For startups to survive, having a highly technical founding team, being frugal with money, and having a strong emotional attachment to the startup's success are crucial.

For startups to survive, having a highly technical founding team, being frugal with money, and having a strong emotional attachment to the startup's success are crucial. (discussed at 0:49 in the video)

Clearly define the problem your product solves in one or two sentences, ideally having experienced it yourself, and narrow down the specific group you're solving it for first.

Clearly define the problem your product solves in one or two sentences, ideally having experienced it yourself, and narrow down the specific group you're solving it for first. (discussed at 5:48 in the video)

Understand who your customer is and how frequently and intensely they experience the problem you are solving; this helps identify if a business is viable.

Understand who your customer is and how frequently and intensely they experience the problem you are solving; this helps identify if a business is viable. (discussed at 16:51 in the video)

Founders should be willing to charge for their product from the beginning, as this helps identify users who truly have an intense problem and value the solution.

Founders should be willing to charge for their product from the beginning, as this helps identify users who truly have an intense problem and value the solution. (discussed at 25:07 in the video)

Use event-based metrics products like Mixpanel or Amplitude to track specific user actions, rather than just website traffic, to understand how your product is being used.

Use event-based metrics products like Mixpanel or Amplitude to track specific user actions, rather than just website traffic, to understand how your product is being used. (discussed at 47:22 in the video)

Start with 5-10 simple, clearly named stats to track, ensuring everyone in the company understands and can access these metrics.

Start with 5-10 simple, clearly named stats to track, ensuring everyone in the company understands and can access these metrics. (discussed at 50:22 in the video)

Integrate measurement into your product specifications from the start, rather than adding it later, to effectively track product performance.

Integrate measurement into your product specifications from the start, rather than adding it later, to effectively track product performance. (discussed at 51:55 in the video)

Adopt a structured product development cycle with short iteration periods (e.g., two weeks for apps, one week for web), a clear KPI, and detailed specs to avoid unproductive arguments and wasted effort.

Adopt a structured product development cycle with short iteration periods (e.g., two weeks for apps, one week for web), a clear KPI, and detailed specs to avoid unproductive arguments and wasted effort. (discussed at 59:17 in the video)

How to Create a Product Roadmap

0:00 without any further delay I will

0:02 introduce to you Michael Seibel the CEO

0:04 of Y Combinator the founder of companies

0:07 like justin.tv and twitch and Socialcam

0:11 to begin what is going to be a deep dive

0:14 into product over the next several

0:17 lectures

0:17 Michael so before I begin its kind of

0:31 had a conversation with Jeff and I and I

0:33 wanted to say a couple of things about

0:35 my experience at justin.tv and and

0:39 twitch so what I will say is that we

0:43 broke many if not all of the rules that

0:46 are about to tell you at various points

0:50 during our company the things that

0:55 allowed us to survive were one our

0:58 founding team was extremely technical

1:03 Justin Emmett and Kyle all were amazing

1:09 to work with and basically what I found

1:12 amazing about them is they were not

1:13 intimidated by any technical challenge I

1:17 think that I would not be standing here

1:19 if I wasn't

1:22 privileged to work with them and so I

1:24 think this is something that a lot of

1:26 companies a lot of startups a lot of

1:29 sort of founders don't truly understand

1:31 like that fact allowed us to break a lot

1:34 of rules the second is we didn't spend a

1:36 lot of money we moved out when we were

1:40 21 22 22 and 23 we lived in a

1:44 two-bedroom apartment that apartment

1:48 cost $2,500 a month we were each given

1:51 $500 a month walking around money which

1:54 technically is against the law because

1:56 it was below minimum wage but who cares

1:59 about laws and that was it that that was

2:04 the game

2:05 Emmett got his own bedroom Kyle and

2:07 Justin slept in bunk beds I slept in the

2:10 living room and sometimes on the balcony

2:12 we just didn't spend

2:14 much money that gave us a lot of ability

2:17 to screw up and make mistakes and then I

2:19 say the last thing that was kind of

2:20 interesting I only realized later is

2:22 that our ego was highly tied to our

2:26 startup we were not doing a startup to

2:30 have a cool resume item it was really

2:32 the only thing we had done on our own

2:34 and so I think at various points during

2:37 the company when it looked like we would

2:38 fail basically our startup failing was

2:41 our life failing right it was like well

2:43 this is the only thing you've done so

2:45 far and so if it fails you get F on life

2:47 and I think that we all had that feeling

2:51 very internally and therefore we just

2:54 couldn't really conceive of giving up so

2:57 I think more than anything I want to say

2:58 in the rest is presentation those were

3:00 the three things that saved our company

3:02 made our company work and strangely I

3:05 don't even think if you take one of

3:07 those things away any one of them we

3:09 would have died so this isn't one of

3:11 those things where it's like oh you can

3:12 grab for one or two and that's pretty

3:14 good we needed all three or else game

3:16 over so as I get into product I'm gonna

3:20 tell stories from justin.tv from really

3:25 early days at which when I was still

3:26 there and then also from a YC company

3:29 from a couple batches ago named poppy

3:31 it's a company that I've advised since

3:33 so invested in did YC great founder name

3:36 off knee and weirdly I just feel like I

3:40 needed to do a case study outside of my

3:42 own story somehow it's gonna help share

3:45 these lessons a little better so I

3:49 always like to start with what problem

3:51 are you solving because when I'm pitched

3:54 by founders most often they just want to

3:56 tell me what their idea is what they're

3:58 gonna do what their product does I think

4:00 what's interesting is that like

4:03 oftentimes they don't even know why they

4:08 don't know what's the problem that they

4:10 expect to be solved at the end of what

4:12 they're doing now I think that for some

4:14 businesses is totally fine

4:15 right I think that especially if you're

4:19 early on especially if things are still

4:21 in project phase whatever but I think at

4:24 some point pretty early on you have to

4:26 figure out

4:27 what are we doing and what do we expect

4:31 the result to be so at justin.tv the

4:34 first thing the problem we were solving

4:36 was entertainment we were making TV

4:37 shows Justin was the first one broadcast

4:40 his live 24/7 this was to be a TV show

4:42 so actually pretty easy for us to

4:44 understand whether that was working or

4:44 not is anyone watching right that's the

4:47 problem we were solving people watch TV

4:48 shows no one was watching so we didn't

4:51 solve the problem then when we pivoted

4:53 to an open platform the problem became

4:55 can we let anyone broadcast live but

4:59 that was problems trying to solve anyone

5:01 can broadcast live on the Internet and

5:03 once again once we understood that it

5:06 was very easy for us to judge whether or

5:08 not someone could do it we had this open

5:10 platform as anyone using it but I think

5:12 that like that was key to what we were

5:13 doing and then sometimes when I talked

5:15 to founders there's something they want

5:17 to do in the world there's a problem

5:18 that they're kind of vaguely interested

5:20 in or there's an idea that are vey

5:21 interesting but they really haven't

5:22 nailed down what's the actual problem

5:24 we're solving if you don't know the

5:25 problem you can't know whether you

5:26 solved it

5:28 the first thing I asked founders can you

5:30 state the problem clearly in two

5:32 sentences if you can't you don't know

5:34 the problem right in fact it's really

5:36 only take you one sentence so if someone

5:39 asks you a problem you're solving and

5:40 you find yourself delivering an essay

5:42 you're doing it wrong

5:44 - have you experienced the problem

5:46 yourself this is not always required but

5:51 is certainly helpful I've met a lot of

5:54 founders who are trying to solve a

5:55 problem for someone else who they've

5:57 never met never talked to and don't

5:59 truly know whether that person exists in

6:01 the world and so all things being equal

6:04 this is a great hand that you're on to

6:07 something well at least one person has

6:09 had this problem before the next one's

6:12 can you define this problem narrowly

6:15 what's interesting is when you get

6:17 started you can't really solve this

6:20 problem for everyone who has it

6:22 so when justin.tv first started we

6:25 couldn't let anyone broadcast live video

6:27 you had to have a laptop you had to have

6:29 go to internet connection you had to

6:31 have a webcam there are all these kind

6:34 of things you needed and so can we

6:36 actually now talk about all right we

6:39 want to make live video for everyone

6:41 but let's talk about the people that we

6:42 can address first who can we help first

6:45 and I think oftentimes founders kind of

6:48 want to skip that step they want to

6:49 solve the mega problem like I want to

6:51 cure cancer I'm only talking about when

6:54 everyone's cured as opposed to like what

6:56 can we address immediately how do we get

6:59 the first indication that this thing is

7:00 working and then the last one is the

7:04 problem solvable so here's what I'll

7:07 bring up with poppy so poppy is a

7:09 company that's essentially uber for

7:12 babysitting they make it really easy for

7:14 babysitters I'm sorry for parents you

7:16 need babysitter's to get babysitters

7:20 poppy is a very interesting company

7:22 because you need babysitter's for a lot

7:24 of different types of things some people

7:27 need a babysitter five days a week while

7:31 the parents at work right that looks a

7:32 little more like a nanny some people

7:34 need a babysitter whether it's an

7:35 emergency

7:36 oh I have you know I have a medical

7:40 emergency and I need a babysitter right

7:42 now because I need to go to the hospital

7:44 some people need it because there was

7:46 miss planning oh I thought well housing

7:48 was gonna be at this time and it wasn't

7:49 I thought I was gonna be here this time

7:51 it wasn't I need a babysitter some

7:53 people need a babysitter because they

7:55 have an infant right and so this baby

7:57 students have a bunch of skills some

7:58 people need a babysitter to watch their

8:00 15 year old to make sure they don't get

8:01 you know out of the house different

8:03 skills and so what's interesting is that

8:05 if you just start with oh we're gonna

8:08 help people get babysitters it's not

8:11 really good enough to understand what

8:14 you can address right away right which

8:16 one of those use cases do you want to

8:17 address if you were to state the problem

8:19 more narrowly though let's say we wanted

8:22 to start out with infants right we want

8:25 to make it easy for parents to get

8:27 babysitters for infants then we can

8:29 really ask the question is the problem

8:30 solvable I think one of the things that

8:32 poppy discovered when operating their

8:34 business is that the level of skill that

8:36 you need for a parent to trust you with

8:39 their infant when they haven't met you

8:41 is very very high so the idea that

8:45 you're going to have a rotating set of

8:47 people you haven't met watch your little

8:49 baby hard very very hard they have to be

8:52 very skilled and then on the flip side

8:55 uber model only works because there's a

8:57 whole bunch of basically replaceable

9:00 people with a common skill well it turns

9:03 out the people who've got the skills to

9:05 watch infants and make parents

9:06 comfortable with that tend to have nanny

9:09 jobs where they work lots of hours and

9:12 tend to be paid fairly well especially

9:14 in up-and-coming cities and so now we

9:17 have this disconnect where it's like

9:18 well we want to solve the problem of

9:20 infant watching for moms but that talent

9:24 pool who can solve the problem the

9:25 supply of babysitters they might not

9:28 exist problem might not be solvable and

9:31 so going through this exercise in real

9:34 time like with your products out in the

9:36 world you should be thinking about these

9:38 things you should be thinking how have I

9:39 narrowly defined the problem I want to

9:40 solve first and you should be always

9:42 asking yourself is it actually solvable

9:44 I think a lot of founders just don't

9:48 want to think about this because it's

9:49 hard it's hard to think about who you

9:51 want to talk to first it's hard to

9:53 understand oh maybe I can't solve that

9:54 problem I have to move on Abney was a

9:57 mother of two kids and she was really

9:59 pissed she couldn't solve this problem

10:00 because it was her problem but it's

10:03 turned out to be very very hard problem

10:06 to solve you know young infants

10:07 on-demand babysitters all right the next

10:12 question I always ask is who is your

10:14 customer and really you don't understand

10:19 the problem you're solving until you

10:21 understand who you're solving it for a

10:23 lot of times people just want to say

10:25 everyone everyone's the customer right

10:27 and that seems like it makes sense in

10:30 some cases right if you're building a

10:31 social network or a search engine right

10:33 everyone uses those things now what I

10:36 will say is that in almost all of the

10:38 products that everyone uses now there

10:40 was a time when almost no one used them

10:42 and the creators of those products had

10:45 to figure out who is the ideal first

10:47 customer and so if you don't have a good

10:50 answer to this question you're gonna be

10:52 lost do you have no idea who you should

10:54 talk to to ask them whether this problem

10:56 has been solved and you have no idea who

10:58 to talk to to figure out who this

11:02 product is for and I'd be surprised the

11:05 number of founders who are just building

11:07 something as if

11:08 they were writing a creative novel where

11:12 it's just a product of their own brain

11:13 and no interaction with anyone on the

11:16 outside and it's not even problem that

11:18 they're solved that they've experienced

11:19 themselves don't do that don't be one of

11:23 those founders like you can talk to your

11:25 users you just have to figure out who

11:27 they are the next question I often ask

11:31 is how often does your user have the

11:34 problem what's so surprising is when you

11:37 talk through startups with people

11:40 sometimes they choose problems without

11:43 quite understanding who the user is or

11:46 the frequency of the problem so give you

11:48 an example a lot of people will come to

11:50 YC and a popular idea back in the day

11:53 was to build a car shopping website now

11:58 if you guys have been on car shopping

11:59 websites especially about five years ago

12:01 they all basically suck they're they're

12:04 hard to use they're not very transparent

12:05 you kind of want to have this almost

12:07 Tesla experience of just buying a car

12:09 but they never actually work out that

12:11 way

12:11 and what's interesting is that like a

12:13 lot of founders come back to this

12:14 problem

12:15 over and over and over again and they

12:16 always think that their customer is the

12:18 person buying a car now the reality is

12:21 is that when you go buy a car assuming

12:23 it's not a complete lemon you typically

12:26 keep that car for seven years so what

12:28 happens if I told you I'm going to

12:29 create a start-up and if I home run with

12:32 my customer if my [ __ ] customer loves

12:34 me they're gonna come back seven years

12:36 from now that's hard it's very hard it

12:41 turns out a lot of the car buying

12:43 websites are not built for the person

12:45 who's shopping for car because that

12:48 person doesn't have a problem very often

12:50 they're actually built for the person

12:52 who's selling a car that person has a

12:55 problem every day every day the

12:57 dealership has to hit their numbers and

12:59 so you don't see as a customer how that

13:03 product helps the real customer the

13:07 person who's trying to sell cars and so

13:09 by doing this analysis I really try to

13:12 push founders in understanding who is

13:15 getting the most value out of this

13:16 product and it's really helpful if

13:18 you're trying to help someone with the

13:19 problem they have frequently

13:21 if you think about the products that you

13:24 use on a daily basis they tend to be on

13:26 the front screen of your phone you tend

13:29 to use them without even thinking they

13:30 become almost extensions of you if you

13:33 think about apps that you've installed

13:34 and then they're kind of on the third

13:36 page in the back or they're buried on

13:38 the second page is some folder those

13:40 tend to be the ones you don't use very

13:42 often hopefully they don't need you to

13:46 use them very often or else they're

13:47 probably not very good businesses the

13:50 next question I always ask is how

13:52 intense is the problem I find a lot of

13:55 founders think they have a good idea but

13:58 they don't do this frequency and

14:00 intensity analysis and so if you have

14:03 both an infrequent and low intensity

14:05 problem that you're trying to solve

14:07 you're gonna have a problem getting a

14:09 lot of customers even interested in

14:11 talking to you um all things we go if

14:14 you graph problems it's nicer for them

14:15 to be higher intensity higher frequency

14:18 let's think about a company like uber

14:19 for example usually when you are

14:22 somewhere and you need to go somewhere

14:24 else it's a pretty intense problem I

14:27 have to go to work I have to go to the

14:29 doctor I have to go pick up my kids

14:31 they're so intense you might have bought

14:33 like a twenty thousand dollar car to

14:36 help you do those things right so it's

14:38 an intense problem and then when you

14:41 think about frequency how often do you

14:43 move more than a mile more than walking

14:45 distance every day happens a lot and so

14:49 if you think about that even though you

14:52 look at the taxi market and you say ah

14:53 taxis unit before Hoover taxis that's

14:55 not that big of a market if you just

14:58 look at the customer you say high

14:59 intensity problem that happens very

15:01 often there's probably a good business

15:03 here the last one is are they're willing

15:07 to pay so many founders who come into YC

15:10 their first thought is completely wrong

15:14 on this front their first thought is I

15:15 need to give it away for free because

15:17 that's the only way I'm gonna get users

15:19 one of the things that I always push

15:21 them to do is think about it this way if

15:23 you want to know whether you have a good

15:26 product it's a lot easier to make it a

15:29 little bit harder for your users to use

15:31 it and then see if they use it anyways

15:34 because if I am

15:36 really intense problem and you say well

15:38 it's gonna cost 100 bucks for the person

15:40 with extremely tense problem they

15:41 probably think that's a deal if you have

15:44 extremely and if you don't have

15:46 extremely tense product problem you

15:48 charge zero dollars you can have a bunch

15:49 of users who come in who don't really

15:51 have the problem but they're just trying

15:54 something out if you try to learn from

15:56 them on how to improve your product

15:57 oftentimes they'll lead you astray so

16:00 strangely starting with a higher price

16:02 or a price is almost always better than

16:05 starting free almost always better and

16:09 if you have to start free you need to do

16:11 this analysis of how do you talk to the

16:14 users where the problem is actually

16:15 intense I talk to the users who are

16:17 using a product frequently in production

16:19 for real-world purposes as opposed to

16:21 the hobbyists talking your customers is

16:24 good but talking to the wrong customers

16:26 very very very bad I've seen a lot of

16:29 companies that are basically hijacked by

16:32 bad customers especially companies where

16:36 there's real-world costs so for example

16:40 you know if you have a company like

16:42 poppy there's real-world costs in

16:44 recruiting managing and working with all

16:46 of these baby sitters and if you have a

16:49 company if you have a customers trying

16:51 to basically take advantage of that

16:53 system being laid being non-responsive

16:55 being rude to the baby sitters that's

16:58 not going to help you run your business

16:59 and you'd be surprised at how many

17:01 hijack customers there are out there the

17:04 last question I always ask people is how

17:06 easy is it for your customers to find

17:07 because inevitably you're gonna need to

17:09 reach them and what's interesting I look

17:11 at the the last batch there are two b2b

17:14 companies one b2b company was here in

17:17 America and it was very easy for them to

17:19 find customers they could basically go

17:20 to LinkedIn find their customers on

17:22 LinkedIn find the name addresses email

17:23 them they get email thousand customers a

17:25 week another company doing b2b was in

17:28 China and interesting enough reaching

17:32 out over email and the b2b context in

17:34 China just isn't - well done practice

17:37 getting access to people's email

17:38 addresses is actually not very hard not

17:40 very easy and so strangely they had this

17:42 challenge of a relatively simple

17:44 business to explain a real intense

17:46 problem that happened often if they had

17:48 no way to reach their customers

17:50 and to basically invent new ways to do

17:52 it and so I often want to ask this

17:54 question because if your customers are

17:57 ridiculously hard to find

17:59 you better have a solution for that

18:00 upfront you can't build the whole thing

18:03 and expect for them to find you and so

18:06 often times you had a situation where

18:09 someone's trying to build a product for

18:11 either an imaginary customer or a

18:13 customer who can't really hope to use

18:15 the product I'm trying to get water to

18:18 people in the middle of a desert in the

18:22 Sahara all you need to do is download my

18:25 app and go online and then they can put

18:29 their GPS location and then we'll

18:30 deliver water to them that's not gonna

18:32 work but you'd be surprised at how many

18:35 people just don't think through those

18:36 logical steps all right next up does

18:40 your MVP actually solve the problem that

18:43 you want to solve this one is so

18:46 hilarious how often it comes up because

18:50 in the process of building an MVP things

18:52 just go weird and squirrely so you had

18:55 this problem and then you started

18:56 building it and then you talked to other

18:58 users and then before long you're

19:00 launching something and then you realize

19:02 it doesn't actually do the thing that

19:04 you promised and even the thing that you

19:05 want to do so part of your process of

19:07 building the MVP it's really helpful to

19:11 do these pre steps first it's really

19:13 really really helpful because then you

19:15 can always gut-check yourself on am I

19:17 actually solving the problem the other

19:19 thing is that it's really helpful to

19:21 build your MVP quickly typically the

19:24 longer it takes the more you're gonna

19:25 have MVP and problem drift or customer

19:28 drift if you decide to only build your

19:31 MVP in two weeks it's a lot easier to

19:33 stay on task and make sure actually

19:35 solving that problem for that customer

19:36 the way you test this by the way is you

19:40 give your product to customers like you

19:43 you have to do that that is a required

19:46 step what I find interesting is that a

19:49 lot of people think of their product as

19:51 a painting as something that could be

19:54 appreciated as a piece of art as

19:57 something that even if it's appreciated

19:59 by one person is special that's not what

20:01 you're making products are not paint

20:04 they're not art if users don't find

20:07 products useful then the products are by

20:09 definition not useful and they're a

20:11 waste of your time to build and I think

20:14 a lot of people want to be artists the

20:18 startup world is very unforgiving to

20:21 artists and I think that interestingly

20:22 after the fact a lot of people are

20:25 painted as artists right like Steve Jobs

20:27 is painted as this like magical artist

20:29 right and the other day he had to figure

20:31 out how to get make a phone that

20:32 millions of people would buy if only one

20:35 person bought the iPhone he would be

20:37 seen as a failure so the definition of

20:40 art is it only has to be appreciated by

20:42 one or maybe even none that's not the

20:45 appreciate that maybe just the Creator

20:47 that's not the definition of a

20:48 successful product so this is what you

20:51 should always be got checking does every

20:53 piece solve the problem the number one

20:54 problem with this question is that it

20:56 hurts the answer hurts you're gonna find

20:59 that a lot in startups where the answer

21:00 hurts you know doesn't solve the problem

21:02 but as long as we don't talk about it

21:04 maybe nobody knows it doesn't solve the

21:06 problem a lot of the answers inside of

21:09 startups are feel that way which

21:11 customers should you go after first a

21:13 lot of founders are very confused by

21:16 this question what I find interesting is

21:21 just like the instinct is to go after

21:23 customers by making the product free for

21:27 some reason I find a lot of people think

21:29 that their instinct should be to go

21:31 after the hardest customers first almost

21:34 as if it's like a proof like if I can

21:35 get this impossible person to use

21:37 something then like it'll be easier I

21:40 know that I've made something good I

21:43 like to start from a different point

21:45 it's an MVP you know you've made

21:47 something bad like that's the definition

21:49 of MP it's bad so the real question is

21:52 like how do you find people who are

21:54 willing to use a bad product right there

21:57 have to be the most desperate the most

22:01 desperate and so a lot of times I talk

22:03 to founders I really pushed them towards

22:05 who are the most desperate customers and

22:08 how do you talk to them first that's

22:10 what I define as easy desperate if

22:13 you're having like a you know if you're

22:15 trying to sell a simple piece of

22:16 software to someone

22:18 thousand dollars a month and you're

22:19 engaged in a six month conversation with

22:21 a company that's not a desperate company

22:24 move on in fact when you're doing

22:27 enterprise sales early as a start-up

22:29 like you're looking for even more

22:31 desperate customers just because

22:33 literally takes so long to sell them so

22:35 if you don't feel like you're dealing

22:37 with desperate people if you feel like

22:39 you've dealt you're you are trying to

22:40 get impressive customers who aren't

22:42 desperate you're probably doing it wrong

22:45 literally the number one thing I often

22:48 tell found is just like whose business

22:50 is gonna go out of business without

22:52 using you which people out there are not

22:55 going to be able to get to work or to

22:57 watch their kids how do you find the

22:59 people who are just literally are

23:00 screaming for something like this and

23:03 then how do you talk to them and not

23:05 talk to your friends I had a bunch of

23:08 friends who were using social cam right

23:12 my company was doing video for for

23:14 sharing with friends and they weren't

23:15 really using it they were using it

23:17 because it was like my app and there

23:19 were friends with me I literally had one

23:20 friend it was like super honest about

23:22 this Steve the CEO of Reddit when we

23:24 sold Socialcam he literally said thank

23:26 God now I can delete this app from my

23:28 phone so the perfect definition of

23:34 someone you should not be trying to get

23:35 product feedback from right and so he

23:39 didn't have the problem we were solving

23:40 many of your friends won't have the

23:42 problem that you're solving make sure

23:44 you find D and by the way the kind of

23:47 community of start-up people and/or

23:50 investors usually don't have the problem

23:53 that you're solving so if you're using

23:54 investors as a trigger for am i solving

23:57 the right problem or like do they find

24:00 this useful it's almost never the case

24:02 almost never the user of a product that

24:05 comes into I see and so ignore your

24:07 investors ignore your friends like they

24:10 will lead you 100 percent astray out of

24:13 good intentions they'll try to be

24:14 helpful well at you know I've never

24:16 lived in the Sahara I've never been

24:18 thirsty but maybe it should work like

24:21 this right live like horrible run away

24:24 run away once you start having customers

24:27 I think it's a very helpful exercise to

24:30 try early to identify

24:31 by bad customers these are people who

24:35 are blasting your support these are

24:38 people who are constantly

24:40 constantly complaining

24:43 my co-founder Justin he had a company

24:45 that was basically on demand personal

24:47 assistant and he was the first one who I

24:50 met who actively fired a customer is

24:54 basically uber for personally support

24:55 systems called exact and literally a

24:57 customer would have the exact do

25:01 something like like crazy like something

25:03 you couldn't do right like we organized

25:05 my house the way I want things organized

25:07 and I'm not gonna tell you I want them

25:09 organized or go shopping for me but I'm

25:12 gonna critique every single like piece

25:14 of fruit and vegetable that you picked

25:16 out she was like completely unrealistic

25:17 expectations and so after refunding the

25:19 person four times for four different

25:21 tasks the person did a fifth task on the

25:24 product right because he's getting a

25:26 bunch of value for free and Justin Cosmo

25:28 says you're fired you can't ever use

25:29 product again like look for those people

25:32 because if you aren't delivering

25:34 anything of value there will be people

25:36 trying to exploit that value and some

25:39 might be doing it not out of the

25:41 goodness of their heart so don't let

25:45 these people lead you astray we talked

25:47 about this don't discount now here's a

25:51 caveat on discounting Parker from

25:56 zenefits came to I see a couple years

25:58 ago and he gave this great talk about

26:01 enterprise sales and zenefits his

26:03 product that's given away for free so

26:04 it's actually kind of a interesting

26:06 enterprise sale and one of the things he

26:09 said that really got to me was that

26:13 there are ways to convince organizations

26:17 basically you can structure discounts

26:20 and incentives into your sales pitch if

26:23 you basically understand what value are

26:26 getting back so his example was he would

26:30 try to sell to a company to switch on to

26:32 zenefits for their health care and he

26:35 would say look because of this third

26:38 party let's just say AWS has given us a

26:41 discount who knows why right we just

26:43 bought dedicated instances so now

26:45 we have 40 percent lower AWS bills so we

26:48 can actually pass on some benefit to you

26:51 but only for the next 30 days now I feel

26:55 horrible even telling you this because I

26:57 want you to take as much time as you

26:58 need to buy my product I would just hate

27:00 if you bought it on the 31st day and I

27:03 couldn't give you this discount now this

27:05 isn't a let me give this away for free

27:08 like because I'm afraid people won't use

27:10 it this is a very structured process

27:13 that he did he basically incorporated a

27:16 deadline based on some third party

27:19 providing a benefit to the customer and

27:22 he knew that when he said this to the

27:25 customer every time that this was talked

27:27 about internally the deadline would be

27:29 brought up and the discount would be

27:31 brought up and suddenly this has now

27:33 become not a way to be afraid right - oh

27:37 I'm not sure how many customers minutes

27:38 make it free became a way to speed up

27:40 the process and his discount was just

27:42 baked in like he just priced the product

27:44 15% higher so it's like literally like

27:48 that is the way to do it

27:50 the way not to do it is I'm afraid no

27:52 one's gonna use it so I have to talk not

27:54 charge any money okay

27:57 the next is how to set up metrics how

28:00 many of you are using Google Analytics

28:03 as your primary metrics products raise

28:05 your hands okay you are doing it wrong

28:10 yes so setting up metrics is something

28:16 that's like super important very early

28:20 in your company because it's how you

28:21 know whether your product is being used

28:22 or not and it's one of the number-one

28:25 sources of new product ideas and

28:27 inspiration so Google Analytics I would

28:30 say is a great product for knowing how

28:32 many people came to your website today

28:33 and how many pages they viewed which

28:35 used to be relevant and it's not really

28:37 relevant anymore and where they came

28:39 from

28:40 um what it's not a great product for is

28:42 identifying what people's actions were

28:45 when they were using your product did

28:47 they click this button did they see this

28:49 screen how long were they on the page

28:51 for before they did something else did

28:53 they leave something in their cart for

28:55 all of those things you want an events

28:57 based metrics product

28:59 Mixpanel amplitude heap I think we've

29:01 funded like 50 of them they're like 100

29:03 of them out there you should be using

29:05 one of them if you're not you can't be

29:08 sophisticated at building your product

29:10 this is just kind of a prerequisite so

29:13 get on it and this goes back to the

29:15 early thing that I mentioned which is

29:16 technical teams for a technical team

29:19 implementing Mixpanel is ridiculously

29:21 easy for a non-technical team it's

29:22 basically impossible this is just one of

29:26 the many advantages of having a highly

29:28 technical team you actually know what

29:29 users are doing without this you're just

29:31 missing a huge part of what you need to

29:34 know the next thing and Suhail from

29:38 Mixpanel gave a great talk about how do

29:40 you set up mix panel one of the

29:42 challenges of setting up mix panel is

29:43 the second that you're sitting there

29:44 saying I want to track what my users are

29:46 doing you can come up with like 150

29:48 things your users can do with your

29:49 product and you want to track all of

29:50 them that's often a mistake if your

29:55 analytics product is got too many

29:57 analytic sitting in it in the beginning

29:58 it will be hard to use and part of what

30:01 you're doing if you've never used

30:02 product like links panel before is

30:04 learning how to use it and most

30:06 importantly teaching your employees and

30:08 your co-founders how to use it because

30:10 this product should be a product that

30:11 everyone in your company understands how

30:13 to use is everyone your company should

30:15 understand how the product is

30:16 functioning this is not something that

30:18 like the CTO uses and creates reports

30:20 from this is interactive product that

30:22 everyone can use

30:23 so to start pick five to ten simple

30:25 stats

30:27 let's take Instagram as an example right

30:29 if I went to pick five to ten simple

30:31 stats for Instagram let's say open the

30:34 app trade in an account took a photo

30:37 applied any effects share the photo

30:40 that's probably all I need in the

30:42 beginning right I mean the number one

30:45 mechanism for Instagram is taking a

30:46 photo and sharing it I can track that

30:49 I'm pretty happy the other thing that I

30:53 will warn you about is that if your

30:55 product is good the naming conventions

30:58 for these stats are going to become very

30:59 important because one day there will be

31:00 a hundred or even a thousand stats you

31:02 track so think a little bit ahead of

31:04 time and don't name something something

31:07 that only you'll understand make sure

31:09 that you if your company's good many

31:11 many people have to look at the

31:12 stats make measurement a part of your

31:15 product spec oftentimes when I talk to

31:18 founders they say we built it on this

31:20 release and we'll add the measurements

31:22 some point in the future I don't

31:24 understand how that works you build

31:26 something you want people to use but

31:28 you're not incorporating the measurement

31:30 that tells you whether people are using

31:31 it that doesn't work building

31:35 measurement is part of a product spec so

31:38 when you spec out a product you better

31:40 spec the stats you expect to be tracking

31:42 and you should also spec the stats that

31:45 you think are going to improve when

31:48 you're building that product that should

31:50 be part of the spec it should be part of

31:52 the first release otherwise you're

31:53 flying behind and this is just countless

31:57 times a justin.tv this is screwed us

31:59 okay

32:01 part development cycle so justin.tv was

32:05 and twitch was three yell kids and one a

32:09 mighty kid yell probably the most

32:12 productive skill you're taught is how to

32:13 argue with other yell kids and so the

32:16 number one way to get products developed

32:18 at justin.tv was to win an argument with

32:20 the three yell kids

32:22 kyle disliked this so much that he

32:25 actually switched his sleeping schedule

32:27 so that he wouldn't have to be involved

32:29 in these arguments so we were awake from

32:31 about 8 a.m. to about 12 midnight he

32:34 would wake up around 11:00 midnight and

32:36 write code all night and then go to

32:39 sleep in the morning so he wouldn't have

32:40 to argue with us on what stupid thing to

32:43 build one of the classic arguments at

32:46 justin.tv that lasted approximately

32:47 three months was the background color

32:52 for the original site so the original

32:53 site is just one page Justin wanted a

32:57 black background I wanted a wood grained

33:00 background three months of bait we

33:04 settled on changeable backgrounds so

33:10 there were five background options

33:13 clearly idiotic like I said we made many

33:15 of these mistakes um we didn't actually

33:18 really learn how to do product

33:19 development cycle until we failed at it

33:21 for about five or so years

33:26 and during that time this is what bad

33:29 product-moment cycle looks like one we

33:32 would release every three months for a

33:35 web only product that is horrible second

33:40 we would have a product meeting and we

33:41 wouldn't write anything down right it

33:45 was just four of us can't you remember

33:46 you're an idiot if you can't remember a

33:48 conversation before people had right and

33:50 if you forget something just ask one of

33:52 the other four people in the room right

33:54 No so as a results during the first

33:58 month of the dev cycle we'd all go off

34:00 working on slightly different versions

34:02 of the thing we wanted to build because

34:04 we didn't write down this back then at

34:06 the end of that month we'd come together

34:08 and we'd be like oh wait this isn't

34:11 we're not really building all the same

34:12 thing and then we'd have another product

34:14 meeting where we didn't write anything

34:16 down and go off and build again for

34:19 another month at this point right two

34:22 months in we probably have about three

34:24 weeks of productivity and about five

34:26 weeks of just stuff that's gonna have to

34:28 be thrown away at this point in we kind

34:31 of come back together and realize that

34:33 we're not 2/3 of the way done through

34:35 this sprint were less than 1/3 of the

34:37 way done and we're starting to get sick

34:39 and tired of this feature that we're

34:40 building so then we basically say

34:43 alright slash-and-burn

34:44 let's just make a shitty version of it

34:46 and we take another month to do that now

34:49 we've worked on this product for three

34:50 months if you had any good or new or

34:53 interesting ideas during that three

34:54 month period of time you were told we're

34:57 already working on something else so

34:59 your ideas are worthless just write them

35:00 down somewhere

35:01 whatever we're working on this thing

35:02 right now at the end of the three months

35:04 instead of wanting to iterate we were

35:06 sick of the da TM feature we just spent

35:08 three months building poorly so we would

35:11 launch it and if it wasn't used right

35:13 away we would come up with some new

35:15 brainstorm on some brand new feature

35:17 that would rescue the company this is

35:19 the wrong way to run a company it was

35:22 absolutely horrible I was talking to

35:23 Jeff earlier the major product decisions

35:25 that justin.tv made that carried through

35:28 to twitch - today was chat on the right

35:31 video on the left we decided that in

35:33 2006 it is the same way 2018 the vast

35:38 majority of the products issues we made

35:39 were

35:40 and never saw the light of day because

35:41 they went through a process like this so

35:44 if your process revolves around arguing

35:47 revolves around not writing speck

35:50 revolves around long dev cycles you are

35:53 doing it wrong you are 100% doing it

35:55 wrong what I'm gonna give you is a model

35:58 of how we figured out how to solve the

36:00 problem steal as much or as little of

36:02 this as you want but understand that if

36:04 you have any of the symptoms I'm talking

36:06 about you need to solve them or else

36:08 your company is just going to be much

36:10 less productive than it could be first

36:12 you need to actually have a number that

36:15 you track that reflects how good your

36:18 company is doing almost always if you

36:21 ever are going to charge money to your

36:24 customers this number should be revenue

36:27 almost always if you are never going to

36:30 charge your customers money like

36:32 Facebook then maybe it should be a usage

36:36 based metric like how often do your

36:37 customers come back every day like DEA

36:40 use it is almost always one of those two

36:44 usage if you will never charge the

36:46 customers money if you will charge the

36:48 customers many people will invent

36:50 reasons why these two metrics don't

36:53 apply to their business one percent of

36:56 them might be right 99 percent of them

36:57 are probably wrong

36:59 whatever this KPI goal is make sure

37:01 you're measuring it make sure that

37:03 everyone in your company knows what this

37:04 goal is every day helpful to put it on

37:06 some screen somewhere if an investor

37:08 asked you what your KPI is you not only

37:10 should be able to say what it is

37:11 sheíll to say what the metric is you say

37:14 where the metric is now where it was

37:15 three months ago where it was when you

37:16 started this is kind of table stakes the

37:22 next thing that we would do is as the

37:25 kind of product person I would come into

37:27 the meeting and I would say this is the

37:29 KPI we're looking to improve this this

37:31 cycle at Socialcam the top level KPI was

37:36 da use and the three ways that we

37:38 thought we contributed to da use was

37:40 either new users retention of users and

37:43 new content created those three things

37:46 so every cycle we ran one of those three

37:49 numbers moving that number so that the

37:51 right direction was the goal and we'd

37:53 run an open brain

37:54 brainstorm for us would take a couple

37:56 hours and it was a real brainstorm it

37:58 wasn't the brainstorm we're like you say

38:00 what about this and your co-founder says

38:02 that's a dumb idea that's not a

38:04 brainstorm

38:05 the real brainstorm is that any idea

38:07 that's stated is written on the board

38:09 the cool thing about these brainstorms

38:12 is that everyone's computers were always

38:13 open to Mixpanel

38:15 so if you had an idea or you had a

38:18 thought you could always just go in and

38:20 check the metrics and see like oh is

38:21 that right or is that wrong you'd be

38:24 surprised at how much value there is and

38:26 seeing your idea on the board not

38:29 everyone's gonna get to have built what

38:31 they want to have built but the fact

38:33 that your idea was considered and added

38:35 to the board actually makes people feel

38:37 a lot better than otherwise people feel

38:39 horrible when their ideas are shot down

38:42 CEOs their job is to make their

38:44 employees not feel horrible all the time

38:46 sometimes I think CEOs think their job

38:48 is to shoot down ideas it's not it's not

38:52 gonna help you at all I mean everyone by

38:54 the way in our company participated in

38:56 this brainstorm at that point that was

38:57 for people so easy to do the next thing

39:01 was we did what's called easy medium

39:03 hard

39:03 so our brainstorm was actually typically

39:06 split up into three categories new

39:08 features or iterations on existing ones

39:10 bug fixes and/or other maintenance and

39:13 tests a B tests we want to run we have a

39:17 whole board filled out with ideas on

39:18 these three categories and then we go

39:20 through and do what's called easy medium

39:22 hard for us hard mint it would take one

39:24 engineer most of the dev cycle to build

39:26 medium typically meant to take a day two

39:29 days easy means we could do multiple in

39:31 a day this is extremely important how

39:34 many of you in this room do not know how

39:36 to write code raise your hand there we

39:38 go I am one of you it's extremely hard

39:42 if you don't know how to write code to

39:44 figure out whether your idea is easy to

39:46 build or hard to build that's something

39:48 that you actually learn as a skill over

39:50 time and this process basically is the

39:53 process that can help educate you it

39:56 turns out that easy ideas get built way

39:58 faster way more quickly than hard ideas

40:00 and it turns out that most hard ideas

40:03 can be restated as an easy idea if you

40:06 just understand what

40:07 bits of your heart idea are both useless

40:10 and hard and most of the time there are

40:13 useless and hard bits and hard ideas

40:15 that can just be removed and so for us

40:18 this was educating everyone the team and

40:20 for us we had a cross-functional team so

40:22 someone might not realize that this is

40:23 really hard on the video system it might

40:25 be an easy web feature but hard on the

40:27 video system or vice versa so this was

40:28 basically educating everyone a team on

40:30 what's easy medium hard it also created

40:32 an objective standard by which to start

40:34 thinking about these ideas instead of

40:35 just based on the argument ability of

40:37 the person delivering them it was like

40:39 well your idea is like really freaking

40:42 hard and seems like it wouldn't move the

40:43 numbers that much based on Mixpanel

40:45 whereas like this other person's idea is

40:47 super easy and probably can move numbers

40:49 a lot the next thing we would do is we

40:52 decide hard first so we look at all the

40:56 Harz and we say which hard is gonna

40:58 impact the KPI the most and then we move

41:02 to Easy's and then we move to one with

41:04 moved at mediums and then we moved to

41:06 Easy's what was interesting is that like

41:08 just with the ideas on the board and

41:10 with easy medium hard a lot of the ego

41:12 was removed from the debate because one

41:14 you knew your idea has been considered

41:16 and two you'd sent some objective

41:18 measure about how hard it was and three

41:20 because the board has a bunch of ideas

41:22 on it now it's probably pretty easy for

41:25 you to find an easy idea that you really

41:26 like and so you're just gonna be excited

41:29 that that's probably gonna get in and

41:31 you're really hard idea that's fine if

41:33 it doesn't the next step is you have to

41:37 write the spec this is where everyone

41:39 [ __ ] up the meeting might be going on

41:41 for four hours now and this is the step

41:44 no one likes you actually go through and

41:46 you actually write down what do we mean

41:49 by we're adding video filters to

41:51 Socialcam what do we mean by were

41:54 allowing people and justin.tv and twitch

41:56 to chat with one another what does that

41:58 actually mean how is it going to work

42:00 this is really important and once this

42:04 is done you can then distribute tasks to

42:06 the team now we would run these cycles

42:08 every two weeks at Socialcam because

42:10 back then summing to the AppStore took

42:12 longer if you're doing a pure web

42:15 product you can run these cycles once a

42:16 week the rule that we had to make the

42:19 team not hate these really

42:21 meetings is the only meeting we had this

42:24 is the only meeting and so it might take

42:26 two hours it might take six but for that

42:28 to occur period of time there were

42:30 northern meetings in fact for me being a

42:32 non coder my number one job was to shut

42:34 the [ __ ] up because I could create a

42:39 problem right

42:41 we're busy working we have this written

42:43 spec everyone knows what they're doing

42:44 if I have some brilliant idea during

42:47 that two weeks right that'll throw the

42:50 whole thing into chaos

42:51 suddenly the written specs not important

42:53 we're back to the drawing boards or

42:55 we're changing things Yeti ID Etta what

42:57 I had to realize is that every two weeks

42:59 we do this over again so for that

43:00 burning idea just [ __ ] wait two weeks

43:02 and we're gonna have the meeting again

43:04 and then we can get it in and turns out

43:05 you're burning ideas probably wrong so

43:07 it's totally fine to wait two weeks to

43:09 try to kiss people do something wrong is

43:11 totally fun as opposed to like having

43:14 this cadence meant every two weeks we

43:16 had success every two weeks if we built

43:18 what we said we were gonna build we felt

43:20 good and then that cadence meant that

43:23 we'd go into the next cycle and do even

43:24 more this cadence is extremely important

43:28 because it's going to take you guys a

43:29 long time to find product market fit

43:30 maybe trying a lot of things we

43:33 iterating a lot and if that process

43:35 doesn't feel fun you're getting very

43:38 frustrated

43:39 this made the process feel fun because

43:41 we had goals and accomplish them versus

43:44 iterate a lot of YC companies a lot of

43:49 founders in general will tell me our

43:51 thing isn't working it's been two months

43:54 it's time to pivot when I think about

43:57 that statement it blows my mind right

44:00 you're building a new product for a

44:03 customer who might not have ever used

44:04 the product before you're oftentimes

44:06 exploring a problem that you only know

44:08 to some degree or you've only

44:10 experienced it personally what makes you

44:12 think two months is enough time to know

44:14 whether you figured something out what

44:17 impressive thing only took two months to

44:19 build so if you're not thinking that the

44:23 process of coming up with a solution for

44:24 this problem is probably the more like a

44:26 two-year process you're doing it wrong

44:29 if you are unsatisfied with significant

44:31 progress in under two years you're

44:35 probably doing it wrong it's going to

44:37 take time you're doing something hard if

44:40 it was really easy someone else would

44:42 have done it so I define pivot as

44:46 changing the customer or changing the

44:49 problem this should be rare this should

44:52 happen infrequently many times this

44:56 means you should start a new company i

44:59 define iterate as changing the solution

45:03 it turns out you had the right customer

45:06 you had the right problem your MVP was

45:08 shitty and it didn't work

45:10 we need a new solution it turns out

45:13 maybe your MVP was great but it didn't

45:14 solve the problem with a new solution it

45:17 turns out you showed the product to your

45:18 customers and they didn't want to use it

45:19 even though they have burning problems

45:21 we need a new solution often times I see

45:23 this in Reverse people think solution

45:26 first and when cut the customers they

45:29 thought didn't like their product they

45:30 try to find some other random customer

45:32 who does he might even have a completely

45:34 different problem and they tried

45:36 shopping around their solution because

45:37 they think their solution is the genius

45:39 part I think the problem is the genius

45:42 part I think identifying a problem that

45:44 other people haven't figured out is

45:46 worth working on is the genius part

45:48 right Facebook wasn't first a social

45:51 networking and Google wasn't first a

45:53 search engines their genius was

45:55 understanding that the people who came

45:57 before them hadn't solved the problem

45:59 and if they could solve the problem

46:00 better they'd built huge companies their

46:04 genius wasn't oh we built this cool

46:06 thing let's just figure out who might

46:08 want to use it wrapping up a little bit

46:10 here I always tell the story about fake

46:13 Steve Jobs versus real Steve Jobs a lot

46:16 of people think that Steve Jobs is this

46:19 person they should emulate but they have

46:21 a false picture in their heads of what

46:23 Steve Jobs was they think that like he

46:26 dreamed perfect ideas out of his head

46:28 and into the world and what's funny is

46:31 that I think often times people look at

46:32 the iPhone as perfect example of this

46:34 but they look at their iPhone today your

46:38 iPhone today is [ __ ] magical the

46:41 first iPhone sucked in almost every way

46:44 and they don't realize that Steve Jobs

46:47 wasn't somebody who

46:48 was just not iterating who just imagine

46:51 your perfection minute one Steve Jobs

46:53 was iterating at every step so I like to

46:57 remind people what the first iPhone did

46:58 first iPhone no 3G back when 3G was a

47:02 standard feature so you have this great

47:05 internet browser but you can only use it

47:06 on edge which means it [ __ ] sucks

47:08 right one carrier oh you don't have this

47:12 carrier sorry switch carriers figure

47:14 that out

47:16 horrible battery life screen cracked all

47:19 the time

47:19 no App Store you can't even download

47:22 other apps that was the first iPhone

47:25 everyone forgets that iPhone so if you

47:28 are the person in your company who is

47:29 being fake Steve Jobs is saying the

47:32 product has to be this way because what

47:34 I said [ __ ] the customers [ __ ] everyone

47:36 else [ __ ] you make the product the way I

47:39 want it to be you're being fake Steve

47:41 Jobs real Steve Jobs released a shitty

47:45 MVP that was revolutionary but still

47:47 fairly shitty and every year iterated it

47:50 until you have the thing in your pocket

47:52 right now which is pretty damn good real

47:55 Steve Jobs iterates and talks to

47:57 customers fake Steve Jobs just dreams

48:00 and creates art don't be fake Steve Jobs

48:03 okay so with all this I want to go back

48:07 to the beginning what I said the

48:11 beginning still holds justin.tv the only

48:13 reason why I actually even know any of

48:15 these rules is because we broke all of

48:16 them the one thing that justin.tv and

48:19 twitch had was a really strong technical

48:22 team with high ego in the product and lo

48:25 burn when we started figuring things out

48:28 with twitch it was very interesting

48:31 gamers had been on our product the whole

48:34 time gamers had been streaming on

48:37 justin.tv since almost the beginning at

48:39 any given time there were twenty percent

48:40 of our traffic for years we ignored them

48:44 we ignored them we ignored them we

48:46 ignored them ignored them they still use

48:48 the product we didn't build features for

48:50 them they still use the product they

48:52 must been pretty [ __ ] desperate

48:53 because they still use the product year

48:55 after year the number one thing that

48:58 changed when we started working on

49:00 twitch

49:02 we started talking to them and what's

49:03 weird was it's not like we were talking

49:05 to other users and the only reason we

49:08 didn't talk to any users we had this

49:10 like crazy part development cycles we

49:11 couldn't do that with talking users too

49:13 so what we did in the beginning was we

49:15 literally just sat down with these

49:17 gamers and we said what did you want and

49:19 what's funny is we didn't build them

49:21 anything very special they were like oh

49:23 like lag socks or the couple they wanted

49:25 like little things what was great about

49:27 it was they realized we were now going

49:29 to build something for them and no one

49:32 on the internet was building things for

49:34 these gamers and they realized that when

49:36 we said we're gonna build something it

49:37 came out when was the last time that you

49:40 talked to someone building a product

49:42 that you like and you said can you do

49:44 this and they did it was last time you

49:47 suggested a feature to mark at Facebook

49:49 and then the feature came out never like

49:52 it's one of the magical things you can

49:53 deliver as a startup is you can talk to

49:56 a passionate user and then you can build

49:57 what they want and then you can say here

49:58 it is and they will fall in love with

50:00 you even if those features are

50:01 relatively mundane because let's switch

50:04 today chat on the right video on the

50:05 left the same product what was great

50:09 about this process was by talking to

50:11 them they realized that we were on their

50:13 side we realized they were building

50:14 something for them so they tell their

50:15 friends that was the major change if we

50:20 didn't have the technical team if we

50:22 weren't sheep if our ego wasn't involved

50:25 I never would have gotten to that point

50:27 and if you look at the history of

50:29 justin.tv in the first five years it

50:32 went from being worth nothing to being

50:34 worth about twenty four million dollars

50:36 in the next three years it went being

50:39 worth twenty four million dollars

50:40 touring worth a billion like that's what

50:43 software can do when you when you hit

50:44 the right customer let's do a couple

50:46 questions in the back

50:48 so you've mentioned that product so the

51:06 question is put it more generally should

51:10 you be going free if your final idea for

51:13 a product is to be free what I would say

51:15 is this if your users are users who you

51:19 never plan to charge then it's totally

51:22 fine free to be free but if you do plan

51:24 to charge them in some way it's really

51:26 helpful to charge them as soon as

51:28 possible because you want to know

51:30 whether or not they're willing to pay

51:32 and certainly if their business depends

51:34 on it it's especially helpful to charge

51:37 them so that's the measure that I would

51:39 use and they're all kinds of little

51:41 tweaks and and so and so forth but at a

51:43 high level do you ever plan to charge

51:45 them I charge them if you never plan to

51:48 charge them you can plan to monetize

51:49 based on ads which is really usually the

51:52 way that you never plan charges you can

51:53 monetize with ads if you're not going to

51:55 monetize with ads you probably should

51:57 start charging them all right next

51:58 question

52:18 revenue if you're metric if your KPI is

52:23 metric do that if your KPI is revenue

52:28 and the number is zero should you still

52:31 be tracking that as your top-line KPI

52:33 the short answer is yes you should be

52:35 depressed looking at that number every

52:36 week too but that's that's the answer

52:39 now let's be clear like I said with

52:41 Socialcam right there are contributing

52:43 numbers to that right and so whereas

52:46 like DEA use was our top-line metric

52:48 right we also the things we thought

52:50 contributed to that new content new

52:52 users retained users those numbers we

52:55 can move so if you're in a sales type

52:57 business your KPI number one revenue if

53:00 it's zero that should [ __ ] you in face

53:02 zero zero zero horrible but then you

53:04 should ask yourself maybe your three

53:06 other metrics are how many conversations

53:08 did we have this week how many people

53:10 are in contracting how people aren't

53:12 onboarding right and those can be your

53:14 three numbers that will if those numbers

53:17 move you expect the revenue number to

53:19 move and they can keep you motivated but

53:21 your your top-line KPIs no [ __ ] like

53:24 absolute no [ __ ] good we're a

53:28 hardware company pre-launch and so our

53:31 users our target market experience our

53:33 problem five to nine hundred times a day

53:35 and the intensity is can we pay our rent

53:38 or not but we'd like to offer pre-sales

53:42 as a way of getting the product to

53:43 market do you guys have any tips on

53:45 doing that do we have any tips on

53:51 pre-sales what I would say is that there

53:53 are many many tips on pre-sales I would

53:55 email some founders who've done it

53:57 before that's one of my best tip is to

53:59 email them and ask them what they did

54:00 the number one mistake I see with

54:03 pre-sales is discounting the number one

54:06 mistake is that basically especially

54:08 hardware founders will misunderstand how

54:10 much they need to charge so they don't

54:12 lose money and so their presale becomes

54:14 their death so I'd avoid that all right

54:18 two more quickly and the way back what

54:29 was the hardest part of having a slow

54:30 burn well we were young so it wasn't

54:33 hard at all we were living like we lived

54:35 in dorms I think it's a lot harder now

54:39 right I've got a kid I got a wife got an

54:42 apartment got a car it'd be really hard

54:44 now and so I think really the hard part

54:49 about slow-burn is can you adjust your

54:52 lifestyle if you've already leveled it

54:53 up and if you haven't leveled up your

54:55 lifestyle yet and you're still working

54:57 you know if you're young and you're

54:58 still working at a company that's paying

54:59 you well not leveling up your wife style

55:02 is a big way to stay ready to do a

55:04 startup adding on the mortgages and the

55:07 cars and the vacations makes it a lot

55:09 harder I just know a lot of people who

55:11 never could come back from that

55:13 absolutely

55:15 her back going from beta how do you go

55:27 from beta to early MVP I don't really

55:29 know what those distinctions are like

55:30 all I know is are people using your

55:32 product yes no like if people are not

55:35 using your product get to the point

55:37 where people are using your product

55:38 extremely quickly once people are using

55:40 your product there's all different

55:41 labels for it beta pre-launch alpha blah

55:44 blah who cares right it's just that's

55:47 the dividing line like are people using

55:48 a product the next question is are you

55:50 actually solving their problem that's

55:52 the next question not like are we

55:54 following this line of launching things

55:57 most YC companies will launch many many

56:00 many many times so that progression

56:03 isn't really that important are people

56:05 using your product great BAM

56:08 you're launched congratulations call

56:11 whatever you want all right this is a

56:44 great question and I have I'll have an

56:48 unsatisfying answer so the question is

56:50 basically how do we figure out what to

56:52 build next here's my answer the reason

56:56 why you have a part development cycle is

56:58 that you can work on multiple things

57:00 usually there isn't a right answer

57:02 usually all of the things that you want

57:04 to build won't work so what you need to

57:08 do is you need to create a process in

57:09 your company to build things quickly so

57:12 that you can actually see whether

57:14 they're work or not and then you can

57:15 iterate them from there so it's far more

57:17 important to have attack tactically

57:20 talented team that can build MVPs

57:23 quickly in a non frustrating way and

57:26 then measure the results than it is to

57:28 be

57:28 a super genius who can imagine what's

57:31 going to happen in the future without

57:32 actually knowing now I'm the big picture

57:34 you have to have that imagination for

57:36 your vision for where it's gonna be ten

57:38 years now you have to have that

57:38 imagination for the little technical

57:40 like tactical move in the next three

57:43 months like it's really hard to nail

57:47 those if you have a process that can rip

57:49 out things quickly and then only iterate

57:51 the things that are working that'll

57:53 serve you far better our mistake was

57:55 that at justin.tv

57:56 it was thinking every time we've got the

57:59 home run let's only swing for home runs

58:01 and of course it would take three months

58:03 to do it because we got to make it

58:04 perfect right and then the whole spike

58:07 spiral of death alright um the last

58:09 thing I'll say is my email address is

58:11 Michael at Y Combinator comm strangely I

58:14 tell people that I answer every email

58:15 and people mostly don't believe me and

58:19 the ones who do email me and I reply and

58:21 so everyone I talk to you and everyone

58:24 online there's really only two

58:25 categories of people the people who

58:26 don't believe me in which case great

58:28 continue your lives and the people who

58:30 will believe me and if you need help

58:31 they'll email and I will try to help you

58:34 the best I can

58:34 it's really that simple you don't need

58:37 to have networked with me Y Combinator

58:39 is fairly easy to spell and so is

58:41 Michael so you should be able to figure

58:42 that out

58:42 thank you

58:44 [Applause]

58:55 you

Key Takeaways

Start a conversation

How to Create a Product Roadmap