0:02 the average person probably doesn't know
0:03 what segment is mm-hmm
0:06 so could you explain for sure so segment
0:09 helps companies give their customers a
0:11 better customer experience and we do
0:12 that by helping them organize all of
0:14 their internal data about all their
0:16 interactions with the customer so for
0:19 example if you go to the bank they
0:21 interact with you at the ATM at the
0:24 teller via a phone call center they have
0:26 a web web app a mobile app they send you
0:27 emails they're interacting with you
0:30 across this huge surface area and they
0:32 need to be able to coordinate that
0:33 interaction right they need to know that
0:35 if you encountered an error on the ATM
0:37 the teller needs to be able to say like
0:38 I'm so sorry you encountered an error
0:40 about to be able to help you and so what
0:42 we do is we help sort of bridge that gap
0:43 of having a single record of all those
0:45 interactions with each customer mm-hmm
0:46 because previously companies would build
0:49 all this in-house or not at all maybe
0:50 yeah there's there's sort of two worlds
0:52 one is they would build it all in-house
0:54 exactly I'm gonna be a rat's nest of
0:55 data pipelines from one place to another
0:57 and so the engineering team would spend
0:58 all their time building these data
0:59 pipelines rather than actually building
1:01 things for the customer that's one world
1:03 the other world is one where it really
1:05 used to be a one-on-one relationship
1:07 with a bank branch manager for example
1:10 and they might keep the information in a
1:14 CRM mmhmm right but if you are similar
1:16 would be an optometrist right you go in
1:18 they have your past orders etc but now
1:20 the world is moving much more to a Warby
1:22 Parker kind of world where you're not
1:24 interacting with a person so a CRM
1:25 doesn't even make sense it's like not
1:27 the right technology for understanding
1:28 what interactions are having with a
1:29 customer and instead it's all these
1:31 different digital channels so that's
1:32 where we come in and your first
1:33 customers were they large scale
1:36 companies like banks or who did you get
1:37 in the beginning at the very beginning
1:39 we actually launched as an open source
1:42 library on Hacker News yeah and it took
1:46 off there blew up basically overnight so
1:49 to be fair it was kind of like the long
1:52 road that's not a year and a half of
1:54 dark times yeah yeah we shouldn't we
1:55 shouldn't totally go over that but okay
1:57 yeah so anyway you you blew up overnight
2:00 again once it once it went live on a
2:03 Chen it blew up overnight and so our
2:04 first customers were the folks hanging
2:08 out on on hacker news it was basically
2:09 small companies for the most part
2:10 founders who were looking for better
2:11 ways to insta
2:13 their web applications and mobile apps
2:15 mm-hmm with this sort of analytics
2:17 tracking and so the initial growth was
2:20 that was just completely unpaid
2:21 customers right so they're just using
2:23 the open source whatever you put up on
2:25 github right so the strange thing about
2:27 the open source library which is sort of
2:29 this data router so you put in one piece
2:31 of data customer did X and then we turn
2:32 around transform it and fan it out tall
2:34 it from places downstream and what's
2:36 funny about the library is if you want
2:37 to turn on a new tool if you want to
2:39 send the data to a new place you need to
2:40 recompile the library and redeploy it to
2:43 your website so it doesn't actually the
2:44 open source library doesn't quite
2:46 actually solve the problem which is
2:47 really a market or a product manager
2:50 wants to use a new tool so what happens
2:51 is you really want to use the hosted
2:53 version yeah so almost no one actually
2:56 uses the open-source version as was
2:58 addressed by design no no this is
2:59 completely accidental really yeah
3:00 there's just no way to make a good
3:02 open-source product around it yeah
3:05 that's crazy so okay so what did you
3:08 apply to icy with we actually applied as
3:11 a classroom lecture tool so the idea was
3:12 to give students this button to push to
3:15 say I'm confused and the professor would
3:17 get this graph over time of how confused
3:19 the students were we thought it was a
3:20 really cool idea we were college
3:23 students at the time and we had a bunch
3:24 of professors who were excited about it
3:26 at MIT and elsewhere I'll never forget
3:29 our in our YC interview we were pitching
3:31 this and PG was getting pretty excited
3:35 and then he turns to professor Miller
3:38 from from MIT who we had talked to and
3:39 had given us the original idea for it
3:42 and says hey would you use this and
3:49 we just rolled with the punches and said
3:51 yeah well you know we talked like twenty
3:52 other professors and and they were all
3:54 excited about it oh man but then you
3:56 went through YC with this all right yeah
3:57 building it out right we went through
4:00 the whole whole YC with this idea built
4:01 it out hundreds of thousands lines of
4:03 code super complicated class or
4:04 electrical product to have like
4:06 presentation view and people could ask
4:07 questions it's very complicated
4:09 we were actually even raised money at
4:11 demo day with this idea about 600k
4:13 finally we deployed it in the classroom
4:15 as the fall semester got started after
4:17 demo day it was just a total disaster
4:19 all students are my laptops and then I
4:22 went straight to Facebook so the way we
4:23 discovered this is we were standing in
4:24 the back of the classroom
4:26 counting laptop screens it'd be looking
4:27 over the shoulders of the students near
4:30 one two three and yeah we discovered at
4:31 the beginning of class about 60% of
4:33 students were on Facebook and by the end
4:35 about 80 percent were on Facebook oh me
4:37 in other words they were supposed to be
4:39 using your desktop app the whole time
4:41 that's right in the professor at the
4:42 beginning class had been like can
4:43 everyone please get out their laptops
4:45 we're going to use this mute and all of
4:46 a sudden all these students are
4:47 distracted by Facebook so we had
4:49 accidentally sort of put in an attention
4:52 attention bomb if you want yeah and you
4:55 hadn't brought on any users during why
4:56 so you were in the suit wait I wrote it
4:58 you were in the summer 2011 bat that's
5:00 right summer 2011 and you weren't
5:03 testing during the batch we tried
5:04 testing but there's not that many
5:05 classrooms that are in session during
5:08 the summer right the school year starts
5:10 in September so we had beta tested in a
5:12 few summer computer science classes at
5:14 both Stanford and Berkeley yeah but
5:16 there was always technical difficulties
5:17 and other things that sort of prevented
5:18 us from getting like a real sense of
5:20 what was happening often times it's
5:22 pretty Yolo I don't I just we didn't
5:24 really get tests rolling until right
5:25 until the fall okay
5:27 and so what was the like the come to
5:29 Jesus moment where you realize you had
5:31 to change the product standing in the
5:33 back of a BU classroom as an
5:35 anthropology class and I remember
5:38 arriving at the 60% and the 80% and we
5:40 went up and apologized to the professor
5:43 and and at that point you're just like
5:45 okay we got to kind of shut this down or
5:47 figure out what to do with the money yes
5:49 right well we had we had just gotten
5:51 wires for these for these checks for
5:52 this money like literally a week before
5:55 right so we called back all the
5:56 investors and we were like well it turns
5:58 out this is a terrible idea so what do
6:00 you want us to do with the money and
6:02 almost all of them said well you know we
6:03 invest it for the team so so go find
6:06 something else okay um okay next day
6:08 next day we're like well what are we
6:11 gonna do and we realized we should have
6:12 been able to figure out some of this
6:14 analysis by not just standing in the
6:15 back of the classroom like we should've
6:16 been able to see some of this digital
6:18 you couldn't see it in the analytics
6:21 metrics at all so we decided hey let's
6:22 build an analytics tool let's build a
6:25 better web analytics mobile web app
6:26 analytics tool to compete with Mixpanel
6:30 and Google Analytics the idea was to
6:31 give really advanced segmentation
6:33 because we also wanted to understand how
6:35 some computer science classrooms at MIT
6:36 we're using it differently than
6:38 anthropology classes would be
6:40 and we couldn't do that analysis and the
6:42 tools we had so that was the idea was to
6:44 build an analytic tool mm-hmm we spent
6:45 about a year building the infrastructure
6:49 necessary to do the analytics and really
6:51 we're not succeeding in getting any
6:53 customers during that timeframe so you
6:55 had I mean did you have a whole
6:57 onboarding process was it like a landing
6:59 page what did it look like oh yeah we
7:00 had a landing page I was going on little
7:02 sales trips I was meeting with people
7:04 tell you so you were trying oh yeah we
7:07 were trying okay but it was it was not
7:09 going well basically what happened is
7:10 people would say well I already have
7:11 this other analytics tool installed so
7:14 like it's not that interested you
7:16 definitely you have the time you were
7:19 not 10x better than Google no no yeah
7:22 yeah were you at parity we were at
7:25 parity in certain dimensions and we were
7:27 exceeded in other dimensions but they
7:28 just weren't them dimensions that
7:31 mattered apparently right okay so you're
7:33 just kind of like blow up overnight with
7:36 accra news and that like this is well so
7:38 positive we're not even this is just the
7:39 analytics tool which we realized in
7:42 December 2012 was failed so we're like
7:43 well over a year into the analytics tool
7:46 we're now have tried two ideas both have
7:48 failed neither is clearly gonna work at
7:50 all yeah we realized that we're screwing
7:52 up so we decide to have office hours
7:55 with YC again I would come back walk
7:56 around a little cul-de-sac by YC with
8:00 with PG he started comes ta stop and
8:01 says so just to be clear you've spent
8:02 half a million dollars and you've
8:06 nothing to show for it gulping moment
8:10 yeah I guess that's true and it was a
8:12 good sort of come-to-jesus moment though
8:15 and you got a pause there and then
8:17 rewind all the way back to the first
8:20 week of YC and it was in that first week
8:21 that we'd been like well we should have
8:23 analytics tools on our classroom extra
8:26 tool say we've googled analytics and we
8:27 found Google Analytics Mixpanel and
8:29 KISSmetrics and we're looking at them
8:30 we're like we don't know which one of
8:32 these things we should use all right
8:33 like they're kind of all similar but
8:34 Google Analytics is a little more
8:35 marketing a KISSmetrics is a little more
8:37 revenue we mix panels a little more
8:38 product II in terms of the sorts of
8:42 insights they can give you and at the
8:44 end of the day that they all collect the
8:45 same data they all collect
8:47 basically who is the customer and what
8:49 are they doing mm-hmm so we decided to
8:51 write this little tiny abstraction that could
8:52 could
8:54 data to all three I was like 50 lines of
8:55 code among the hundreds of thousands for
8:57 this classroom lecture tool and then we
8:58 decided we'll just send it at all three
8:59 we'll look at all the tools and we'll
9:01 just pick the one that we need so it'll
9:02 give us a lot of option allottee
9:04 basically for free yep and then we
9:06 forgot about it for like four months
9:07 so months later we clean it up a little
9:09 bit more four months later it cleans up
9:10 a little bit more as I mentioned now
9:12 we're struggling at that point we're
9:13 struggling with this question of why I
9:14 already have mixed panel installed so I
9:15 don't really want to use your analytics
9:19 tool yeah so my co-founder Ilya has this
9:20 idea he's like remember that little
9:21 library wrote that sort of abstracted
9:23 away the differences between these tools
9:24 and what a Sun data to all three what if
9:25 we added ourselves just the fourth
9:27 service that I could send data to you
9:29 and then every time someone has that
9:30 objection we hit him back with an open
9:32 source library that they can use to send
9:34 to us and them that cause seems like a
9:35 clever growth hack it like gets us
9:37 around this problem so we did that
9:40 cleaned it up open sourced it and people
9:41 started replying like all this libraries
9:43 great we'd love to use it couple weeks
9:45 later we'd follow up and be like hey
9:46 well we saw you're using you know the
9:49 open source library but all you have to
9:51 do is copy paste our API key so that you
9:52 can use our analytics tool could you
9:54 just copy paste the API key yeah no like
9:58 yeah so we started just feeling like
10:00 there's some traction on on this little
10:02 routing library which is maybe up to
10:04 like 25 stars on github or something
10:06 like that like not not much but some
10:08 people are issuing pull requests and
10:09 it's the first time we've ever faults
10:12 are like pull like it if you always we
10:13 weren't just pushing a boulder uphill
10:16 this is a little different but subtle we
10:18 had this conversation with Paul Graham
10:20 the next day we sit down this is our
10:22 second time with this we're like okay we
10:23 have 100k left in the bank what's our
10:26 final shot and my co-founder Ian is like
10:28 you know what I think there's a big
10:30 business behind analytics j/s which is
10:33 this routing library like that is
10:34 literally the worst idea I've ever heard
10:37 like it's 500 lines of code to grow a
10:38 little bit's 500 lines of code in it's
10:42 already open source I have no idea how
10:44 you build a business around that yeah
10:47 and so we fought about it all day long I
10:50 was four of us I was the most skeptical
10:51 I think just literally like brutal
10:53 brutal fighting I went home and I was
10:53 trying to figure out how to kill the
10:56 idea was awake half the night finally
10:57 figured it out came in the next day I
10:58 was like alright guys here's we're gonna
11:00 do we're gonna build like a beautiful
11:01 landing page really gonna pitch the
11:03 value of this analytics j/s open-source
11:05 library and it'll have
11:07 a form at the bottom so that we can get
11:09 people to sort of express interest put
11:10 it up on Hacker News and we'll see what
11:12 happens I was thinking like totally kill
11:14 it right I think that's well in hacker
11:17 news right yeah so we'd build a landing
11:18 page put it up on Hacker News and this
11:20 is where we have this you know you're in
11:21 half in the making overnight explosion
11:24 yeah and that that kind of segues into
11:27 your whole startup school talk right
11:30 about legs right basically real product
11:33 market fit yeah and up to that point in
11:35 your life had you launched anything
11:37 where the market I mean this is a Marc
11:39 Andreessen quote because he kind of
11:41 coined the term it pulls it out of you
11:43 right that's right I think you
11:45 experiencing hmm I think that's an apt
11:46 description I had never experienced it
11:48 before and it feels very much like
11:50 losing control
11:52 right like previously you're like
11:53 building a thing and you roll it out in
11:55 a building a thing you're pushing it out
11:56 and all of a sudden you like put a thing
11:58 out there and people start running away
11:59 with it and using it in ways that you
12:01 didn't necessarily expect and you're
12:03 sort of like what it's just an it's just
12:05 a man like stop stop it like we need to
12:06 fix these other things because otherwise
12:08 it's like this feeling of losing control
12:11 and almost like the market is dictating
12:13 to you now what the rules of the road
12:15 are and what needs to get built interest
12:16 so would you're a different feeling yeah
12:18 would you differentiate that from
12:20 overwhelming demand for one particular
12:24 feature versus like we're just gonna
12:26 take this and use it however we want but
12:28 there's a ton of demand there would you
12:33 separate those two things not really I
12:35 think the people always want more
12:38 features yeah but the the thing that
12:40 flipped was people would previously tell
12:43 us they wanted a feature but not use it
12:44 whereas now people were using it and
12:46 they would want a second feature and the
12:48 it's a super important distinction I
12:49 think a lot of founders get caught in
12:51 this sort of I call the the death spiral
12:53 of user feedback yeah where they keep
12:55 going and showing someone their product
12:56 and asking them for feedback they give
12:58 them you know some feedback about how
12:59 they could make it better but they don't
13:01 use it and then they bring it back with
13:02 those fixes and they ask if this is
13:04 better and it's just like the death
13:05 spiral where it never gets anywhere but
13:06 once someone starts using it they'll
13:07 have more requests and that just means
13:08 they're gonna pay you more over time
13:10 right yeah I like how you put it in the
13:12 lecture where you basically if you have
13:14 to ask yourself it's not product market fit
13:14 fit
13:16 yeah it's it's you really can't miss it you
13:17 you
13:21 and said this was now six years ago five
13:23 five or six almost almost six years ago
13:25 yeah right and half you and you're so
13:27 feeling the same way yeah yeah and since
13:29 then we've had a few more sore a
13:30 secondary product market fit moments
13:35 like what about two years in we
13:36 discovered that all of our most valuable
13:39 customers were sending their data to an
13:41 s3 bucket there's basically they're
13:44 keeping log files of the raw data which
13:45 was a little weird because typically
13:47 what people were using the data for was
13:48 to send to an analytics tool an email
13:49 marketing tool and a CRM and help desk
13:51 like places where a business person is
13:53 deriving value log files is a little
13:55 different that's a little weird it's
13:57 unclear what the use case is so we went
13:59 on this sales trip to New York myself
14:03 and our first sales person RAF we met
14:04 with five customers that were using this
14:08 s3 bucket let me just ask them oh my god
14:09 what are you doing with the s3 bucket
14:11 the first customer was like well you
14:13 know we have a data engineering team
14:14 that's taking the data out of the bucket
14:17 and converting it into CSV files and
14:18 then they're uploading it to our data
14:20 warehouse which is a redshift cluster so
14:21 basically they were using it as the
14:23 initial endpoint input into a ETL
14:26 pipeline like oh that's interesting but
14:28 man to the next meeting second customer
14:30 is like well we have a data engineering
14:31 team who's taking the data out of s3
14:35 converting it miss yes okay drops and
14:38 that's interesting and then the third
14:39 fourth and fifth ones all said exactly
14:41 the same thing and and that was the
14:42 point at which I started becoming a
14:44 conspiracy theorist yeah it seemed it
14:45 seemed like some pre meeting had
14:47 happened now but they were all doing
14:48 exactly same thing so it was really
14:52 obvious we just built a way to load data
14:53 directly from segments into a regio
14:56 fluster huh and that was a huge thing
14:58 like you you was really go out that it
15:00 was product market fit again yeah it was
15:01 very explosive
15:03 so we you know we'd grown revenue from
15:05 zero to two and a half million in the
15:09 first year and then we launched this red
15:11 shift connector and the next year one
15:13 from two and a half to ten but people
15:16 weren't asking for that that's right it
15:18 was one step too far for them to realize
15:20 that we could do it easily yeah their
15:22 mentality I think was that oh segment is
15:23 a way that I integrate marketing tools
15:25 and so a data warehouse is in a
15:28 marketing tool it's a bi tool surely
15:29 segments can integrate that just didn't click
15:29 click
15:30 ahead to go
15:33 by asking hmmm interesting and and your
15:36 growth how how how does that happen is
15:40 it has it come through developers I go
15:41 to market model is primarily through
15:45 through engineers yeah we talk a lot
15:47 about sort of the way that we've built
15:49 our infrastructure over Jim
15:50 we obviously process a lot of data so
15:51 there's a lot of interesting
15:53 infrastructure problems I think now
15:54 we're processing you know hundreds of
15:57 thousands of user actions per second so
15:58 there's a lot of data going through
16:00 there we write about that
16:02 that's generally interesting to that to
16:03 the hacker news and an engineering crowd
16:05 and yeah typically an engineer is the
16:07 one that brings this brings us in
16:09 sometimes that really technical product
16:11 manager but it's someone who's like yeah
16:12 this is gonna solve this weird rat's
16:14 nest data pipeline problem that I've got
16:16 well and how much of it is open source
16:19 still a good portion of it is is open
16:21 source but most of the value that we
16:22 deliver is actually by running the
16:24 hosted version right cuz it's like at
16:26 the end of the day it's not just the
16:27 develop like you're you're saving the
16:29 developers time but it's these business
16:32 people that really need it yeah and and
16:34 frankly most of the complexity is hidden
16:35 away and how you actually operate and
16:37 scale a data pipeline that that is
16:40 processing the data yeah so you know our
16:41 JavaScript libraries open source our iOS
16:43 SDK is open source or Android SDKs open
16:45 source II wouldn't sort of collected
16:47 data from anywhere and those collection
16:49 libraries are our open source but the
16:51 sort of core infrastructure pipeline is
16:54 not hmm okay and so right before you
16:56 guys launched on HN where did this like
16:59 small tiny micro launch I whatever it
16:59 might be
17:02 were there other avenues that you were
17:05 considering pursuing like in that debate
17:07 in the day before mmm were you thinking
17:11 about other stuff I think the debate was
17:12 whether to build out the full product
17:15 and then test for product market fit by
17:17 trying to sell it to people versus this
17:20 super super lightweightt MVP landing
17:21 page that we wouldn't put on Hacker News
17:23 to see if there was interest in the
17:26 concept yeah and and what drove us
17:27 towards the super super lightweightt
17:29 test was actually the fact that there
17:31 was a skeptical divide among the
17:34 founders and since the founders couldn't
17:36 agree the only way to answer the
17:39 question was to go to customers a SAP
17:42 and get an answer okay it's tough like
17:44 the the launching early thing is always a
17:44 a
17:46 because I think there have been
17:48 instances where people are like I will
17:49 just launch this early but because
17:51 they're like 10% off of what that
17:53 product ought to be or they're not very
17:55 good at communicating it they never
17:57 really get the feedback that they need
17:59 right like how do you how do you kind of
18:01 balance that out like this is kind of
18:02 like fully formed enough or were
18:05 communicating it clearly enough that we
18:06 can launch it like how do you determine
18:11 that usually that test is so cheap to
18:14 run that it's worth running even if you
18:17 decide that it was inconclusive and you
18:19 should go a step deeper mm-hmm but I
18:21 also think that the way frame fit is
18:22 actually an excuse that a lot of
18:25 founders use for not doing the cheap
18:27 early tests when in fact they they
18:29 should yeah I I get the the same
18:31 questions like almost every single
18:34 podcast and I should be a little bit of
18:36 a devil's advocate here but yeah these
18:38 are kind of like straw man arguments
18:40 right I think the really big product
18:42 market for moments for every company are
18:44 pretty unmistakable like the Dropbox
18:45 founders have called a stepping on a
18:48 landmine I just I really don't think you
18:51 can mistake it it really happens in a
18:52 way that you you lose control it's very
18:54 obvious every much metric goes haywire
18:57 people are talking about it a lot it's
19:01 not it's not mistakable or like oh well
19:02 you know this person said that it looked
19:03 valuable and was really exciting and
19:05 blah blah blah if they're not using it
19:09 like it's not there yeah okay well okay
19:11 so there's a question from Twitter it's
19:12 clear that a bunch of people have
19:15 watched your your lecture so this first
19:18 one from Aaron Evan Farrell he asked you
19:19 mentioned in the startup school lecture
19:22 that you had to pivot to analytics j/s
19:26 to find Prada market is it possible to
19:28 purely iterate on something like that to
19:30 find product market fit or should be
19:32 clear from the outset if a new idea is
19:34 something people want there's two
19:36 versions of this I will say the Airbnb
19:38 version of product market fit is much
19:40 more iterative they struggled for years
19:43 and years and made slight iterations and
19:44 iteration that iterations and finally I
19:45 caught on and obviously they're they're
19:50 their runaway success my feeling is that
19:54 that's extremely rare and that again
19:57 this is a really dangerous place to be
19:59 because you can stay in this iterative
20:02 mode for years yeah and it is unlikely
20:04 that the iterations are going to get you
20:05 to a good place
20:07 so I remember very clearly early on
20:09 being really inspired by the Airbnb
20:12 story and it being a logical reason why
20:13 we should keep plugging away at a bad
20:16 idea and I think we abused the Airbnb
20:18 story to just keep stringing ourselves
20:20 along on a bad idea so I would be very
20:22 very careful of following their B&B
20:23 example I don't know many other
20:25 companies that that hit proud of market
20:27 fit that way right and so how long do
20:29 you give an idea at this point I
20:32 actually don't think that it's quite the
20:33 right frame to think about it in terms
20:35 of how long to give an idea okay I think
20:37 what you want is someone either yourself
20:38 or someone else on the founding team
20:42 who's a skeptic so someone who is going
20:45 to have enough context with whatever the
20:47 specific idea is and whatever the sort
20:51 of regime or market you're in someone
20:54 who's skeptical who will question and
20:57 push for the fastest reasonable test so
20:58 in other words if you have an optimist
20:59 and a skeptic and they both agree on
21:02 what a valid test is then I think you
21:03 actually lend up with a good test but if
21:04 you have three optimists in a room who
21:06 all agree on what a good test is I don't
21:08 believe that that's a good test did over
21:10 skeptics but yeah do you have you
21:12 recruited skeptics or did you just kind
21:13 of like luck into that I think we lucked
21:17 into it the first time yeah I do think
21:19 we have some folks on the team that
21:22 segment some early folks who are who are
21:25 skeptics usually about future product
21:26 market fit moments that we've had and I
21:27 think it's been enormously helpful
21:29 that's really interesting what how do
21:31 you test for that like in an interview
21:33 scenario we didn't test for we just got
21:34 lucky again but yeah okay so not even
21:36 just with co-founders like with early
21:38 employees as well that's right yeah
21:40 how interesting like how hurtful can it
21:41 be if someone is like well I really
21:42 think you haven't thought this through
21:44 there's like these three things that you
21:46 should really test ASAP because I don't
21:48 really believe that you have product
21:50 market fit here right that's what you
21:52 want you want someone who's gonna be
21:54 like pushing it and you're like what happen
21:54 happen
21:56 yeah how would we and then who's willing
21:57 to collaborate with you on how you
21:58 should reasonably test whether those
22:00 things are the case so the sorts of
22:03 tests that we have run for example with
22:05 this mindset recently and even in the
22:06 past year yeah where should we switch
22:08 from a technical buyer to a marketing buyer
22:13 unclear how to test that well so the
22:15 this early skeptic who's amazing her
22:18 name is Diana she I was like well I'm
22:19 just gonna go to a conference with
22:20 marketers and I would try pitching a
22:21 bunch of marketers just flew to Florida
22:23 and pitched a bunch of marketers came
22:26 back she's like nope not a good idea and
22:29 I read my own set of tests so like the
22:31 the hacky ways to test these things I
22:32 think are very valuable in it and it it
22:34 comes from having skeptics who and
22:35 different perspectives of people willing
22:37 to go test those things okay
22:39 and so I imagine these these tests from
22:42 skeptics occur on a maybe daily but
22:44 probably like at least a monthly basis
22:46 right in terms of you guys working on
22:48 your product yeah I'd say maybe more on
22:50 like a pre idea basis so like if we're
22:51 gonna launch a new product then it's
22:53 really helpful to have a skeptical
22:54 perspective of like here's why this
22:56 might not actually be a good idea and do
22:59 you rely more heavily on data or actual
23:01 customer interaction like in the early
23:02 part of the product development process
23:04 it's all qualitative it's all talking
23:06 with customers okay because this is the
23:08 thing that bugs me more is like when
23:09 people are just like putting up landing
23:11 pages left and right and like thinking
23:14 that they can like kind of uh I forget
23:17 what it's I will call it like what do
23:18 they do it doesn't know yet like
23:20 Imagineer your way towards them like
23:24 winding down this path to finding it in
23:26 an advanced inefficient with time yeah
23:30 they're scared yeah and and when you
23:31 actually go and talk to a customer if
23:33 you have that conversation in the right
23:35 way you'll learn a thousand times more
23:36 from that conversation than you will
23:39 from them from putting up a landing page
23:42 yeah and I think ultimately we learned a
23:44 lot more from talking to our customers
23:45 after the hacker news landing page than
23:47 we did from the landing page itself yeah
23:49 totally so what are your what are your
23:50 tactics when you're talking to customers
23:53 yeah I'd say the main thing is most
23:55 founders are not familiar with how a
23:58 sales process is actually run and you
24:00 basically want to run a sales process so
24:02 the the sort of typical founder motion
24:04 with running a sales process is the they
24:05 come in and they say okay I'm going to
24:07 give you a demo and it's like really
24:09 shiny polished pitch and then the
24:10 customer decides at the end of that
24:12 pitch whether they're interested or not
24:14 that's not actually how good sales works
24:17 at all the way good sales works is you
24:18 do qualification up front
24:20 so you have some method
24:22 of understanding the customers problem
24:23 better than they understand it
24:24 themselves and then you do the
24:25 computation in your head as to whether
24:27 your product is a fit for their problem
24:30 there's a lot of methodologies for this
24:32 the methodology that we use that segment
24:36 is one called medic ME ddic it's
24:37 literally just a list of sales
24:39 qualification criteria and this is this
24:40 is what sales reps do if a sales rep
24:41 comes back and we're like we're gonna
24:44 close you know this this this deal the
24:46 sales manager says okay well let's go
24:48 through and metrics what are the metrics
24:49 by which this company is going to judge
24:50 whether or not the product works for
24:52 them and if the sales rep can't answer
24:54 metrics economic by our decision-maker
24:58 decision process the identified pain and
24:59 doesn't have a champion if they can't
25:01 have all six of those things yeah
25:03 there's no deal and so when you're
25:04 searching for product market fit you can
25:06 just go through all of those things by
25:07 asking the customer a ton of questions
25:09 mm-hmm and then you can grade whether or
25:10 not you're actually going to build a
25:11 product that will solve the problem
25:14 right well this is it kind of ties into
25:16 this like skeptic versus like optimist
25:18 idea right you have someone who's like a
25:21 champion of the product and in many ways
25:22 I mean maybe this is you like the
25:24 optimist who just sees the world and
25:26 they see this future and it looks
25:28 awesome it's amazing but you need that
25:29 skeptic who sees the world as it really
25:31 is that's right and a sales
25:33 qualification criteria is is a way of
25:34 almost putting the skeptic out as they
25:36 like structured process that enforces
25:39 some level of skepticism yeah I think
25:41 it's so dangerous when you're the author
25:43 because I fall into this camp for the
25:46 most part like when you get good at
25:48 sales you can kind of sell many people
25:52 on almost anything but if that product
25:55 doesn't exist yet it's very easy to just
25:56 kind of mold it in the way when you're
25:58 reading someone you're like oh I can
26:00 totally kind of want it to be like this
26:02 so I'm gonna kind of go down this path
26:04 but then when you actually show them the
26:06 product and they're like like you said
26:08 they won't even install it then you see
26:09 the world that's it really is
26:11 yep and that's the thing and so we you
26:13 guys are just like going out and how are
26:14 you still having these conversations
26:17 with people personally sometimes yeah
26:22 for sure yeah yeah because this is one
26:24 of the things that like people I think
26:26 in large part because they're influenced
26:28 by your startup school talk they have so
26:29 many questions about it and so Benjamin
26:32 Liam asked like you know how do you even
26:34 how do you find that you have the right
26:36 messaging around you're explaining your
26:38 product oh man this is super hard I'm
26:39 not even the right person to ask about this
26:40 this
26:41 I should Diana who I mentioned before in
26:44 our VP marketing Holly are the two
26:45 people who have really refined our
26:48 messaging over the years and we're
26:51 always trying to refine it so I don't
26:53 know how you know that you have the
26:56 right messaging you know that whatever
26:57 messaging you can sort of test whether
26:59 alternate messaging is going to work and
27:01 you can do that yeah qualitatively in
27:03 interviews with customers you can try
27:04 explaining it one way and see if their
27:05 eyes light up you can try explaining
27:07 another way and just sort of see what
27:09 what resonates I think a really talented
27:13 early salesperson will also have this
27:17 sort of pattern in their in their habit
27:19 of how they pitch that they'll always be
27:20 testing different ways of explaining the
27:22 product that was definitely true for for
27:25 the first salesperson that we hired he
27:26 was fabulous it just like constantly
27:27 experimenting with different ways of
27:28 doing it so I don't know if you there's
27:30 you never know if you have the best
27:33 messaging but you are constantly
27:35 searching and testing for four different
27:37 ways of explaining it okay but but again
27:41 like if it if it's about you know really
27:43 finding a good product market fit yep do
27:45 you think that leave these like minor
27:46 changes and how you communicate
27:47 something will will make the difference
27:49 I don't think I don't think minor
27:50 changes then it will make the difference
27:52 now once you have product market fit
27:53 then sure you can optimize the messaging
27:56 okay so then we should talk about idea
27:57 generation because that seems more
27:59 important yeah and these like minor
28:03 deviations yep weird where do you begin
28:06 yeah I think the best ideas that we've
28:08 had come so there's a big difference
28:12 between the first idea and these sort of
28:15 like follow-on ideas and and the reason
28:16 so the first idea meaning like the core
28:18 product that's and then the individual
28:20 features that's right okay and not just
28:21 individual features you might have
28:23 entirely new products that come along
28:27 but those are much easier right the the
28:30 problem with the first product and
28:31 product market fit is that you can move
28:33 the product and you can move the market
28:34 because it's fit between these two
28:37 things and so it's unclear and they move
28:39 in some crazy multi-dimensional space
28:41 and so what's the the issue is that to
28:45 get them to both match up you can always
28:47 move either one and in different conversations
28:47 conversations
28:49 in one conversation you might shift the
28:50 product and you might in a different
28:51 conversation realize you need to shift
28:54 the market so that's super tricky I
28:56 don't think there's a repeatable way to
28:59 do that mmm I think you just have to go
29:01 very very deep into a particular market
29:03 and understand the problems that people
29:05 have in that market so do you have a
29:08 particular a process for idea generation
29:11 or you just you get into something and
29:13 you're like man just go super deep yeah
29:15 for that first idea you just have to go
29:16 super deep you just have to understand
29:18 the market and the ecosystem and the
29:20 customers upside down backwards better
29:23 than they do themselves okay so you were
29:26 booted from segment today do you know
29:28 where you would start you'd have to
29:29 start with something that was
29:30 interesting to you personally and then
29:33 you'd go dig in in a deep direction I
29:35 think that it becomes more repeatable
29:38 when you are finding a second product so
29:40 at that point you've mostly lost the
29:41 market side right because you already
29:42 have a buyer you already have a go to
29:44 market motion you already have like an
29:46 area of interest which for us you know
29:48 is these sort of data pipelines and data
29:49 infrastructure customer data
29:51 infrastructure then it's much easier
29:54 because you know exactly who you need to
29:56 go to and you know roughly the like type
29:57 of questions that you need to ask and
30:01 then you can run a process which is much
30:03 a much deeper x-ray of the customer than
30:05 you might be comfortable with at least
30:07 it was much deeper than then I was
30:09 comfortable with when I first got
30:12 started like as an example we recently
30:13 were testing product market fit for our
30:15 product we're gonna announce at our user
30:18 conference in September and that's now
30:20 in beta and it's doing really well but
30:22 the initial way that we were sort of
30:23 testing fit there
30:25 we would go in and say like oh hey do
30:27 you have you know a problem with with
30:30 data cleanliness and the person would be
30:31 like oh yeah yeah totally that's one of
30:33 our big problems yeah I get cool cool
30:38 like okay - that is and we might ask
30:39 like two or three more questions but
30:41 like that was sort of the depth but the
30:42 actual level of question needs to be
30:44 like okay well like what do you mean
30:45 Didache like how do you do you currently
30:47 invest in data cleanliness at all no
30:49 like oh well yeah we actually you know
30:51 we have a team of like six people who do
30:54 data QA con all the time like oh well
30:56 those data QA people like where are they
30:57 based or they're based in LA Oh
31:00 interesting so they have like real
31:01 salaries and their knobs
31:03 overseas they're like real like us
31:04 salaries like yeah yeah okay so what
31:07 like 80k a year 100k yeah yeah yeah
31:08 that's about right so like okay so
31:09 you're spending you know like 750 K a
31:12 year for this data QA team and like tell
31:13 me more about that process like what are
31:15 they queueing exactly oh well they're
31:16 clicking this button in the app and
31:18 they're like well which button yeah and
31:20 then are they like in the OP where they
31:22 do when they find a bug and so we would
31:23 ask like literally 45 minutes of
31:26 questions like this and now we actually
31:27 understand their problem and we
31:28 understand what they're doing we
31:30 understand where their cost centers are
31:30 we understand how this thing and then
31:31 we're like oh well what if we did a
31:34 product that did X yeah which is exactly
31:35 what they just explained to us for the
31:36 previous 45 minutes and they're like
31:39 that would be amazing they're like okay
31:41 now wait now this is that was both sales
31:44 qualification and discovery yeah which
31:45 is a standard sales process but now it's
31:47 being used for product development and
31:48 that's such a good learning because
31:51 people aren't gonna tell you no way I
31:52 think a lot of people just get scared to
31:54 like ask these questions totally the
31:56 customers will tell you yeah especially
31:59 if you if you take the champion part of
32:01 medic their last one the C and you just
32:03 start by asking like what's your vision
32:05 for X thing that you do John will tell
32:07 you you're like oh we our mission is
32:08 similar because that's why you got the
32:10 meeting in the first place that person
32:11 is instantly aligned with you they'll
32:13 talk for 45 minutes about their problems
32:14 before you have to tell them anything
32:16 yeah I think that's one of the things
32:18 that most people don't realize it like
32:20 many of the best salespeople don't talk
32:22 that much the best salespeople at
32:24 segment ask why to the point of
32:26 uncomfortableness from everyone else on
32:28 the team including myself yeah
32:30 interesting yeah I wonder what the
32:31 correlation is between sales and
32:34 skepticism it's probably pretty high and
32:36 people who are questioning things and I
32:40 can see the angle yep hmm all right next
32:43 Twitter question so Danny Pro first of
32:46 all he says go Peter and his question is
32:49 about a culture so he says what values
32:51 and standards do you have in place for
32:52 your team at segments and how do you
32:54 actively build that culture into your
32:57 company yes we have four values that
32:59 segment that we're quite dedicated to
33:03 the first is karma which is we want to
33:04 have a positive impact on the world and
33:06 that manifests itself in a bunch of ways
33:10 one of those ways is we really care
33:12 about the customer having sort of
33:14 getting value out of our entire process
33:16 so you'll notice that all of our
33:17 marketing materials for example are
33:19 often like highly educational we have a
33:22 really high bar for what an education
33:25 piece of educational material looks like
33:27 even in the sales process we want to be
33:29 helpful if we're not the right fit will
33:30 tell you and sort of like refer you to
33:32 the right places
33:36 separately we really care about doing
33:39 the right thing by the end-user this is
33:41 still within karma and that's from like
33:42 a data privacy perspective so we're very
33:43 interested in helping companies
33:45 understand all of their own first-party
33:46 data so all their interactions with
33:48 their own customers within their four
33:50 walls we're super uninterested in
33:51 helping companies data broker data
33:53 between different companies schedule a
33:55 we call it data gossip it's gross we
33:56 don't want anything to do with it there
33:58 are plenty of other companies out there
34:00 that that have stuff like that it's
34:01 gonna go away and die eventually
34:04 so that's Karma we care a lot about that
34:07 the second one is tribe which is a
34:09 segment we're all there to support each
34:10 other we're all there to accomplish the
34:14 same thing and so what we expect is that
34:16 and what we value is that people really
34:19 support each other both when they may be
34:21 struggling with something but also
34:23 giving them giving them crit so really
34:25 try to reward folks who are who are
34:26 willing to go the extra mile to give
34:28 create when it may be hard could be
34:31 giving crit across teams or up several
34:33 levels or whatever that's that's really
34:36 something that we value hmm the third is
34:39 Drive much more self apparent we like to
34:41 get done we value people who are
34:42 getting done and the fourth is
34:44 focus which is not just sort of the
34:46 ability to sit down and get stuff done
34:48 but more power through something but
34:50 actually thinking carefully about
34:52 prioritization we've done a lot of
34:54 research around how to make the office
34:56 environment you can actually focus so
34:58 check out our blog we've we've written
35:00 about sort of sound decibel levels that
35:01 we've measured around the office and how
35:02 we've mitigated that did pretty well
35:04 that piece did pretty well read yeah and
35:06 and it was a surprising result for us to
35:07 discover the different parts of the
35:09 office had very different sound levels
35:10 that were not correlated with people
35:12 talking but were just correlated with
35:14 the sort of acoustic shape of the of the
35:16 office and so just moving people around
35:18 into different places helped a lot
35:19 depending on how much noise they were
35:21 willing to tolerate and sort of need it
35:24 in their role hmm so anyway those are
35:27 those four values we they are literally
35:28 the things we value and so we pushed
35:30 into all the places where you would
35:32 expect what you value to have an impact
35:33 so it's who gets highlighted at all
35:36 hands we have a citrus prize which is
35:38 someone who's living all the values
35:40 promotions hiring we have a strict
35:43 interview in the hiring process we have
35:45 performance we need a strict interview
35:47 sorry we have a culture interview where
35:48 we have these four values and waste
35:49 specific ways that we're going to test
35:53 for them when we run performance reviews
35:55 the performance of you is literally the
35:57 four values are usually like these this
35:58 is what we value and therefore it's what
36:00 we test and measure by and I think
36:03 ultimately it's that cycle of giving
36:04 feedback and measuring by it that is
36:07 what drives huh culture to stick and it
36:09 has this been something that like came
36:11 natural to you like building culture or
36:12 did you have to learn it I don't think
36:17 so I think we I think we learned it mmm
36:20 we got to about 25 people before we
36:22 realized that it was something that we
36:25 should write down and we went off site
36:27 four founders went off site and we tried
36:30 to synthesize the values out of what it
36:33 was that we really liked yeah there was
36:35 Apple Reddy happening and what it was
36:37 that we didn't like that we had a scene
36:39 already happening and not just among the
36:40 team but amongst ourselves to like what
36:42 were we not proud of that we had done
36:44 and what were we proud of that we had
36:46 and that that ultimately was what got
36:48 synthesized into those four values hmm
36:50 and so those were just interactions with
36:52 other people or like literally product
36:53 building no interactions with other
36:55 people and and interactions with
36:57 partners and customers and things that
36:58 we were proud of
36:59 mm-hmm that we wanted to see more of
37:03 right on so next question
37:07 Ashwin doke asks how is GDP are impacted
37:10 segments business model so GDP are for
37:12 those who don't know is a new EU
37:14 regulation which basically gives
37:17 end-users a lot of rights about the data
37:20 that's collected about them and first
37:22 off I think it's an awesome regulation
37:24 both as a consumer but also wearing my
37:28 segment hat it's interesting in that it
37:30 impacts the entire globe because if you
37:32 are storing data about in use it isn't
37:33 it doesn't matter what Roche diction you
37:35 run your company in you were still
37:38 responsible to do that for an EU citizen
37:40 the biggest impact
37:42 broadly on the
37:43 overall ecosystem is it really
37:45 negatively impacts third-party data and
37:46 third-party data brokers because they
37:49 have no real consent path to the user
37:51 for sharing and buying and selling the
37:54 data because we help companies purely
37:57 with their first party data it's not
37:59 like a existential threat to us in any
38:00 way and in fact it's something that
38:03 we're really sort of aligned with for
38:04 another reason as well which is because
38:06 we're routing the data out to all the
38:07 different places where people are using
38:09 it so we're routing it out to an
38:10 analytics tool to an email marketing
38:12 tool to a data warehouse to a CRM to a
38:14 help desk to ad conversion pixels if
38:18 that user shows up and says hey I want
38:21 you to delete me from your system well
38:23 it's actually like 20 systems for most
38:25 companies and we're already plugged into
38:27 those 20 systems so it's actually now a
38:29 feature of segments that we can go to
38:31 those 20 systems and delete whatever
38:33 user is requesting it and clean up that
38:35 record across all those different
38:38 systems so for us GDP r1 is like aligned
38:41 with our values philosophically - is
38:44 actually an exciting new feature sort of
38:46 requirement that we can support and a
38:47 sort of value that we can provide to our
38:50 customers so we're huge fans nice that
38:55 was a unsuspecting sir but uh people
38:56 have been stressed out about it my
38:59 friend makes Instapaper and they have a
39:01 big issue with it's a big problem in
39:02 publishing where they relying on
39:04 third-party damn yeah especially these
39:06 like little tiny products that are parts
39:08 of really big companies even they didn't
39:10 necessarily know and yeah another
39:14 everywhere cool all right so next
39:17 question and roopa cool asks any advice
39:19 that you have on asking for more money
39:21 than you're comfortable asking for this
39:23 is part of your startup school lecture
39:26 well I guess one of your sales reps was
39:28 forcing you to ask for more a lot more
39:30 yeah yeah we had a sales advisor who is
39:33 well I gotta back a little bit we were
39:35 initially selling our product for ten
39:38 dollars a month and you know hundred
39:40 twenty dollars a year and we brought on
39:41 the sales advisor and his first advice
39:43 was well you have to ask for $120,000 a
39:45 year and I was like that's a thousand
39:48 acts that's crazy so we were going to
39:50 the first sales meeting me and him and
39:53 is with a company called xamarin
39:56 and I've since told NAT this story which
39:58 he found amusing but now was the CEO of
40:01 xamarin and as we're walking up our
40:03 sales advisor says okay you have to ask
40:04 for a hundred and twenty K yeah him this
40:06 meeting and I was like that's the most
40:07 ridiculous thing I've ever heard I'm not
40:10 doing it and he's like well if you don't
40:11 do it then I quit as your sales advising
40:15 alright I guess I'm asking 420 K so we
40:16 go in we have to you know demo and
40:19 everything yeah and he says okay well
40:21 what's the price and it's 820 K and I
40:23 turned beet red and he says well how
40:25 about 12 K a year I said okay well how
40:28 about 18 and he's like okay fine
40:30 so from his perspective 85% off from my
40:34 perspective I got 150 X and it was a
40:36 successful negotiation I I think I think
40:38 it's really hard to offend people with
40:39 price at least if you're sitting in the
40:41 same room or or on the phone it's
40:43 probably not a good idea to share
40:45 pricing information via email if you do
40:46 that then it's really easy for them to
40:48 hang up but if you're on a phone call or
40:49 in person there's a bit of a social
40:50 contract to a continue engaging
40:52 particularly in person you can recover
40:55 so I would encourage you to not be
40:57 scared of offending someone with a high
40:59 price yeah but maybe just start in
41:00 person which is probably the most
41:02 uncomfortable place to do it but gives
41:03 you the most opportunity to recover
41:05 right and the thing is like if it
41:06 actually banished your business then
41:08 that's just what it costs yeah you're
41:10 gonna have well and you have no other
41:13 way of assessing the value yeah yeah and
41:15 then in fact what will happen is when
41:17 they say that's crazy then you say why
41:19 and then they'll explain to you how they
41:20 actually value the product and then you
41:22 say okay and you value it according to
41:24 their logic and you ask for that price
41:26 and how long did it take you well are
41:30 you charging them 120 now for sure yeah
41:32 we have customers to get way more value
41:33 than that out of it now yeah exactly and
41:35 so how many customers did it take you to
41:40 reach that six-figure amount a dozen at
41:42 most yeah so it was amazing yeah yeah
41:47 cool one carlos garza asks how did YC
41:50 help to get segment where it is right now
41:51 now
41:55 YC was super helpful the most impactful
41:58 thing early on is just demo day you're
41:59 not going to find a bigger concentration
42:02 of investors who are excited about
42:03 investing in startups creates a
42:05 compelling event structures the timeline
42:06 in crowd
42:07 way helpful for a first round of
42:09 financing that can easily get strung out
42:11 and waste a lot of your time yeah that's
42:13 the first thing the second thing really
42:16 is the is the founder network there's
42:18 not only a lot of reasonably high
42:21 profile companies now that you can learn
42:22 from or companies that are sort of
42:23 farther ahead that you can learn from
42:26 now but there's companies at all stages
42:28 so there's almost always a group of
42:33 people in your area or in your market
42:34 that you can learn from and share from
42:36 so all the there are tons of little
42:37 groups that spring up you know like a
42:39 group of enterprise founders that are
42:42 all between like 70 and 100 people in
42:44 San Francisco and you can have dinner
42:46 once every two months yeah and that
42:49 becomes an incredible support group and
42:52 sort of way of learning about about
42:54 what's going on have you stayed in touch
42:57 with people from your batch a few yeah
42:59 exact same some good go right on yeah
43:01 I've heard of like these informal
43:03 founder meetups happening quite a lot
43:05 and again yeah it seems to be great and
43:08 uh it's a trusted network there's no
43:10 replacement for that yeah totally I
43:13 definitely didn't get that from college
43:16 all right Juan has another question in
43:18 the early stage what's the thin line
43:20 between ignoring a customer's suggested
43:22 feature or moving a customer's requested
43:25 feature to the core of your application
43:28 or product I think what I think what he
43:30 is trying to ask is basically like at
43:33 what point do you say like hey this
43:35 customer is requiring or asking for this
43:37 feature and we have to kind of hold the
43:39 line because we don't want to become a
43:40 custom dev shop
43:44 so should we integrate this or tell them
43:46 to you know find someone else the best
43:49 defense against that is having a clear
43:50 product vision for where your product is
43:52 going to go long term and if you have a
43:54 clear product vision for where it's
43:55 going long term it's a very simple
43:58 question of yeah is this thing in that
44:00 picture long-term or not and if it isn't
44:03 that picture long-term then you can
44:05 prioritize it to be sooner or later
44:07 depending on whether a customer is going
44:09 to pay for it or not yeah and if it's
44:12 not then it's not and you probably
44:13 shouldn't build it right yeah I think
44:14 that's like that
44:16 the infamous customers you don't want in
44:18 Scenario yeah where you just have to let
44:19 them go yes I
44:20 I guess the important thing is like
44:21 imagine the entire timeline of
44:23 everything you're ever gonna build feel
44:25 free to move things around we do this
44:26 all the time we move things around based
44:27 on like what customers actually want has
44:29 it's a reasonable signal of what's
44:31 actually more important yeah
44:32 but I wouldn't I wouldn't add major
44:34 things or remove major things from it
44:36 just based on one customer mm-hmm so
44:39 since you've done YC and it's been
44:41 several years now would have been the
44:43 biggest learnings since oh my gosh so
44:47 many a huge bucket or a huge area of
44:50 learning for me as weird as finance like
44:51 I mean I came from an aerospace
44:54 engineering background and then we were
44:55 doing software engineering for the first
44:57 couple of years and so you just start
44:59 completely unprepared for preferred the
45:02 like business side of of things so I've
45:04 learned a tremendous amount about about
45:06 finance as we've raised money and and
45:08 and you know learned to manage your
45:10 business with a piano and all those
45:12 things not that you should rush into it
45:14 but it's it's a huge area that can be
45:17 leveraged I think and you would have if
45:19 you were to do it again hired someone
45:20 earlier on who knew what they were doing
45:22 in on the finance side I actually think
45:23 we did a reasonably good job of that so
45:26 we hired a part-time CFO around town
45:27 that raised our series a so we were
45:29 about at about a million in revenue and
45:31 we were raised to fifteen million
45:32 dollars here he's a we had had a
45:34 bookkeeper up until that point but we
45:36 were like I feel like we should have
45:38 someone like you know point us as to
45:39 what we should be doing with the money
45:42 and maybe have like a plan or a model or
45:44 something so that was definitely the
45:46 right time to hire a part-time CFO and
45:50 Jeff berkland was super impactful over
45:50 the years
45:52 intro would have been the other big
45:53 important hires for you that I made like
45:56 a huge difference I'll have remembered
45:58 the exact hammer a whole bunch of people
46:00 but advisors maybe is an interesting
46:02 category sure part time CFO I think is
46:03 in that bucket
46:05 we had an HR advisor who was really
46:07 impactful we've invested more in HR than
46:10 most startups yeah of our size and I
46:12 think that was the right thing a lot of
46:15 startups like uber for example do not
46:16 and end up with I think really high
46:18 prices for this way I think it's a
46:20 challenge right because if you if you go
46:21 around and start googling like should I
46:24 hire CMO should I hire CFO should I hire
46:26 XYZ I think you can always find someone
46:29 strongly advocating for any particular
46:32 role but the challenge is like okay you
46:33 know you only have so much
46:35 and so much time and you can only find
46:37 so many great people so like where do
46:39 you and where and when do you decide to
46:42 hire those like optimal people for this
46:44 stage in your company right and so yeah
46:46 just kind of curious if there are any
46:50 big turning point moments for you it was
46:52 a huge turning point around 10 million
46:54 in revenue when we hired the first team
46:57 sort of execs to the team yeah one is
46:58 our VP of engineering and the other was
47:00 a VP of people they were the first
47:02 people who had previously been managers
47:05 and our VP of engineering had managed a
47:07 team of 150 at Dropbox so we went from
47:09 literally zero management experience
47:11 aside from what had been picked up along
47:13 the way going from zero to 50 people to
47:15 having someone who really knew it or two
47:16 people who really knew what they were
47:20 doing that was hugely impactful and we
47:21 should have figured out a way to do that
47:23 earlier 50 people in 10 million revenue
47:24 or whatever it was was way too late
47:28 yeah cool if you weren't working on a
47:29 segment right now do you have an idea of
47:33 what you would oh man I get to
47:35 occasionally invest in YC companies nice
47:38 and there's a lot of cool things
47:41 happening there I was blown away by this
47:43 like breadth of things that are
47:45 happening in the batch this batch there
47:46 was a really exciting company building
47:49 in space rocket engine and another that
47:51 was doing industrial inspections by
47:53 drone I just can't imagine a world where
47:55 we continued have people in harnesses
47:57 hanging off of wind turbines that's I
47:58 can't imagine that that continues for a
48:01 long time so that seems like an obvious
48:05 market opportunity so flying things I
48:07 have a background in aerospace
48:12 engineering cool man so if people want
48:14 to learn more about segments where
48:15 should they go if they want to learn
48:17 more about you yeah segments just go to
48:21 segment com or you can tweet at me on
48:25 twitter i'm ryan pique rei NPK okay cool
48:26 yeah we'll link it all up all right
Chrome Extension