00:04Stanford University hello and welcome
00:09back to Yi 145 technology
00:11entrepreneurship I have a couple of
00:13special guests that I'm excited to
00:15introduce to you today
00:17max marmar and Bjorn Herman they're from
00:20the startup Genome Project and the goal
00:22of the startup Genome Project is to try
00:24and map the DNA of entrepreneurship and
00:27innovation in Silicon Valley and so I
00:30like to bring them in and I'll ask them
00:33a couple of questions and then turn
00:36things over to them to explain to you
00:37what they found so max and Bjorn
00:40thank you very much for joining me today
00:43yeah and so how did you guys get started
00:46with this project how did you meet and
00:48what was the inspiration for the start
00:49of Gena I think that's a long story but
00:53can keep it short with the inspiration
00:58which was simply seeing the massive
01:02explosion of entrepreneurship globally
01:03and also seeing that this explosion of
01:06entrepreneurship had a massive impact on
01:08our economy more than 90 percent of all
01:11job growth in the u.s. comes from
01:12high-growth ventures and most of them
01:15being tech ventures and at the same time
01:19we saw that there was an enormous
01:21failure it's more than 90 percent of
01:23these startups fail and most of the time
01:25they fail because of self destruction
01:26rather than competition and that's where
01:29we started thinking well isn't there
01:31maybe a way of how we can solve this
01:33problem and reduce this failure rate in
01:38initially we saw accelerators who were
01:42able to overcome some of this failure
01:45rate and we saw companies like Zynga
01:48that took a more data-driven approach in
01:50overcoming the failure rate of the new
01:52games that they released and so I had to
01:54put a lot of these different puzzles
01:56together and eventually the site of
01:57genome came out great
02:00and and so max we first met when you
02:02were a student at Stanford yeah so how
02:05did how did you get started with us and
02:07and what's what's been exciting to you
02:10about what you found yeah I think a few
02:12years ago was trying to find a really
02:14big problem that I could sink my teeth
02:17into and was spending a lot of time
02:20looking at accelerating technological
02:23change and seeing this sort of
02:24exponential increase in the power of
02:27technology to in our society and it's a
02:30main driver and changing society
02:32throughout human history I was how do i
02:34how do I participate in this trend and
02:37basically saw that most of the
02:40technological innovation was being
02:41carried out into society by tech
02:43startups but at the same time as Ben
02:45mentioned there was this very high
02:46failure rate so is how do we come up
02:48with a more systematic way of increasing
02:52the success rate of startups and at
02:55Stanford I was doing a independent study
02:57with Steve Blank and looking at I
03:01thought he had the first really new
03:03innovative approach for looking at
03:05startups in a more rigorous scientific
03:07way with his book four steps to the
03:09Epiphany and that sort of circulated all
03:11around the valley and was called the
03:13Bible for entrepreneurs no so Galia this
03:15this is what I've been looking for
03:17and that was sort of the the initial
03:19starting point and we've started to
03:22synthesize a lot of the ideas from many
03:25different startup thought leaders from
03:27Steve Blank to Eric Ries to out
03:30Osterwalder and Shaun Ellis and
03:32basically trying to build a more
03:35generalized model and understanding of
03:38how startups work and how they evolve
03:40and then build things that are very
03:42applicable for entrepreneurs to to use
03:45on a daily basis great well I think this
03:48is a really exciting project and I think
03:50this is really valuable useful stuff
03:53that you guys are working on so let's
03:55get into some more the details of it and
03:57have you guys present what you found and
03:59what your advice is to startups out
04:01there to increase their success rates
04:03yeah sounds good max now going to talk
04:05about the key learnings from the startup
04:07Genome Project we have by now looked at
04:12more than 15,000 companies and have
04:15assembled two reports and launched the
04:17first software application that allows
04:19you to assess your own company and the
04:22following slides are based on that
04:24research that we've been doing all over
04:29the key problem that we've identified
04:30was with software startups over the last
04:34year in our research is that I'm the key
04:37reason for failure is premature scaling
04:39which is basically a proxy for assessing
04:44how well a company mitigates risk and
04:47overcomes uncertainty in the key
04:51solution that we have found is learning
04:53which just basically means how fast can
04:57you learn from your ecosystem from the
04:59environment especially from your
05:00customers in what's the right valuation
05:03what's the right pricing what's the
05:05right team to execute etc etc now max is
05:10going to talk about how we assess
05:12prematures getting you know one of the
05:15primary problems of why companies end up
05:19prematurely scaling is because they
05:23create a plan and then just end up
05:26executing it and scaling right away
05:27because that's what a lot of traditional
05:32business management theory has taught
05:35the problem is startups have so much
05:39uncertainty that many of the assumptions
05:41in the plan end up being wrong and so
05:44what what pretty mature scaling is a
05:48meta concept of more systematically
05:51defining the stages that companies
05:53evolve through and helping them test
05:55their assumptions more incrementally so
05:58to break down a little bit more
06:01concretely how premature scaling works
06:04we basically look at the startup as a
06:07type of organism that exists in a
06:10environment and the organism of the
06:15startup evolves through four
06:17developmental stages so those four
06:20stages are discovery validation
06:24efficiency and scale
06:26and basically in the discovery phase a
06:29startup is trying to figure figure out
06:32whether they have problem solution fit
06:34so they're trying to understand whether
06:35they understand the problem correctly
06:38whether people actually have this
06:40problem and whether the solution
06:41hypothetically solves that problem
06:43in the validation stage they're working
06:45on putting out a Minimum Viable Product
06:47something that actually gets something
06:50in front of customers so that they can
06:52run real quantitative experiments and
06:55see if people actually like it in the
06:58efficiency stage startups are have
07:01already found product market fit and
07:03they're just working on optimizing a lot
07:05of the product from the user experience
07:08in the backend to the customer
07:10acquisition models and then finally once
07:14all of that is figured out they can step
07:15on the gas pedal and scale so the
07:18problem with which companies often do is
07:22they don't align this external feedback
07:24from from the market which are which we
07:28place in these buckets of discovery
07:30validation efficiency and scale they
07:32don't align that with the internal
07:35dimensions of the company that they can
07:36control so they scale the the team up to
07:4220 or 30 people and they haven't
07:43launched a product yet or they write a
07:46very complex product and haven't talked
07:49to anyone about whether they want it so
07:51we actually look at each of the
07:53dimensions of the company we have five
07:54they are the business model the
07:56financials the team the product and the
07:59customer relationship and we look at how
08:03far a long a company is on each of these
08:05dimensions and try to assess whether
08:07it's in line with the response they're
08:09getting from the market
08:11so one of the best examples of a company
08:14that's done this recently prematurely
08:17scaled is color when they they were in
08:22the validation stage from their response
08:24from the market in that they just put
08:26out the first prototype of what they
08:29thought was a good application to in the
08:33sort of local social network market but
08:37the problem was that they launched a
08:39huge PR campaign they raised over 40
08:43million dollars in venture capital and
08:45they had a large team of over 20 30
08:48people and this basically they launched
08:52this product and it turned out that
08:53people weren't as interested in it as
08:55they thought they ran a huge campaign
08:58for the Royal Wedding in London and
09:00there were almost no pictures taken and
09:03this makes the company much less able to
09:09adapt when they have when they invest so
09:11much in these big plans with twenty or
09:14thirty people and that what the theory
09:18of premature scaling recommends and what
09:22we advocate is to take a more systematic
09:24approach where they would have released
09:26this Minimum Viable Product with just a
09:29few users tested it got the value
09:32proposition right started to optimize
09:34the user experience and only then would
09:36they scale because this allows them to
09:39more systematically fine-tune a lot of
09:41the assumptions of what their business
09:43is going to be yeah I guess two theories
09:47that go align in inline with the idea of
09:51premature scaling and that are more
09:54specific in how you can overcome
09:55premature scaling as the Dean startup
09:58from Eric Ries and customer development
10:01Steve Blank now let's talk about a few
10:07sort of differences that we have found
10:09in our data between companies that scale
10:11maturely versus companies that scale
10:15prematurely and you can see here the
10:18difference and like let's see the blue
10:24bars and all the following slides always
10:26are the company's scale properly and red
10:31ones are the companies are scared
10:33prematurely and what we can see here is
10:36that companies that scaled prematurely
10:39never really overcome the 1 million in
10:44revenue they get stuck before which
10:46means that they are still they can be
10:49successful in some way but they hit a
10:52limit and don't move beyond that what we
10:57can also see is that companies escape
10:59prematurely half even a much faster user
11:02growth in the beginning but then drop
11:05off and come vicious scare properly and
11:08that go through this whole systematic
11:09cycle of doing customer development
11:11understanding about the position etc
11:13have a little bit smaller user growth
11:16than beginning but then take off and are
11:19much faster and much more successful and
11:21then companies are scared prematurely we
11:24can see very similar trend in valuation
11:26we can see that companies that scale
11:29prematurely or value themselves
11:31enormously in the beginning but then
11:34turn off turn out to be much less
11:38valuable later on you can also see same
11:41trend and funding companies that scare
11:44prematurely erase much more funding they
11:47people who like very excited and and
11:53convinced of the idea but they didn't
11:56really test it and they didn't really
11:58figure it out so they get much more
12:01money early on but then they fail when
12:04they have to scare so summing up
12:09startups are exposed to an overwhelming
12:12uncertainty which means they don't know
12:14what their market will be they don't
12:16know the pricing they don't know the
12:18exact model position to what kind of
12:21customers they don't know what is the
12:22team they don't know how their product
12:24will work and how big it will get or
12:27what are the key features and so on and
12:29all of these things they have to find
12:31out and that describes the search
12:33process that makes described earlier in
12:35those four stages where we put startups
12:37as we assess them and the big challenge
12:41for a companies to overcome this
12:43uncertainty and to understand all the
12:45things that they don't know in order to
12:48mitigate risk and to overcome all the
12:52challenges they have ahead of them and
12:53to become eventually a large company now
12:58there's a number of ways of how you can
13:01learn for yourself to identify premature
13:04scaling on your startup or on your
13:07friends or other people startups
13:10so we can see that companies that are
13:15scaling prematurely often monetize a
13:19much greater percentage of their user
13:21base early on and this is a problem
13:25because startups that are often trying
13:27to monetize early are in much more of a
13:31execution mindset rather than a search
13:34mindset and the the first three stages
13:37of a startup discovery validation and
13:39efficiency are really characterized by
13:41the search process rather than an
13:43execution and when companies get too
13:46stuck on trying to monetize they lose a
13:49lot of their listening ability and
13:52ability to adapt because they're so
13:55focused on selling rather than listening
14:00and figuring out what works and what
14:02yeah so you will always meet founders
14:06who are really pushing to make money and
14:09that's fine they probably will make
14:10money but if you want to build a
14:13high-growth company and really the key
14:15asset of the company is the valuable
14:17position that you have to your customers
14:19and if you push too early or too much
14:23towards making money instead of trying
14:25to figure something figure out what
14:27exactly your position is going to be
14:29you're much more likely to scale
14:32prematurely the startups just add one
14:37more more point that their growth curve
14:40and revenue curve are hardly ever linear
14:43so it's really about getting the the
14:46product right getting the customer
14:47acquisition right and only then stepping
14:50on the gas pedal and that's when you see
14:51the hockey stick type growth another
14:54really interesting indicator that we
14:56found with a lot of companies that scape
14:58prematurely is that they tend to
15:00outsource a lot of data or much more of
15:02the development than companies that
15:04scale properly and the just from our
15:09personal experience of building
15:10companies and working with a lot of
15:13the typical profile of companies that we
15:16see that alters the development of
15:18people who don't know how to put the
15:20product and the the result is that you
15:24the feedback loop from talking to your
15:27customers and learning how you should
15:29change your product to actually
15:31implementing those changes into the
15:33product becomes very very long which
15:36means become very complicated and
15:37there's a lot of friction that is
15:39created for especially in the early
15:42stages where you do a lot of changes all
15:44the time which then leads to a constant
15:48trade-off should you spend time and
15:50effort and to talk to your outsource
15:53developers to change something in the
15:55product or should you just focus on
15:59pushing and do what you're good at and
16:02talk to customers and and sort of let
16:04the developers go and and the result is
16:07typically that you have some the result
16:10of that is typically that the product
16:11turns out not to be as good as it should
16:15be yeah outsourcing just really
16:20decreases the startups ability to adapt
16:23because now there's a much longer
16:26feedback loop for them hearing something
16:27from customers to then having to change
16:30the specs to the outsource firm and
16:33startups really in these early stages
16:35want to increase their learning the
16:36feedback loop and experimentation cycle
16:40of experimentation so they're doing a
16:41lot of experiments very quickly
16:44another very important distinction that
16:47we've seen to identify companies that
16:50are scaling prematurely versus ones that
16:52are scaling properly is that ones
16:54companies that scale prematurely often
16:56focus on product development rather than
16:58customer development so they again have
17:01more of a execution mindset where they
17:05just want to build the product that they
17:08think will work and there's not as much
17:12listening to customers trying to figure
17:15out if their vision is actually
17:18something that people want to buy and
17:22they they consequently are much less
17:25adaptive yeah we can see the the result
17:32of of that slide can be seen here were
17:37we measured of experimentally in like
17:42the amount of lines of code written by
17:46different companies in the different
17:48stages and there and what we can see is
17:50that companies that escaped prematurely
17:53tend to write lots of lines of code very
17:56early on much less than companies that
17:59scare properly and the typical question
18:05of this you will see with engineers
18:10especially who start the companies and
18:11who know exactly the type of features
18:14that the company or that the product
18:16needs to have in order to actually be
18:18able position to the customers and they
18:20spend one year two years to build out
18:22this perfect product which only then
18:25they start testing with their customers
18:27and they find out that it actually
18:29doesn't work and and that's what you can
18:32see very explicitly here were companies
18:35already in the first stage tend to ride
18:38huge amounts of code before they have
18:43even validated their product where the
18:45the goal for entrepreneurs here is
18:48really to test their their assumptions
18:51as fast as possible and minimize waste
18:54where they where they write code that
18:56the end up having to throw away because
18:57it doesn't meet any need that that in
19:02the market yeah and I guess that's
19:03normal for almost every single startup
19:05that they will rewrite the code a few
19:07times and we'll have to throw out a lot
19:11features that they've spent lots of time
19:13on developing another big difference of
19:18characterizing companies that scare
19:20prematurely versus companies that scale
19:22properly is their self-reported
19:25valuation where we can see the companies
19:28that didn't really go through this
19:30reality check of talking to their
19:32customers many times over value the
19:35company to a huge degree and it's normal
19:39that in the beginning that's happened as
19:40we can see also companies that scare
19:42properly or value themselves but it's
19:46happens to a much greater degree with
19:48companies that scare prematurely one of
19:52the another reason why this valuation is
19:55so out of whack with companies that
19:58scale prematurely is many times these
20:01founders commit this mental fallacy
20:04where they mix up the the present with
20:07the future where they imagine the the
20:10company being worth a hundred million
20:12dollars or a billion dollars and they
20:14think that they're worth that much today
20:15and this creates a lot of confusion when
20:19they're talking with investors and
20:20they're splitting up equity with other
20:22co-founders but they don't recognize
20:24that they haven't even launched anything
20:25yet and the company is not anything
20:28close to that yet we can see a strong
20:30trend for companies that scare
20:33prematurely to spend a lot more money on
20:36custom acquisition before they have gone
20:39through like cost optimization that
20:43typically happens in what we call
20:44efficiency stage 283 we're after your
20:49validator your product you're going to
20:50try to optimize you input output before
20:52you start pushing in a lot of money so
20:55we can see that here companies are
20:56scared prematurely tend to overspend and
20:59before they have done this and that's
21:02practically throwing money out of the
21:04window without any effect and yeah you
21:09can see here that that especially in the
21:12hundred thousand-plus spendings per
21:14month it's mostly premature or it's
21:17mostly companies that scare prematurely
21:23the focus in the in the early days of
21:25the company the first three stages are
21:28again really about testing initial
21:30assumptions going through a search
21:32process and customer acquisition is is
21:36much less relevant for in until you get
21:39the model right then that's the time to
21:41to step on the gas pedal would really
21:44invest in growing the business no yeah
21:48basic you can learn depending on what
21:51kind of business you are you don't need
21:53millions of users in order to get your
21:56model right you just need fifty maybe a
21:58few thousand and and that's enough so
22:02you don't really need to spend a lot of
22:03money in customer acquisition or to get
22:05a critical mass of people to work with
22:07and to get the product to a level where
22:10it makes sense to spend a lot customer
22:12acquisition now summing up the key
22:18findings from our research has been that
22:22the key challenge that companies have is
22:25this over one an uncertainty of not
22:27knowing where they're going and what
22:29they're doing and who they're doing this
22:31with and and the key reason for their
22:34failure premature scaling which
22:37basically describes companies very
22:39typically very effectively executing
22:42something that is completely unnecessary
22:43and them investing a lot of resources
22:46into something and that turns out to be
22:49a waste of money later on and the key
22:53like strategy of overcoming this is to
22:56try to learn as much as possible in the
22:59very systemized way and there's two
23:03thought he said I've written a lot about
23:05that who's one of them is Steve Blank
23:08who wrote about the customer development
23:09and a more structured way of testing
23:13your assumptions with your customers and
23:16then the Lean Startup for my agrees that
23:19is also talking a lot about that and you
23:22can read more on all of these topics on
23:24our website there's a compass that CEO
23:29and lock that up compassed at CEO we
23:31have two research reports that we
23:33published on the block and we have an
23:37application that will allow your startup
23:39to to see whether you are scaling
23:42prematurely or not and if you have any
23:46questions send us an email at feedback
23:47at salad compass lets you or you can
23:50follow us on Twitter thank you for more
23:53please visit us at stanford.edu
![Welp, Here's what the Fed *Just* Said [Fed FOMC Minutes]](/_next/image?url=https%3A%2F%2Fi.ytimg.com%2Fvi%2Fl8u23oq1E9E%2Fhqdefault.jpg&w=3840&q=75)
