'm joined this week by Gustav. You want to tell us a little bit about yourself? Maybe your background?

Sure! So I work here at YC as a partner. I've been here for two and a half years. Before that, I had most of my growth and product experience at Airbnb, where I spent almost five years on the growth team as a product manager, and spent time at different companies before that, working on growth as well. So I feel like I've seen some really good stuff at Airbnb and some really hard stuff at other companies, and that's where my learning comes from. I also deliver several lectures to the YC core participants during the batch on growth, and I'm kind of the main person that our founders turn to to ask questions. So you've got the right person in the room today!

We're going to take questions from the Startup School forum, we're going to take some from Twitter and YouTube comments if we have time, but we've got some from the forum already. So let's jump right in.

First up: from My Education Club, asks, "We're building a peer-to-peer lending platform. Christophe, you were talking in the video about starting to get paying customers as soon as possible and not to use paid growth until you have paying customers. How does this work for social network platforms or platforms with network effects?"

So historically, if you look at the last ten years, many of the social products were actually based on advertising. So the way that you make money would not be to charge users but to charge advertisers to advertise to them. What that led to is this idea that you should not try to monetize early; you just try to grow, grow, grow. And that’s sort of how many of the early products like Twitter, Facebook, LinkedIn, and many others grew. Snapchat, that's how they grew in the early days. They were getting users to use the product and then monetizing them once they grew.

Nearly all of them were growing without using paid marketing, so they were all growing through viral invites or word of mouth. LinkedIn used email and SEO is the way the LinkedIn grows. So, basically, two primary channels to LinkedIn growth will be those two, and I would say for products that have a kind of network effect, where for every user that gets on the product, the product becomes better, it really matters to get to that first million, a couple million users.

So LinkedIn in the early days was useless unless you had some coworkers to connect with. It's just hard to build a new advertising base. That’s how they like to make money. So correct—the business model for companies is either you're charging users or advertising or some other business model. And there are just a lot fewer companies today that get started using advertising as a business model.

The main reason is it's extremely hard to compete with Google and Facebook. They are the best advertising platforms, and it's really difficult. It’s not like competing with TV or newspapers anymore; you’re competing with two really great companies. So most companies probably shouldn't be going after that business model, which means they're going after a different business model, most likely charging your users in some way.

And if you're going to charge your users, I guess a good example is something that’s like a relatively new business model that evolved after the ad-driven kind of placement: they enable creators to actually charge money, to link streamers to actually charge money to people consuming it, and they earn, you know, a percentage of the transactions.

Right? So let’s go back to the question: how does have an ROI calculation? That means you know how much money you make back from every user you acquire with paid marketing, or you don't. If you do know how much money you make back, you can scale paid growth as long as you're making enough money back to pay back the money you've paid on marketing. If you don't know how much money you're making back, you can't really grow on paid growth. Everything you do is no good; you just spend all your money. Everything is just like an experiment, but it's a very time-boxed and limited experiment because it’s just not a sustainable growth, and no one's going to want to fund you unless they see you an extremely clear path ahead that you’re going to make money.

People often have this misconception that, "Oh, I just grow with paid growth, and then in the future, I'll come up with how I'm going to make money." I don't believe that's going to work, and I think that's sort of a failure mode for startups these days.

Next up, Michelle from Critique Match asks, "Is it good to focus on growth as much as possible before implementing a monetization model that might hurt growth?"

A monetization model that hurts growth doesn’t seem like a very good model to begin with. Obviously, you are going to want to align the way that people get value from your product and the way that people pay for your product. If you're paying for, say, unlocking features, then that’s not the best model. It’d be better if you pay for something that actually is very aligned with repeat usage or retention of your product.

What are some examples? Like companies that we've seen recently that have implemented paid monetization still in a company that needs to grow and it's not at scale? Yes, I think the truth here is like let’s take The Athletic as an example. The Athletic is a YC funded company; they’re doing really well. They are building an app for local news, and the app has a subscription feature where you really can't use the app and read the news from 150 journalists in the US unless you pay them every month.

The truth for mobile apps, and for many other companies that are trying to grow, is that paid growth is becoming a more and more important part of the growth mix every year. Back in the day, people would look down on paid growth and say that’s a negative way of growing or that’s not a good way of growing. They would usually say that because it’s sort of like they think that the product is not good enough, yet the truth is that if you look at the two platforms through which most of our attention is—Google and Facebook—they have shifted their tactics dramatically in the last five years.

It was very hard to get free growth, and paid on Facebook, if you look at searching Google, the first four or five listings are going to be paid ads. So what those pathways are doing is they’re trying to monetize your usage, which means, as a founder of a new company, you should lean into that and try to figure out how can I grow sustainably?

The problem with doing that is sort of like you’re paying for every user that you get to these two platforms, but the reality is some of the platforms that we know about—a big portion of their growth is coming from these two platforms. Instagram, Facebook, Google are a core component of growth, and if you look at something like Airbnb, search marketing through Google was always very critical to our growth strategy and is extremely core to many other travel companies.

So increasing paid links—that’s what they're doing every year. To grow organically on Google is getting more difficult as well. So back to the question: the monetization model that you implement can’t really hurt your growth, and it should be there early so you can optimize it. Otherwise, it's going to be difficult to say it’s going to be difficult to grow organically from the beginning.

You can do that to a certain level but I would argue that there are certainly a set of ways to do paid growth sustainably, but that it requires you to make money. I’ll kind of synthesize a couple of things that Gustav said here. By putting in a paid model early, you're beginning to get knowledge about what your conversion funnel costs. So if you know how much money you're going to make per user on your site, that gives you the ability to experiment more effectively with paid marketing because you know how much money you're going to earn on the other side.

If you don't have that, then, as Gustav said, you're just running these experiments that don't have viable outcomes—you can't actually benefit. Once you learn how much it costs—like The Athletic, as an example. I don't know how much to charge per month. Let’s say they charge $15 a month for their subscription for local sports.

If you live in Toronto and follow the Toronto Maple Leafs, you might be targeted with an ad on Facebook saying, "Download The Athletic app and get local news from independent journalists on their app for $15 a month," because they know what the conversion rate is for that click that comes from Facebook, and they know how much each click is roughly worth. They can actually scale up paid marketing and do it pretty effectively. Just an example, but this is how many companies today are growing.

This ties in pretty well to the next question: Suren from Ivanka asks, "What tools or software have contributed to most of the success as a product growth leader? What are the CRMs that you use?"

Lead automation? Every company and every category of companies have a different set of tools. If you're a consumer company or a SaaS company or a hardware company, you're going to use different tools to measure the product. I would say at the very basics, you would want to know what's going on. And knowing what's going on for most companies means having some kind of event analytics—like Amplitude, or Segment, or some combination of those, or Heap. Those are all good tools that you can use for the very basic stuff, in the sense of how many people are coming in, what are they doing when they come in, how are they flowing through my product.

If there are core flows in your product—like an example of The Athletic—what is the onboarding or sign-up flow? What is the payment conversion flow? What is the returning user metrics? Those tools will solve all that for you. At the minimum, those are the things that you need. Google Analytics is typically not enough for most companies. It will be good for Google-specific stuff. But I would recommend something like Amplitude or Mixpanel, and you can use Segment too to integrate those.

That’s kind of like a baseline. It's hard to find a company that shouldn't have what they're building, SaaS, whether you're building it for enterprise or even software that you're individually onboarding. The first few users, you do want to have that baseline analytics that you can always reference.

Then, depending on this, and this goes from day one—you be this, you're put in from day one—then depending on the stage of the company, you use different tools. You're not going to do automated sales tools until you reach the stage where you can automate sales. In the beginning, most of the time, you're going to be doing things that don’t scale.

And I recommend you to read that Paul Graham article—"How to Do Things That Don’t Scale." You’re not going to solve your sales problem from your computer, necessarily, if that makes sense. You're trying to find a customer and fail customers. But over time, as you're building things out, you can start to sort of build—use tools to build these for P, to build things that you do.

But most people that are asking for these tools in the very beginning probably are not thinking about it right. They should probably just go out of the office and meet their first 50 or 100 customers and get to know them more than trying to interact with them through a tool.

Here's an example: I was talking to a founder of a YC company the other day, and they were asking about how to automate an email system. I was like, "How many emails are you sending out, or who’s the target for these emails?" And they were like, "Oh, well it’s actually for other YC founders." I said, "Whoa! That is totally not the right direction to go at the beginning when you’re forming. Like an individual, when you’re selling, just send the email from your own account."

Think about it this way: you get a lot of email; the ones that you respond to are probably from your friends or family or other co-workers, and they look like real emails, just text. It turns out that those actually have a much higher conversion rate, especially in the early days as you're forming these one-on-one connections.

Another good example is customer service. So anyone who's trying to build out a Zendesk or some kind of forum from the beginning isn’t attempting to learn enough from the users. So I would ask the founder of a company: "I would want to receive all the emails from all my customers directed into my email inbox, and I want to reply with my name so that they feel surprised that they're hearing from the founder of the company."

There’s a great story from Airbnb here—Joe and Brian actually shared their cell phone numbers with what they shared with the first couple hundred hosts, and they've kept calling them for years. Joe didn't change his number, so his number’s out there, and that was not a negative; that’s a really positive thing. He learned so much and engaged with these hosts over those years.

So as a first-time founder, like as a founder getting started, you don’t really want a lot of systems aside from tracking how your activity on the product is happening to you. You want to just do it directly.

So that leads into the next question quite well. I come from Haystack, asks: "Customers often have a pain, but they don't know how to solve it. This results in pretty vague feedback. How do you take feedback and iterate on a product?"

A common way to do user research is to ask extremely open-ended questions and not interrupt them too much, and not make a lot of assumptions. One thing we used to do whenever we were launching a new product, so we would take that product, either on a laptop or on a phone, and we would just walk down in our office and we would show the screens to random people that sat down there and be like, "Just tell me what you think this is."

Wait—your colleagues?

No, just two random people—like in the lobby?

Or in the lobby! Our product wasn't meant to be for the mainstream, so every being coming into the—you know everyone. Then we’d just slide through the screenshots of the things we were trying to build, and we'd spend time asking, "What do you think this means?"

If they didn't get it—if you ask 10 people and most didn’t get it—it wasn’t good. I mean, you're not going to be there looking over the shoulder with any of the customers. They're just going to be there on their own.

So if most people out of a group of 10 random people aren’t getting it, it’s not good enough; it's not obvious enough how the product works.

The other thing that we did—there’s two other things—we sometimes would use something called usertesting.com. I don't know, actually, not sure if I'll recommend them anymore, but at the time, they were good enough for the problem we were solving. You do the same thing. We would ask users to do something very simple in the Airbnb app, and then you would have a random person who would get paid to do that task and then record themselves doing it.

They'd speak up and say what they’re doing, and it’s like this. Then you move a little. I watched ten of those videos and people were pulling their hair out—the designers and the product and engineers were pulling their hair out because they couldn't believe what people couldn’t get about what they were doing.

But the truth is that most people don’t understand most products. If you come in with assumptions in these meetings, it's not going to work.

So you have to let it out there and learn from that. To play on this, one of the things that I've found is like we always talk about data; we talk about this like, you know, track—get your analytics set up and track all this data.

But oftentimes, like a founder is confronted with like, "Well, what do I do? I have all this data." So I think what you mentioned is actually one way to cut through that. So if the data shows that you're having this problem or if you’re having this conversion—one of the ways you can get like a solid answer is to spin up one of these user testing sessions—like 10 minutes, 30 minutes—and ask the person, just ask five people to go through the pain point.

My hunch is that most of, like most, you don't have to ask a thousand people to get that. It’s like you used to say ten people was good enough to get guidance on whether it’s good or bad.

Because they could be larger or smaller.

Right! They can be your big enough sample or they can be your site and can be a friend.

I think the key thing is do not give guidance—don't tell them what you're supposed to do. Just tell them either go, just ask them to speak up when they're going through the flow.

From my experience, that makes you want to standardize things. That makes you want to use extremely clear language. People try to be smart about language in their app, and clever about how to market the product. Those things tend to go extremely wrong and what you really want to do is just use kind of user interface language like, "What happens here? Click on this button. What happens here?"

That’s my experience of how you make something people actually understand how to use.

The next one is a question that a bunch of you asked: "How do you grow a multi-sided marketplace and retain users on that marketplace?"

I think there are many things in this question. I get this question a lot, so I can talk about it. I'm going to talk about exactly how I would kickstart a new marketplace today, and then why people I think ask this question might not be exactly what I'm going to talk about.

So look, if I was trying to build a marketplace today, I would try to build up the supply first. I would build up the supply, sort of artificially, to a level—a small number of supply—artificially to a level where it would reflect what the idea is that I have it in the product.

Let’s take Uber or Lyft as an example. If I would launch Uber or Lyft today in a new city, I would find 10 drivers—could be friends—and I would make sure that they are driving, and that committing to drive for the next four or five days.

I would go out and get their friends. They’re just going to drive, and by committing to drive for all that time, I basically have now artificially created this like 10 day or something of 10 drivers who is going to be on the streets. Then I'll go and get the demand afterwards.

We actually have an example from one of the companies—it’s called Applause. It’s an online marketplace that lets founders test out their experiments with their target audiences to validate their assumptions, which is almost exactly like we are talking about.

So the person is asking: "How do we get this marketplace going?"

Yeah, so if I would build that idea, I would get 10 really good people—there are my friends—maybe you have to pay them, maybe you don't. Maybe they're willing to sit through X number of these sessions.

I mean you must have a lot of friends who are founders as well, right? Yeah, get them to volunteer a little bit of time, but don’t go out and get strangers for the supply. Yeah, like validate that there’s the demand before you go and figure out how to get the supply.

And then have them subdue it, and see what the demand feedback from the demand. You've got this like early supply, so you've got your friends all volunteering their time on this market.

Let's talk about like how do you—I mean the level of quality. So in the earlier stages, we allowed the supply like listing to sign up without a user photo, without a photo of the house. Like how crazy is that?

Like that was the dumbest idea ever, but it took us a while to figure out that was a really dumb idea. What you really want is like quality—you can’t list unless you have quality. You can’t be an Uber driver unless you have a nice car, or you're a safe driver.

And I think this is one of the things that you can do now when you—before you have scale—so you can actually curate this initial list, like your 10 friends. You don't make them build the profile; let them build the profile for them, post it—just do everything.

And once you demand—it’s applicable in this case of Uber, or in this case of user testing.com. If you got unto people on the demand side, you do it again, like if I use Uber and had a good experience, chances are I do it again in the next 10 days.

And if I had a good experience, and it’s up to you to define what that good experience is for your supply—if I do it again, that’s a really good sign. The people are voluntarily coming back and either doing this experimentation side or the Uber a second time—it’s like on this experimentation question.

So you're building a marketplace—you’ve got your 10 or 15 friends that you’re kind of coercing into being those early people, and now you’re going out onto the demand side and you’re admitting it. Sorry, yeah, the demand side exactly.

So like you’ve got the first people that are coming in the door saying like, "Hey, could you test out this feature?" How do you retain them? Like how do you get them to come back?

I would always ask people that come in like, "What was the experience like? Can you review the site?" Or just like get some quality feedback from these people—whether they had a good experience comes from—it's sometimes hard to get quality feedback from people who had a bad experience because they tend to like just not care.

And sending surveys to people who have retained, or those churned users, is always tricky, but it’s very, very important because you want to understand why people didn’t find that good experience.

And if they didn’t have a good experience, you need to work with demand and try to understand what about the idea you had in your head for the supply is not correct. It’s like debugging the marketplace.

Yes! One of the things that I've seen founders benefit a lot from is asking for phone numbers during the signup flow, especially during the early days. So that if something does go wrong or if you want to add, they actually have a great experience. You can physically like, you could phone them up and get an answer.

Like emails, sometimes it’s easier to push off; it’s easy to blow that off for a weekend. You might be in the dark, but this person had a great experience and I can't get them on the phone. I can't get the word back.

Once you’ve gone through this phase and now demand really wants what you’re offering, you have to figure out the price—literally what is the demand willing to pay for that quality supply?

That is not straightforward either. Like in the case of this experiment, like getting friends to do reviewing experiments or ideas, the friends might not be willing to do that at the price that I'm willing to pay.

So as you suggested, it costs me $45 per test. I'm going to assume that for $20 maybe it would have been the ways to that person that was reviewing me. I think some people want to do that; I don’t see how everyone is going to do that for $20.

You know? So I think this really depends on what you’re really trying to figure out. Most of us would rent out our home for some price; maybe we're not all going to do it for the current market price.

So that’s why I mean not everybody in the world is going to be hosts, for example. Your pricing is part of your product-market fit journey, and you’ll be experimenting and adjusting that as you tune it.

Next up, Daniel from Storylo asks a pretty tactical question: "They make custom photography easy and affordable." So he’s asking, "Can you outline the pros and cons of incentivized growth examples like Airbnb's invite-a-friend for $25 or Uber’s subsidized rides?"

Great! So I spent a lot of time on this, and I worked on the referral program for most of the time I was there. It grew a significant portion of the growth, and from my experience, referral programs have a couple of criteria to make them work well.

The first one: there are products that already have word-of-mouth; they already have people telling their friends about it, so they’re not reliant upon that in essence. They also sometimes require a friend. The product is radical enough that their friend can really help convince them about using the product.

So great examples here are Uber or Airbnb, and some of the things that are just new that I might not trust without a friend telling me about it.

So you’re saying like the products where having another friend recommend it actually makes the new customer more likely to buy or that the product gets better if your friends are on it as well?

Yes! Maybe I wouldn’t trust the product; maybe I wouldn’t trust the idea unless a friend recommended this to me. So I wouldn’t trust, and that’s why you might put in a referral program.

Think of this like the current, the Airbnb today and the Uber of today are very different than the first month of those two products. Most people would be scared of those two ideas the first time they heard about it.

It's only until the minute it gets into the mainstream where you now get comfortable for everyone to do it. In those early days, if the idea was radical, having a friend convince you that that is not scary is a really good fundamental foundation for a good referral program because referrals optimally push them over the edge a little bit to do it.

And that pushes the person they receive the referral over the edge to actually provide the product, and it does a couple percentage points on each side, and that could really matter.

What’s interesting is that you haven't mentioned the actual savings or the percent off. Like that is almost not part of it at all.

So that the money is important, but it’s not just added to any product. I think it does matter that the product kind of requires this in some sense.

The second thing is that the plugs to have a regularly active user base—what I mean by that is probably only used once in your lifetime. Referrals are less likely to work well because you only have one chance to get someone to tell their friends.

Versus every time I use Lyft or Uber, I get reminded that I could give my friends $25.

Yes! Is actually there is some conversion every single time I do that. And for Airbnb, for example, after you made a booking, we had two left over; we would ask someone to invite a friend Kevin B, and I think we had a 20% click-through rate on both those places.

So 20% of everyone who made a review or made a booking actually went into the referral flow, and half of them I think ended up sending invites. So that was pretty effective, but people weren’t coming back to Airbnb regularly, in a way that they were willing to be interrupted.

You only do an Airbnb if you’re traveling; you only do it once a year or once sponsors—people travel here and there.

The reason I mention this is that most people don’t think about this—where do I get notified, or kind of interrupted about the referral program? They just put like, "I’ll just put it in the menu." "I'll just put it in their settings menu."

But they’re not coming back to Airbnb. People don’t like that demographic that you want to track to figure this out is percent of weekly active users that see the link for a referral program.

If you do track that metric and you look at the people who do your settings page, no one’s going to go to your settings page if you put it in the menu!

People are typically in an interaction of doing something else, so you can't put it in like in the workflow that the users are going through because they don’t want to be interrupted when they’re doing what they’re doing.

So that’s the first couple things. The second thing is that referrals is paid marketing in the sense that you need to be able to make the money back to be giving away.

A lot of people get excited about this idea of referrals, but they're not making enough money on the backend to actually make it work. So they’re like, "Maybe they make $10 per user." The truth is that most Americans don’t care to take any action for $5 because it’s $5 on each side.

People just don’t care about it, so then you need to have some minimum number, and I think that like $15 or more. You need to be able to give away; otherwise you just aren't going to care, and it’s not going to work well.

I think right now Airbnb's program is $55.

Yes, here in the US, they also do some of this for the sender. Those are meaningful numbers, and that can be 10% or 50% of your trip as you travel for the first time, so it can be really, really impactful.

We saw Airbnb’s programs that this doesn't matter.

Next question is from Justin at Louda Looks around. Louda Looks provides online tools to home-based beauty and grooming professionals to help them grow and manage their clientele. All of their customer acquisition right now comes from outreach, cold emails, DMs on Instagram, and it’s working! They don’t feel like it’s scalable, especially if they want to grow faster.

At what point do they need to be honest with themselves and figure out what the scalable growth channel is here, and how long can they stay on this kind of direct method?

Great question! When you switch to scaling growth channels, the truth is there are only a couple of really scaling growth channels in the world: paid marketing, different types of marketing, SEO, word of mouth, sales, and viral growth.

And maybe one or more—there aren’t that many others. So first, think of like which of these channels actually fit my product well. In the talk that I gave, there’s on YouTube, I have a slide that I borrowed from another person that is sort of like a court here on how you figure out which of these categories my part fits into.

What you're describing here is probably a product that, on the demand side, fits well into marketing because of the level of targeting that you can do. It seems like Louda Looks has revenue, and that revenue can be used to pay for that paid marketing.

The challenge with paid marketing is that you don’t want to reacquire users. So in the case of Airbnb, for example—we wouldn’t want to go and pay for you to click on our Google ad every time you make a booking. We only want to do that the first time.

After that, we want you to install the Airbnb app, and next time you do a search for travel, you do it in our app. Most companies you think about this that way—the first time you might be acquiring users through paid marketing, but you don’t want to do that concurrently because you’re building adds a lot of risk to your growth strategy.

Much better if you would have some natural entry point that users are going to open—that could be an app, it could be some email, or just some way that I'm not going to have to pay to reacquire that user. So in this case, I think marketing is probably the most likely place; referrals might be the second most likely.

The good news is if you are earning money, you can begin to like experiment with those because you know how much customer is worth after you close them.

Cool! Next up is your experience. This is a question from Malo. In your experience, especially at Airbnb, what are some key product sales activities that you did to reduce lead time, and to what extent did lead time matter?

I think lead time would be like how long it takes to close a sale. In Airbnb's case, this is not super applicable, except for on the consumer side. The truth is that we are all, as consumers, pretty susceptible to things like exclusivity, scarcity, or urgency messages like that.

Now, in some cases, those are dark patterns, and what I mean by that is they're just messing with our minds but actually not really providing any value.

The urgency thing—yeah, yeah. But what we did at Airbnb is that the staff has information. Let’s say you do a search for San Francisco, and there’s only 40% of the listings left. We should tell you that, because that gives you a real sense of urgency that you need to book soon.

Because if you are booking some of the last 10%, then you might not get a good deal. It might be a pretty bad deal because all the good ones are already booked by everyone else.

Everyone is rational. So our view was to surface these tactics: urgency, scarcity, exclusivity, sort of like in an informal way, and not in a way that kind of like overpowering, overbearing, and mess with your head.

Another example is what we call a unique find or rare find. If you found a listing, we would literally tell you, "This listing is usually—it's rare find." And that’s just a positive signal about the listing instead of saying like it’s sold out or something.

Yeah, something like that; that’s really—I like that.

I've seen the verify of the rare find, and the truth is that those work extremely well because, as users, we react well to those messages. That’s the reality. Would I prefer that didn’t work well?

Well, I think you just need to think hard about what is the perception of the tactics that you use as you’re trying to get people to do what you're doing. And all commerce—it's for consumers, some level of these tactics but you need to just make sure that they are user-friendly and they’re not just messing with the psychology of that.

Actually kind of giving you some real value.

Yeah, we got a question from the YouTube comments. Mo asked, "How would you implement a referral program in B2B, where businesses can refer another business for a bonus?"

This is a good question. It’s complicated. A couple of cents—first would be to be an example. If I work for a SaaS service and I refer dollars for another customer through that service, the thing that they think about first is, "What is the likelihood that I know other people in my role that would buy this service?"

So let’s say the SaaS service is a payroll software. What’s the likelihood that I know other people in HR and other companies that would buy?

So the payroll software in the founder community is probably fairly true, but in the larger mainstream world, we’re less likely.

Now, there are exceptions, and there are certainly cases where this could work. There are some legal implications, like you can’t give someone $1,000 to then put it in your pocket and then convince the company my company to start using software.

Then I put the money in the pocket; I think that’s basically bribery.

There are legal problems with doing this. I don't know exactly where the line is drawn there.

I think you should—well, that only really said like Gusto has a thing. That Gusto is payroll software; they have a very prominent refer someone else and you get $500 cash.

Yeah, I think the case with this could work really well is, for example, owner-owned businesses.

For small businesses, it’s very often that person therefore is also the owner of the business. I don’t think there’s an illegal problem if I’m where that would make a meaningful amount like $500 to a beauty salon that’s wrecking ball business, and I refer you to payroll software and you get a few more dollars off your payroll software.

Yeah, you’re the owner of the business that makes a ton of sense because it goes back into the—

Yes! I think for small businesses, this works really well. For larger ones, I think this gets really complicated, but I think for small businesses, it’s not a bad idea at all.

But I think the other thing that I would say is they're like helping people. There are ways that you can be smart about who might like helping people. For example, you can use Google Apps, and they go with Google contacts to fan people IDs in people's heads of who might need the software.

So I think probably too long going into right now, but I think there are things you can do to just plant the idea in the head of like, "Who else do you know that runs payroll, or who else do you know that needs this type of specific SaaS offer?"

People always do! You can do that during your customer interviews because as you’re talking to them or when you’re onboarding them, you can always—if they express interest, if they’re excited about the product, you can always see: "Do you think your friends might be interested? Who are they? Tell me about their businesses."

Sometimes their roles are more fluid, so examples of recruiting. People are more likely to move between companies, and recruiting roles, in my experience, are very good at sharing software and tools between having different companies.

I've got a question from Erica at Kiefer Labs: "How do you reach out and educate people on the amazing benefits of your product and the utmost solution it provides if you have an unfamiliar product and name, like probiotic kefir? How do you break the wall and scale?"

It seems like a general question. What do you think, Eric?

I think that, I mean, I would take this question about like branding. How do you, as a startup founder? Like there are two types of products; there’s one where you need to educate the market and teach people about like this new product or new category.

Then there’s the other one where you're building a competing product to an existing incumbent, and you have to price it or market it accordingly. I would say that this question is more about how do you build enthusiasm for something that’s completely new?

I think it's different in consumer versus B2B. I think for B2B, what I would start with is two things. I would first show them the product. I've learned this in working with a bunch of YC companies: it can really help just including a screenshot or a video of what people are trying to explain this new idea to someone.

So for an existing public, it's not a new problem. Typically, it can be much harder than to just show them the product.

So we found that in B2B SaaS companies, it's like incorporating screenshots or a video of how it works, which is tough to do because oftentimes when people include screenshots, they’re like, "I’m going to put in the entire product."

You need to have like screenshots; you need to be able to get specific things that people will recognize and be like, "Aha! That’s the thing that I can see myself using!"

Yes! So like I can’t think of a million things here. The second thing on the B2B side is just include social proof. A lot of people are afraid of trying new things unless there's someone that they trust who has tried it.

So like would you try this new payroll software? No. Would you try the new pair of software that Airbnb and Dropbox are using? Yes!

To tie this back to your point about referrals, this is where the referral program kicks in. Because if you're selling something that's kind of new, but you have a trusted friend who's the one introducing the product, you will talk about the soil. For example, like I think you're selling a kefir, I think it's like yogurt, right?

So Soylent was a product that was pretty new—like people eat food, but they definitely didn’t need Soylent. So at the beginning, they had this entire community where people participated in this forum, and they would talk about it.

I’m sure those people were the first ones telling other new users that they should try this thing called Soylent!

I think consumer products typically—if they are new enough, they tend to have word-of-mouth built into them. If they don’t, that’s the problem with categories.

But this is difficult. I think it's a very private product; it’s like private products that are brand new. They can't be discussed easily, like hair loss products—in those things, you're typically not broadcasting it.

That’s interesting, they flipped it on its head. They actually like brought it into conversation a little bit, so that might have been a smart move.

Yeah! And then there was one more point on that question. The second thing is like I would, for this question, try to get some brand association in the sense of try to get written up in some major papers, or like some authoritative media that you know about, and then include that.

So like those two things—like if there's a new product, I never heard of it, but if you have press written about it, I’m certainly gonna increase my likelihood of trying it.

They don't have to write about your specific product; they just have to write about the category. If you can attach yourself to a category that's on the rise, that would help.

I wouldn’t go extremely far to get this down, but I would say that for something brand new—having some level of social proof from some you know helps.

Next question: "How do you balance growth against the backlog of cosmetic quality of life bugs and tickets?"

It's a false question! As a startup founder, you always have to do everything, and then it’s really hard to say, "Oh well, I see you allowed me to work on growth, so I don’t have to work on quality!"

This is false. Like that is not sort of like how you think about these things. There’s always tension, but there’s never going to be a black-and-white situation. They are tightly coupled.

If you have bad quality of the product, you're not going to get growth, and if you have lots of bug buds and tickets, people are going to churn, and you're not going to get growth because having retained users is typically what gives you growth.

I think the most important thing that I would advise this founder is to truly try to understand what drives growth and what drives retention, or why, what drives churn, like what makes people unhappy with these new products.

What makes people happy to use your product? It’ll give you an idea of what people actually care about in your product, and from that point, this question gets easier to answer.

Another question from the YouTube stream: "Would you recommend SEO for early-stage startups?"

SEO is interesting because SEO is sort of like still a very big channel because most people in the world still use Google to answer most questions that are not daily questions.

The questions that we ask will change every year. This year we might ask the question like, "Who is..." and then like a politician name that we didn’t ask a year ago.

The search volume and then the keyword star that goes into Google actually change quite a bit—or what is Soylent!

You might have not asked that five years ago. So from that perspective, if you can capture the keywords of something new, it’s possible that this can work really well for you.

If the question is, "What should I do in Paris?" you’re not going to be able to rank for that most likely because there are five other companies that have a lot of companies and a hundred people teams working that are trying to rank for that specific question.

So think about the question of trying to rank for and how new they are. The bigger the chance you can actually succeed. If these are established questions that people have been asking for years—like what is a good mortgage—you’re not going to be able to rank for that because it's so competitive.

You’re talking years of effort that you might not even control to be able to do that. So we met first question like—like is this something, does it make sense to think about it around the idea of a question that people are asking?

Yes! It gives me any kind of keywords, but like the words that people put into Google are the ones that you should think through. The second thing is there are two ways to really grow SEO: one is to write a lot of good high-quality content that gets ranked.

The second thing is to do something that's automated where you have a lot of content that sort of like automatically created.

Most are more on the latter side. Airbnb would have lots of landing pages for different cities, where there are different listings.

Or even just different listings, some of them sell for waste. They would organize those listings. That's more automatically created versus many startups in the other day just would write really good content.

So maybe in the case of hair loss, you would literally write super high-quality unique content about that, those questions, and you would just always start with the keywords itself.

So before you try to figure it out, like in this past, is it a good idea? You want to go to Google; you want to use a tool called the Keyword Planner and search for those keywords and see how much traffic they get.

If they get a lot of interest in social media marketing channels these days, they seem to keep changing—Facebook, Instagram, Snapchat, TikTok—where would you put your focus?

This is probably where I'm the least expert, and maybe one of the reasonably exclusive because they do change so often, so it’s really hard to keep track.

On one end, I would say quite frankly when I run into companies who are successful at getting, say, influencer marketing or some Snapchat filter or something in that world working, I'm very rarely running to companies who said, "We made that work, and now we have a lot of known users," because these tactics tend to be quickly saturated.

The companies that you're building on top tend to not be in favor of giving with free traffic; they want to charge for that traffic. So they don’t really typically allow you to do some hack for too long, and then they tend to remove it.

If you compare, for example, Instagram with previous platforms, Instagram doesn’t really allow links. Why? Because they don’t want spam, and they don’t want people to sort of hijack the app for a bunch of commercial messages.

They want the commercial messages as they exist inside ads that just pay them money. So the one channel in social media marketing that does work is to buy ads from them. That is a truly scalable channel.

And TikTok, for example, I believe was one of the biggest advertisers on Facebook and Instagram the year before—like 2018 or 2019—but they literally just bought ads and spent a billion, and then other people will discover it.

They’ll come in, either at people who discovered or the platform itself will make that getting harder.

Next question is from Jose at Arcade: "Quickly, how would you measure retention for a service that could take a long time? If you have dropouts, is an indication that people don’t like the product?"

So I would start by having probably two metrics: the first metric would be dropouts, and the second metric would be some metrics that predicts dropouts.

So my guess is that the people that would drop out—whether people that didn’t take the course anymore. If you find that there’s a strong correlation between people who stop taking your course and then drop out, then you can actually start measuring that first metric.

Like are people stopping taking the course? Because most likely it will lead to a dropout. So I think I would just sort of like to try to figure out what’s the predictiveness of someone stopping using a service.

And everybody, for example, if someone left a one-star review, you're highly predictive of someone not looking again.

So we should definitely try to avoid people having a one-star experience, or people never searching on the site for another year after their first booking is also quite predictive of them not looking again.

So there are these kinds of predictive metrics, so to speak, that can predict either the next big good behavior or bad for you.

So that would be my recommendation: try to figure out what predicts people drop out.

Michael from Hi Low asks: "Hi Low is building a platform where users share the high and low points of their workday anonymously with their team leaders. He asks, should you say yes to everything?"

For example, we had a company of 100 and 150 employees approach them and want to implement us even at MVP stage. He thinks that if he says yes, he has no idea what they're going to do in terms of actually handling that.

So if you have demand for your product, it’s a great time. I think just ask for a price and see what they’re willing to pay.

If they’re willing to pay $50 a month per employee, or $20 a month per employee, he’s like, "Sure!"

Because if you don’t do that, it sounds like a great first customer. If they’re not willing to pay, now they might not be willing to actually use it. They might not care as much about what you're building, and this idea sounds new and interesting, but new enough that I suspect a lot of people hear about this idea and would say, "That sounds interesting. I think I want to try it!"

But they didn’t say, "I would try that for $50 a month for one of your contract!"

Please be surprised to say that right away.

So in this scenario, I think you need to figure out: what is their value? This might take time to figure out what the value is you can charge for your product.

But the good news is it sounds like people are excited and they’re excited to try even in an MVP. You really want to get that like the danger here would be taking this as positive feedback and then going and continuing to build your product without interacting with the customer.

The right move, I think, is what Gustav said is to go back to them and say: "Would you actually like to be a paid customer of this early MVP product?" If they say yes, then I mean you just got your first customer.

If they say no, you'll figure out what is the Delta between what your MVP is and what they would be willing to pay you for.

Oh, sorry! Oh!

There we go, okay. This is a question from Twitter: "Where to Turi! Airbnb, for luggage sharing! The concept of luggage sharing is quite novel in India, and the market is hugely untapped. How do we scale our product or concept to a larger audience and with investors?"

That's a great question! I actually used my first luggage sharing app last year.

No way! What was your use case?

My use case was I just checked out my Airbnb, and we were staying in Harlem, and we were going to Penn Station to take the train to D.C., and I didn’t want to carry my bag around for a couple of hours.

Yeah, that’s natural! Hotels will do this for you; everybody doesn’t do this for you if you have a checkout at 11 a.m.

So the rest of the day, you kind of have to carry your luggage around. So there's a clear need!

Now, there isn’t actually a great product for this. There are a bunch of products, and I tried a couple of them, but there’s no one that is exactly what I want.

So what I would do here is I would build an app that allows you to—once someone finds you through Google, which is the most likely place you'll find this—install the app, and keep the app on the phone.

The app is going to be the thing that I remember next time I need this. I just go to the app, and it automatically opens the app in the map to see whatever I have in the area where I type in an address.

The second thing is you want to make sure you have some level of trust.

So, one thing experienced storing my luggage actually was just—my luggage in a restaurant sort of at the side, and it felt like they weren’t looking around. Some of them come in to grab my luggage, and I didn’t feel safe that was going to be my computer.

The second thing is you really need to build in some level of guidance to the stores and some level of guarantee that you can offer that makes me trust that my luggage is safe.

This? The third thing is, you don’t need to figure out how can I put this in front of people every single time they check out.

There are baby—you're going to try to talk to everybody, but this is development team. It will be hard if you're a small startup; they might not care about you unless you have thousands of locations around the world.

Well, this is possible, but it would be a pretty good idea to integrate that. After you made a booking, just like you have rental cars, it's time to hotel booking, etc.

But you could also do that without actually talking to Airbnb; you could talk to Airbnb hosts directly. So every Airbnb host sends a little pre-formatted email with every—every time someone checks in.

It has like the address of how to get there and how to get there from the airport. You could maybe convince some of them to include it in there! And then absolutely, that's a good idea.

The cool thing with this idea is you can start just like everybody starts: you start in the big cities!

So there’s a lot of like—do a lot of New York, London, like all the places where people are a lot of travelers.

Wherever you definitely need to consider your luggage, and I think again, think of the cities where you can walk a lot; probably LA wouldn’t be the best place.

Because I can bring—I'm not going to walk around the LA as much as I will do somewhere else. Start with cities where you need to walk a lot and get a few places that you can sort of luggage that are all around the city, but specifically where people are going to go next.

So that's where probably I'm going to look to pick up my luggage—like at the airport or the train station or whatever. That’s where people are going next—that’s where they probably want to pick up my luggage.

There’s a kind of innovation I haven’t seen yet here, sort of like I open the app, I push a button, and within 10 minutes someone comes to pick up luggage!

That is the information I was hoping for! I would have paid like $15 or $20 just to have someone get my luggage.

Which makes sense, like you could piece that together using on-demand—like Uber. You order an Uber and take my luggage.

You heard it here first! Good! And then drive my luggage to where I’m going to pick it up.

I would pay $20! You know, I don’t have to drag my luggage around all day, but no one is offering that!

That does not exist as far as I know but I think there’s certainly an opportunity for someone to build this company.

You just have to figure out how much people are going to pay, but I believe people probably would pay $20 or $30 just to do this for over a day.

Next question! We’ll get through a couple more. Joseph from Metropush: Metropush is an application that sends push notifications from public service broadcasters to their customers. They decided to offer their product with one year free before an ad-free subscription, and they think that's a pretty good deal.

Are they hurting themselves by offering a free trial and only relying on ads for revenue?

Yes! I don’t think you need to offer your product for a year for free. I think you can offer— the longest that would make sense to me would be two weeks or a month.

That would be the longest type of because what are you really looking for? Is someone to start using your product, build some basic, basically onboard themselves and try to figure out what the progress is, and see if they find it valuable.

Maybe they again take the Athletic. I assigned it to The Athletic. I’ve read four articles about Toronto Maple Leafs, and I'm like, "Wow, this is a great app!"

That's all you need to know about how the app works! You don’t need to use it for free for another month. I think at that point you want to start—you want to get someone to pay.

So I think it’s probably a mistake to do it for a whole year. The revenue you’ll make from advertising versus revenue you make from subscribers is going to be, at whatever, like 10 to 20 times the difference.

So you're going to make a lot more money from subscribers even though they’re going to be fewer subscribers. Yet advertisers is just very difficult in a generic ad-based platform.

They’re using some kind of ad network today and making a lot of money—because the CPM cut, the earnings are just not there. You’re going to make from a very active product somewhere between zero and fifty cents a month at the most.

Probably less than that. Yeah, for sure! Most likely you’re talking about a couple of cents to ten cents per user per month, which is just not worth it.

If that’s like $1 a year, just can’t—nothing you can acquire new customers for that. You can’t service them. You can't buy content.

It’s a basically having engineering ad networks today doesn’t build a sustainable company. And I don’t think investors would only fund you if that's what you’re serious about to monetize upon.

Good news is that users, consumers are getting more comfortable with microtransactions—paying for subscriptions.

Like, for example, we have a company called Substack that went through YC about a year ago. They enable anyone to start a female newsletter that you can actually charge money for.

I think this was somewhat of a revolutionary concept a year ago—like who would pay for an email subscription? It turns out a lot of people are because they get reliable content on a regular basis from a high-quality writer, and they feel good because even the person that’s writing it is earning a good wage, and it fundamentally solves a societal problem.

Now the business model is in line with the actual users of the product, and they’re not in conflict, which always—I mean, not just Facebook, but newspapers have some of those challenges.

Absolutely! Next up is from Marina from Growth Channel: "We want to hear more about user-generated growth strategies that can be applied to B2B SaaS companies."

Okay, so the way that you would win as a B2B company today is to make a new product that may solve an existing need and then just apply them—like the very best consumer grade growth skills to that product—because you’re competing with others that aren’t going to do that.

They’re going to do very traditional stuff: case studies, like old-school marketing interests are going to big events. They’re not going to do invite flows; they’re not going to do super-targeted ads on Facebook; they're not going to do a lot of things that they're not a great referral program, and they’re not going to build these things into the product.

So, if I would try to grow inside a company, I’d connect with the Google. I would try to figure out if I can get a hold of the emails—I would just pull some of the employees; I would pick out the emails from the co-workers, like based on the domain.

So maybe like mycompany.com would be the domain that would look like audiences from within the company.

Okay, inside a company, how do you grow? I think it’s a little trickier. Let’s see. For example, the idea that we had: questions should they look at over—yeah, let’s lead rallies!

Yes! There’s certainly going to be intercompany virality.

Let’s say collaboration on documents—they're also rallying that goes to the subset of that—so some documents are by nature or something you collaborate with the broader world with.

So if I would say Airtable, I would try to figure out what are the most viral documents that exist today in the world.

For example, the Burning Man packing list is one spreadsheet that I remember being very viral. It was like 10,000 people and tens of thousands of people.

I can use that I would replicate that on Airtable, and I would think of all the other viral documents that are just getting mass distribution and see if I can do the same.

What’s the effect? That’s an enterprise SaaS product where you pay like—if you win the customer, they’re paying like $25 a month or something like that.

I would just like to make it viral! I would also make it embeddable. That would take the Airtable table and be like, "Can you embed that on your website?"

And so that everyone comes there and are trying to do something specific.

I actually see that Airtable—that's how I would reach like the broader market.

How do you reach between companies? I think it’s a little trickier because you should think hard about what do I do that relates to other employees and other companies? Let’s look at Slack.

Yeah! So like Slack has the ability for you to add an employee from another company into your Slack room without them having their own Slack, and just like you give them an account.

You want to figure out how to do that? That’s a great method! If you could do something like that, that's really good.

So I think for virality, always think about what’s the piece of content that I’m collaborating on versus like that I’m just trying to figure out a way to invite a lot of people.

I think that’s less exciting. So build it into the product rather than just the referral strategy. So you like in Google Docs—you can just like add someone, and they can give a comment in your document.

You can have anybody in your email address book. That’s a pretty viral thing.

You just have to think about what Google Docs can I do that for and then growing inside companies.

I think you really want to find a way to utilize the address book of your Google Assets—you got Google Apps account.

I think that’s going to be the core way to do it. It’s often to get their contact and to give them all accounts.

Well, you want to figure out what you should share with these other emails, and like what does an employee benefit from sharing?

That could be different things for different products. The second thing is you want to make every document very easily copyable.

There’s a YC company that got funded that creates and a link basis. You can take any link and then you can say air slash anything and then you can basically share their own the office.

That company offers that service, but that link becomes sort of like physically viral because they know that air slash something that's going around.

Yeah! There's a company that—basically, there's a physical biology to that idea inside a company, for example.

Cool! Well, that’s it for today. Thanks very much, Gustav! There’s a lot of actionable advice that we got in there, and yeah.

We’ll join you—we’ll be back next week for another live stream. Thank you so much! Thanks!