What is the main reason though? The BigCo being too risk averse? Or employees having too much of a safety net? Or BigCo processes weighing the team down? A combination of these?
I don't know if there is a "main reason" for all instances. For me, on this kind of project at one of the Biggest Co's, it seemed a bit like the safety net point you make and it creeps into everything. There didn't seem to be much pressure because the project was never going to be pulled and had infinite money. This seemed to manifest in "good people" (my opinion) working at about 40% capacity and new hires having a background in companies more like BigCo than InternalStartup.
Edit: To clarify my last point, my assumption being that the point of InternalStartup was to allow deviating almost entirely from BigCo's approach to everything and that those people from OtherBigCo would be less likely to have that mindset.
If you work for someone with $80bn of shareholder capital at risk you can take 1/10000th of the risk at work than if you have $8M of shareholder capital at risk.
That depends on what the risk is. If the risk is to the credibility or business model or relationships of the $80bn core business, then sure, the big company absolutely must be more conservative. On the other hand, if only a tiny fraction of that $80bn is needed to underwrite losses from your skunkworks' niche product taking too long to launch, lack of rapid uptake isn't an existential threat (even if the project gets canned, most of the team will still have jobs and they might even keep the profitable part of the project alive) and resources can be shared with other areas of the business, there are other risks that are easier to take.
Big companies are risk adverse. At the same time, the motivation for the people is missing, in a startup you have the chance to become a billionaire or at least a multi-millionaire, in a big company you can get them a billion and be the employee of the month with a $100 gift card. I saw it myself some years ago when I got my company a ~ 5 million/year saving project all by myself with a one-time cost of ~ 20k, I got a $250 reward on top of the regular salary. That was a lesson to not care too much. 2 years ago I got them over half a million per year in a 3 days of work mini-project, I got nothing, not even a formal thank you note. That project was not regular work, it was none of my business and I came with the idea and the execution alone, so you cannot say I was paid for such work. Translate this into a risk adverse big company play-pretending to do startup stuff.
What's interesting about skunkworks is much it gets invoked by people vs how much it actually happens in practice.
We engineers love skunkworks: working on cool stuff, without accountability and a lot of freedom ? Who would not want that ! But there are very few situations where this is sustainable, for reasons well explained by OP.
Strongly recommend reading the book - they had tons of accountability as critical projects were handed to them on tight deadlines. Yes, working on cool stuff, but delivering at an extremely high level.
>We engineers love skunkworks: working on cool stuff, without accountability and a lot of freedom
I'm pretty sure people who keep saying this have no idea what working at Skunkwork was actually like.
If you read Ben Rich's book, which I recommend you do, you'll find out that they had tons of accountability, the more classified their projects were, the more they were drowning in paperwork. No employee was to be left alone with the blueprints and if one of two needed to go to the bathroom then the plans had to be locked in a safe during that time.
Also, the hours of work and the stress they were under was insane as shit would break unexpectedly all the time.
I'm sure this is not what engineers love, and what they tink they mean by Skunworks is cowboy coding and being paid handsomely to play like kindergarten kids in the sandbox with the latest shiny toys, while leaving work at 5 PM.
>I'm also pretty sure most engineers don't want this, and what hey actually want is cowboy coding, playing like kindergarten kids in the sandbox with the latest shiny toys, while leaving work at 5 PM.
Nope. They want autonomy, mastery and purpose. They want to be able to apply what they know appropriately - not to be micromanaged or frustrated by instructions that make no sense. They want to be able to develop skills that make them valuable to their peers and their organisation - valuable enough to be secure and able to earn sufficient to protect their families. They want to know why they are doing the things that they are doing and to be able to believe that their efforts will contribute to something worthwhile.
If you can provide that all your engineers will both love you and jump into a bonfire for you.
I should have explained a bit better what I meant by accountability in that context: reduced direct oversight from general management, almost kinda "chinese wall" style.
Besides the reasons given by OP, an issue w/ skunkworks is that it requires two things that happen much less often in practice than people like: you need very strong, skilled team, and the need for high impact that management believe cannot be achieved any other way.
Similarly, at the heyday of xerox park, it sounds like Taylor was key to enable true, long term autonomy for his teams.
The most famous version of this (where the name came from), was Lockheed’s. A family member’s father worked there, and had a lot to do with some rather exotic avionics.
I mean, sure, but it also misses the point a bit. "Startup in a big company" doesn't mean "yo take some cashdollar and try build and release something to the public", but it does mean "hey, take a small team and do your own thing with less oversight and process overhead from higher-ups".
The team I worked in at AWS was like this and it felt a lot like startup teams I've worked with. That didn't mean we could take 'brand and legal risk', and if that's what you want then definitely go and start your own thing.
If you want more autonomy and higher velocity than is normal at corporates, then a 'startup within a big company' might actually work out for you.
Yeah, this. People who are used to actual startup culture and talk about how "startups" at big companies don't work have a different internalized concept of what startup autonomy looks like.
The success criteria are different, and measured differently.
While I worked for Adobe, one of the executives there (Mark Randall, serial entrepreneur that was "acquired" by the company, awesome guy) launched the "Red box" initiative that eventually become https://www.kickbox.org/. The red box was essentially a "toolbox" that he created on how to build a startup (set of training materials + $1000USD credit card that you could use however you wish), and the purpose was to get enough proof so that you get executive buy-in to continue your idea - if you achieved that, you got the "blue box" (whose content was a "secret" but was essentially custom-tailored support for your project). The entire process banks on the fact that you validate your ideas _without_ involving the Adobe brand, because if you slapped "Adobe" over it of course people and press would get interested. There were some legal caveats here and there (e.g. you couldn't outright lie and claim you were _not_ with Adobe), but I think it is to some extent the closest thing that you can get to "startup within a big company".
I personally got to the blue box - but then I gave up, when I realized that to get it to actually ship I had to sacrifice nights and weekends. Which would have been maybe fine - except that I knew that I didn't get to keep any upside. I had most of the risks of a startup, but very little of the payoff. Great learning process, but for me - it only managed to convince me that I shouldn't try within a corporation, if I want to try I need to actually do it by myself.
I wouldn't mind, but I don't know what - ask if you have specific questions. There's ample material on https://kickbox.org and https://home.getkickbox.com/ (basically all the contents of the red box, minus the cash), plus it appears one even has access to a "kickbox community" from what I can tell.
In the link[1] virgilp shared it says red box participants get 1000 CHF budget, 20% of working time, and access to experts.
So they want you to prove out your idea with an allocation of one workday per week. I’d bet most people have to add their own nights and weekends to make it successful.
Do note that they appear to have fine-tuned the program. I was one of the early participants - initially the blue box was a "secret" because Mark Randall genuinely didn't know what to put in it. In the meanwhile, the blue box was refined and it appears that a gold box turned up. True to his form, he approached the entire process as an MVP that was evolved & enhanced in time.
I agree that an incentive structure is probably needed as a matter of practice (since few people will want to do extra work with zero upside), but I wouldn't call it exploitative.
You're an employee, you work for a guaranteed salary now in return for someone else shouldering the risk. That's the nature of the deal. If you want to share in the upside, you have to share the risk.
It's not like anyone's being forced to do this extra work (unlike, say, game dev, which is exploitative). It's just an option for people who want to tackle interesting projects.
Reward being tied to risk is only for the little people. One can point at numberous executive compensation packages where they have plenty of exposure to upside, and almost no risk (or even negative risk: a golden parachute if you get pushed out).
We're talking past each other here, so I'll try to simplify where I'm coming from...
It is exploitative by the company to expect you to work nights and weekends, but then all the possible benefits go to the company. If you're going to do work on your nights and weekends, then do work for yourself, not for a company that won't compensate you for it.
The above has nothing to do with whether or not it's like a startup; it's speaking only to the deal in question being exploitative.
(Side note: I don't have a problem working some nights and the occasional weekend for my job. But it's very rarely expected of me; it tends to happen when I make a commitment to delivering something in a certain timeframe and then it turns out my guess as to how long it would take was wrong. Because I'm flexible, so is my company; if I need to take a half day to help my daughter with something, nobody is going to push back on that. I just wanted to make it clear that I don't consider the occasional night/weekend exploitative automatically... just that the situation described by the OP appeared to be so.)
> It is exploitative by the company to expect you to work nights and weekends, but then all the possible benefits go to the company
We’re disagreeing, not talking past each other. (Which is fine!)
Above-market salaries, bonuses and the promise of promotion (within the company or without) are fair compensation for aggressive work expectations. The upside is sharply capped in comparison with start-ups. But the downside has been practically eliminated. Nobody bemoans investment bankers, who have largely this compensation model.
My understanding was that this person's normal job did not include working nights and weekends. It was the act of taking on the "company side project" that necessitated that extra work.
If the person's job includes working nights and weekend and compensates fairly for it, and they have the option of working on an official side project _instead_ of their normal work, then I agree with your analysis of the situation.
If the person's job does not normally include nights and weekends, and they are compensated based on that, and adding the official side project to their work adds this time with no added benefit, then the company is being exploitative.
At the high level...
- if the side project can be done instead of your normal work, all good
- if the side project is done in addition to your normal work, with no additional compensation, then it's exploitative.
The (implied) benefits you get is that you can work on interesting things, and probably get much more freedom in your day job.
Looks very fine to me if done on a voluntary basis.
Sure, in that sense it’s “fair”. It’s also meaningless, as this example shows most people will not make a really great product if there is no upside for them so they will mail it in. So sure you can try to squeeze your employees but maybe if you gave them some more you would get a smaller percentage of a much bigger pie giving a win win. It’s very unlikely big corporations can do that for the precedence it sets.
It's not, at all. You know very well what you get into, and have every option to get out at any point, with zero downside to your career. You get paid, get to learn a lot, get to experience building a product with basically zero risk for yourself. Why would the company also give you the upside? They put a lot of resources into this program.
Also, it appears that Mark enhanced the program in the meanwhile/ I was one of the early participants. There's more structure now, and there's a "goldbox" that suggests he figured a way to give employees some guaranteed upside.
“except that I knew that I didn't get to keep any upside. L
Same at my company. They have similar programs but you also don’t get any stake in the outcome and are still controlled by executives who in the end get the credit.
Same for hackathons they tried to organize. The idea quickly turned from fun projects basically into overtime to check off Jura tickets quickly but with the addition of free pizza.
I think leadership in big companies is almost by definition very controlling. They simply can’t let go. It’s against the instincts that got them into their positions.
It is astounding to me how smart, conscientious adults can be manipulated with kindergarten-level tactics.
I dont know if psychopathy (or the whole dark triad?) can be quantitatively measured but I would not be surprised in the least if after the mean value, any % of incremental in psychopathy is way better than its equivalent in IQ to survive and thrive in the corporate world.
If they were honest and called it a "beat down tech debt day" I would probably be keen for that. I am easily swayed by free pizza, and I love cleaning up hard to understand/bug prone code.
That's not necessarily bad so long as everyone is honest about it. Often tech debt and bugs don't get fixed because there isn't enough time or they don't generate revenue. But we developers hate tech debt and bugs. Dedicating a chunk of time specifically fixing these issues that we hate can result in good outcomes.
Sounds like in your case it worked out, but I would consider AWS to be an insane outlier...also from a company still founder-led (Bezos is an exec with proven startup success--hence why Amazon exists in the first place!).
I think the point of the author is that, inside a big company, the fundamental risk/reward incentives that drive startup innovation are removed. Autonomy alone doesn't drive people to make crazy decisions. Risk/reward incentives do.
Google is never going to say to one of their internal employees:
"We're cutting your salary to $50k/yr, but you get 80% ownership of the business you create if you succeed. But there's a 95% chance you won't. And if you don't succeed you're fired. Good luck!"
That's the equivalent environment needed if you want to mimic the human decisions and behavior that drive innovation inside startups.
The problem is, if this employee is the perfect person to create this business, why are they working at Google and not on their own starting this venture already?
Why waste time letting one of your clearly risk averse in-house employees (hence why they work at Google in the first place), fumble around for 5-10 years trying to build something? While your competitors might be doing the same thing and will actually succeed? And while the rest of your employees start complaining, "why can't we also play startup for 5-10 years on the Google gravy train like John?"
Google can just sit back and let the real startup ecosystem do its thing. Then buy whatever they see that has shown success in areas of strategic value to Google. No PR or legal risk while the startup does necessary but shady things to force itself into being. This is a much better model, hence why innovation is most often acquired, not incubated.
I think the point he was making is it’s a half measure compared to having no safety net. The “adjust as you see fit” would be not operating under the Google umbrella at all.
The simple act of having to do the rounds to repeatedly secure financing and letting that, or the actual performance of the business, gate growth or survival and being a cofounder or employee in this environment has a different emotional and operational strain on the business.
I worked in a company like this btw for about five years, we were started as a subsidiary of a successful medical device company by that ceo as a “what if”, to take the already developed dispensing hardware by our parent and adopt it to the general supply chain management industry. We were doing decent business ($30M ARR for a 40 person company) and were minimally profitable but eventually shuttered by our parent company after it became apparent the hockey-stick like growth was not coming.
I’ve also worked at startup that failed after about 6 years. The difference between how those two companies screeched to a halt was stark. In one case it came out of the blue and suddenly 1/3 of the employees were sucked up into the parent company and the rest got pink slips. In the other case it was a really wild final year with the writing clearly on the wall, multiple furloughs and downsizing for survival.
The word for this is "skunkworks". Even the most independent skunkworks still serves big-company goals, often taking advantage of big-company resources and generating IP for the big company.
In some ways, a skunkworks is startup-like but in many critical ways it's way different.
Just came here to say this. The popular examples were the legendary work at Lockheed Martin, and also at Bell Lab for Unix. I was under the impression that Unix was well supported in the beginning at Bell Lab. Apparently according Ken it's not so and initially they need to piggyback on a grant that did not even mentioned OS or system research due the outstanding issues with the Multics project.
Honestly just working on something without an existing user base makes a huge difference.
There's a certain amount of institutional crud that's just crud, but most of it exists because once you are successful, you can lose what you have more easily than gaining something new.
Millions or billions of people use the major products FAANGs provide. They might get more users if they make these products better, and boy howdy are there a thousand ambitious devs, PMs, and managers trying, but they'll also lose millions of users if they make them worse. Hell, they'll lose millions of users if they make them different, regardless of whether it is better or worse.
90% of institutional crud is "how can we not piss off the billion people who are happily using our products and giving us money, either directly or indirectly?". Users hate it when our product breaks, so we're going to have excruciating review and approval processes to submit code changes, style guides and minimum test coverage and an entire campus devoted to monitoring for problems. Users hate it when their personal data gets leaked or abused, so we're going to set up razor wire and require devs to go on a quest for mystical artifacts before they can do anything with it. Users hate it when the site changes so we're going to require a dissertation in the form of a sonnet explaining why the color of the buttons should change, and then six months of live experiments to be sure.
90% of what you hate as a user about big companies is the result of them "acting like a startup" and radically redesigning UIs or changing the core functionality of a product or breaking existing functionality to add new or just developing a new product out of nowhere and presenting it to you like you should care.
Agreed on this feeling within AWS. The product I worked on came out of an engineering brainstorming meeting, we had to get it funded, small team, etc. Felt very much like startups I've worked at before and was part of the reason I stayed at AWS so long.
From the data science side of things, this works if and only if upper management has the discipline to leave you alone, including letting you procure your own tooling & compute if needed.
Glad you shared your experience. The title of the article is completely misleading and many people in the comments who claim it's impossible are either wrong or don't understand the meaning of this phrase.
I have been a (very happy) part of two such startup-within-a-big-corpo initiatives during my career in investment banking.
Although on different (in my opinion: better) terms, it does happen, although quite rare.
> "yo take some cashdollar and try build and release something to the public", but it does mean "hey, take a small team and do your own thing with less oversight and process overhead from higher-ups"
>but it does mean "hey, take a small team and do your own thing with less oversight and process overhead from higher-ups".
If you are interviewing for a "startup in a big company" job, one of the first questions you should ask is "how much oversight and process do we have from the larger company?" Make sure you ask a lot of questions about how decisions are made, who the product is being built for, how it is being sold, etc.
I've worked in two "startups" in two very different big companies. In both situations, the every major decision had to go through the larger org, which meant basically the whole thing was a waste. The only things that were approved were things the big company was already doing.
I'll provide a concrete example. I was involved in a product where our people had to work with the larger account execs of existing customers. The idea was the account execs could take the product to existing customers easily and grow very rapidly. What really happened was the account execs refused to put the product in front of customers unless it fit into the larger enterprise architecture strategy. They didn't need a startup - they needed a typical huge enterprise software team. So basically the product that got built was a completely half-baked POS that only looked good on slides. It was a market failure.
I was really shocked to see into a couple 'startup in a big company' teams at my company and to realize just how little they actually went off-leash. I don't know the details -- how much of it is not being allowed, how much of it is people who've worked here 10 years believing we're doing things the right way -- but I was really taken aback by the number of things they had to slow them down and the amount of oversight that seemed to be in place, given the general concept.
The creation of EC2 originally (which at the time was not AWS -- AWS was originally the XML-RPC and SOAP APIs that allowed people to interact with the www.amazon.com website) was done by a skunkworks project in South Africa. That team was very, very far removed from Amazon proper and actually delivered.
That's probably the only time I've seen that kind of startup-with-a-large-corporation work. I can think of other examples (like Amazon's google-competing search engine) which tried to copy that and failed pretty completely.
I suspect other successful things that happened after I was there were run as "startups" though (I'd guess Alexa?).
But in general every other time I've seen those happen at other companies they've been completely miserable failures.
Working in IT, usually those "startups" are full of Dunning Krugers who mostly go to war against the corporate IT and don't actually have very good business ideas (and honestly the corporate IT has been run like crap and deserved it, but that doesn't actually help launch your product). They're usually fed a bunch of ego-food about how they're the special children which will entirely transform the next generation of the company, and then they wind up fighting with everyone else.
> The creation of EC2 originally (which at the time was not AWS -- AWS was originally the XML-RPC and SOAP APIs that allowed people to interact with the www.amazon.com website) was done by a skunkworks project in South Africa.
AWS did indeed start out with E-commerce Services (2002) and Alexa Internet APIs (2004), but in mid-2003, when Andy Jassy took over AWS from Colin Bryar, he completely changed its charter to build an "Internet Operating System" instead. EC2 happened in South Africa in 2004 after Jassy and others had ear-marked compute as one of the key building blocks, along with Storage (S3) and Database (RDS / SimpleDB). In fact, S3 launched before EC2 did. SQS launched even before that, in 2004, though in limited beta.
Yeah, it was a bit messier than that though. There was a ton of internal politics between 2004-2006, that Wikipage makes it sound way too clean. You're right though that S3 and SQS launched before EC2 did, but EC2 was highly politically charged internally. I don't know that it would have happened if they didn't ship development off to South Africa.
A lot of affirmations that it is not possible, with examples of where it didn't happen, but doesn't really go into "why" at all. The closest they get was near the end by alluding to risk-aversion and having to wear Google badges to the cafeteria.
I worked for a startup within another company. It got spun out and the CEO used it as his own escape hatch, while selling off the remaining parts of the original company. The startup had a successful exit, although it didn't become a unicorn.
Any time someone makes a categorical statement like these, you know it's not wholly true. There are more permutations to the real world than anyone can think of up front.
Being a startup in a large corp works if you are the "famous" manager inside the corp and get your personal playing ground (skunk works?) project and are building something new.
It works less, when you are being acquired and want to keep your independence. When acquiring the acquirer will look close at you to align the acquired product with the corporate interests.
I'm not sure why everyone seems so focused on the cafeteria thing. I'm not sure people and their attitudes towards amenities are really the problem here.
From my understanding, the differerence is the companywide attitude to risk. A startup has a "grow at any cost, or maybe perish" attitude. If things go south, bankruptcy will take care of the leftover excess risk (barring criminal charges). A bigco cannot easily go bankrupt, even less a single department. There are tons and tons of capital to eat through if the accumulated risk is realized in cost. So a bigco has to do something about that risk somehow, because most of the company wants to keep what capital it has, only a small part of it really wants to risk things. So the only means to have a situation where you don't risk bigco for a single department is not making it a department. Make it a Ltd. in a holding or something.
If people really think of founding enterprises as a gamble with potential infinite reward, but limited downside (as bankrupcy will take care of it), I would say our economy has a problem...
It's this that makes entrepreneurship more abundant in the first place. If you have an economy with no risk taking you have no growth and no dynamism. Sure, 80-90% of businesses will fail, it's essential that people can fail without too much personal risk after all, but the 1 in 100 businesses that go onto grow to employ hundreds or thousands thanks to an effective business model are more than enough to make up for the failures. When someone fails in a startup it's not like they drop out of the economy altogether. Its not like they learned nothing. They go on to become effective employees at other companies, producing value there.
Why? Limited downside does not mean small, it just means limited. It can perfectly destroy a person's finances if they invest all their lifesavings in it. And the reward for a company is normally a % of how much value they provide to their customers (yes, I know, not always, but generally that's true). Unbounded does not mean infinite.
> So the only means to have a situation where you don't risk bigco for a single department is not making it a department. Make it a Ltd. in a holding or something.
I was thinking that as well. Can anyone explain why this isn't common practice?
> I was thinking that as well. Can anyone explain why this isn't common practice?
If you squint a little bit, this is exactly what the whole premise of corporate VC is. Take the funds that you'd allocate to long shots and operate as a VC would, provide strategic distribution where you can add value as a bigco, etc. Problem here is that compared to pure VCs you risk portfolio conflict in a different way that may not be as attractive to founders, but at least it's viable.
It is rare I think because of the parent company's CEO and founders don't like to really lose control to someone else.
A child company is not easily managed (and they shouldn't, but that's scary and risky by itself. The parent company need to have the know-how of an a serial startup builder investor)
The thing with the cafeteria is that it reflects company culture. If you in the office sit at a desk in the Google search engine area or youtube or Waymo is hardly noticable. People have settled in an if the neighboring teams go home at "normal" times, the "startup" team will go home at those times. Also you have a hard time to keep the small team culture when directly embedded. And then corporates encourage changing jobs internal over leaving and hiring somebody else, which means that there are ties on all levels. Pulling out and having your own startup culture is hard. The cafeteria symbolizes that, as it is a social place and culture to large parts is a social thing.
So like, the whole startup thing boils down to being at work late and working overtime, despite all the studies on crunch showing it is inefficient and less productive?
I think that summary is too narrow (i.e. the whole team, maybe even as "us vs. them" thing etc. are part of culture) but yes, for many of those startup guys "do 'hard' work, push things" seems to be an important piece and that's a reason for me to keep my distance.
It's just big company talk to pretend they're cool and innovative. Obviously ridiculous garbage that probably got someone a good annual review score.
A small company I was working with had heard that IBM, who had a customer in common, were operating some sort of "garage" thing that the customer really liked. Rumours were flying around, someone had heard it was a video conferencing app, they were desperate to know more about this secret sauce.
Eventually I got to speak to an opposite number from them on a piece of work that joined with our own. They were calling any team of people a garage (because of the Amazon, Apple, etc stories). That's all it was, just a different word for team, like toddlers playing let's pretend.
When I worked at a startup we copied the Spotify model as well. We called it "functional teams" and so on but the idea was the same - we watched the spotify videos as well. To be honest it helped. I think this is a point in IBM's favor because they saw something better, and adopted it even though "it was not invented here".
Indeed, it was the same in the Watson group where I worked about 6 years ago. It was pure cargo cult activity that ultimately resulted in much lower productivity.
I actually liked my experience with the guild structure, as somebody new to the industry. I had my specialty, was the leader in my team in my particular specialty, but that's mostly because I was the only one in my team with the specialty. The guild structure got me the opportunity to network with people who were more experienced in that area
Where I work we have teams and guilds. Guilds operate every other Friday, they work on cross-team concepts (e.g. automation, ai/ml, security, open source). Some guilds work better than others. The downside is around ownership (e.g. A guild creates a useful tool that ends up being used throughout the org, who fixes the critical bugs that arise?)
We originally cargo-culted the Spotify model, but changed it to fit how we want to work, it seems OK.
Woah I think we've stumbled on something here.. Spotify seem to be the source/example of all these new management fads. I remember having to sit through that hand-drawn sketch video (is there a name for these things?) describing how Spotify does SAFe (if you don't know what this is, pretend you never read it and continue your life as usual, please) and how awesome it is and why we all need to do it.
Opinions are my own, but it’s basically a worst version of agile (safe = scaled agile framework) that removes all autonomy from teams and gives upper management/leadership more power in making decisions at the team level.
If you never heard of Allen Holub look him up on YouTube/Twitter. He’s one of the same voices when it comes to agile frameworks.
I worked within IBM garage, it's not 'any team of people', it's a specific organization that mostly does services work. Their 'secret sauce' is that they're trying to consult on technology and company culture at the same time, in line with Conway's lay.
I.E. If you want to be more agile you'll want to adopt CI/CD, if you want micro services you'll want smaller more autonomous dev teams.
Yes but the parent's point is that the term "garage" has nothing to do with how or where the team actually works. It's a make-believe name chosen because of its association/connotation, not because of any actual grinding in reality.
There is a physical garage (more of a hipster warehouse styled co-working space) that the org no longer fits entirely in, but was still in use up until Covid hit. The plan pre-covid was to build more of these spaces to use for co-working sessions with clients.
If you don't really own your company, you won't ever behave as if you owned it.
It's basically a less rigid version of a big corp job, not more, not less. Which isn't bad, big corps have to rejuvenate themselves in some way, but selling it people as a "win-win" because they get to work in a big corp AND do a startup is a simply a lie.
If you sell your company for millions/billions to a FAANG company, stop lying to yourself, you did it for the money, which is fine. Pack up your things and do something else with that money
> Perhaps Corp-Tech should move to employee share buy back where employees must sacrifice some of their salary for equity or change equity to vest by a product related metric to connect the teams performance with the employee returns.
This doesn’t work all that well in my experience.
Employees focus in the stock which is not in line with the success of the product.
ESPPs aren’t bad. They’re seen as a bonus by most employees and some get excited watching the stock climb. Some may try to use that as a motivator.
But it’s more like watching your favorite team on TV that you’ve bet on with no control. That doesn’t inspire the right behavior.
Vesting stock based on some product metric would still be hindered by the futility of attempting to tie stock price to the actual success of the product.
Ask yourself this, would your "startup within a big company" have managed to lure Mark Zuckerberg or Larry Page to join them instead of founding their own companies? And even more important, would it even let such people have lead a significant part of the company in order to realize their ideas?
If you answer no to either of those, there is your answer to why your "startup within a company" can't hit it as big as real startups.
This happens with companies who are not big enough to take over other companies, so they try to latch on the loophole in capitalism by spawning different products often completely unrelated to original purpose of the company.
Such company will have a competitive advantage over someone starting from scratch with a similar product.
Question is should it be allowed? This ultimately leads to creation of those companies that are too big to fall, that have enough budgets to buy laws designed to keep competition at bay, do creative accounting and other measures that small company cannot afford to do.
Fortunately these "startups" within a company usually fail and are not as effective as take overs.
I think take overs should be illegal unless the technology your are buying is going to be incorporated into the original company line of products and does not spawn completely new business.
I know this is not what the article is talking about, but there have been some success stories of companies incubating startups. An obvious example is Tinder, incubated at IAC.
Wait. In a lot of these startup companies the business plan seems to be grow the customer base and worry about profit later.(youtube, Instagram, waze) Selling to a big company with a lot of capital seems to be the goal or a survival technique.
Johnson and Johnson I’m told handles the aquiring of businesses well (it’s a business conglomeration of hundreds of entities). I know someone who left after their division was sold from JnJ and everything started going south.
There are lots of startups within big companies. Example is Zonky (peer to peer lending startup) within Home Credit - multibillion dollar lending entreprise. But they have been created with a strong team which felt ownership. And as Home Credit was growing like crazy themselves too, they left this internal startup to self-manage too.
This isn’t true. The digital service team for the US Government is indistinguishable from a high quality startup/consultancy that goes around and solves pretty cool problems in modern ways.
Can you give some examples of those projects? I would have thought a government in house software team would be doing things like replacing paper forms with web forms and replacing old COBOL applications with Java, along with being an approval board for government contracts.
I think this misunderstands the process by which any project gets funded. There is not a nice orderly process where the CEO says 'right we are now going to build a website so users can track their order - here is the spec, and the cash go do it.'
The CEO (or other CxO) picks from a multitude of projects that already exist in different forms of completion - both internal and external.
For every conceivable project in a large company, there are already 5 different unofficial versions of the project - two spreadsheets on an analysts desk, one knocked up by a senior lead who needed a solution and 2 being hawked around by a MD who had some spare cash and let someone run with a side hustle. On top of which there are 3 SaaS options and Oracle probably has one to sell you, and McKinsey will do a demo next Thursday.
I will lay good money that when they bought Waze there are two projects in Google already that did the whole 'tell us what is on the road ahead' thing.
But the CEO picked an external buy - and those projects went the way of the Dodo.
By time you have big co sign off to do a 'internal' start up, you have basically hit Seed / series A level. Someone with money believes in you. You are past the major points of startup failure (I don't know what the stats are for failed before raising A and after but I bets its waaay lower)
I have been in both sides - the small scrappy start up, the funded start up and the getting something off the ground unofficially in a bigco.
And they all feel the same until you get 'blessed from above'
The Series A slowdown - this is point where all the crazy starts to slow - you actually have lawyers to read things, etc etc, someone starts to consider holiday pay and HR stuff.
In a bigCo this hits all at once - all compliance needs to come in. People look over your shoulder. But its not that different to the Board suddenly asking new questions.
So, yes entrepreneur startups are different to 'intrapreneur' startups, but not that much. Its a fight to get anything off the ground, usually in spare time, and internal politics looks a lot like marketing plus who you know in the real world.
Finally - yes if google is giving out equity for free, then yes incentives are misaligned. This seems to be a google problem (one exacerbated by the fact that most previous tech giants had smaller tech giants after their lunch within a decade or two - we don't seem to see that in FAANG.)
I worked in a startup lab within a huge multinational company. In many ways it felt like a startup because we had very limited runway to incubate and commercialize our ideas, so that definitely put the fire under our feet.
Despite an effort to firewall us from the larger organization the culture and processes of the mothership always seemed to leak in and contaminate the organic startup flows. Additionally a large organization is extremely risk averse unlike a real startup and this manifests itself in a multitude of ways.
The mindset would be very different. There’s always a sense of urgency to get things done with the realisation that there’s no safety net. You ask yourself different questions going to work and that drives different level of passion.
I'm a VP for a software firm focused on IoT. I work closely with both actual startups and Fortune 500s companies looking to launch new connected products.
I agree there is no such thing as a start up within a big company.
But that's not the point.
The point is to innovate. And a lot of innovation can happen within a true start up. But a lot of innovation can happen inside big companies, too. I see it all the time. Most of it, though, is just incremental innovation, not massively disruptive innovation.
But that's true of most startups, too. Most startups (the ones that survive) are incremental in their impact: they make incremental improvements to some part of our world, and they get bought. Very few transform into the next Fortune 500 giant and have that level of impact on society.
Final thought: I have seen a few world-shaking ideas emerge from corporate America. The problem is it typically cannibalizes the core business, or is so outside the core business, and the C-suite doesn't know what to do with it.
326 comments
[ 3.1 ms ] story [ 288 ms ] threadEdit: To clarify my last point, my assumption being that the point of InternalStartup was to allow deviating almost entirely from BigCo's approach to everything and that those people from OtherBigCo would be less likely to have that mindset.
It's that simple.
We engineers love skunkworks: working on cool stuff, without accountability and a lot of freedom ? Who would not want that ! But there are very few situations where this is sustainable, for reasons well explained by OP.
I'm pretty sure people who keep saying this have no idea what working at Skunkwork was actually like.
If you read Ben Rich's book, which I recommend you do, you'll find out that they had tons of accountability, the more classified their projects were, the more they were drowning in paperwork. No employee was to be left alone with the blueprints and if one of two needed to go to the bathroom then the plans had to be locked in a safe during that time.
Also, the hours of work and the stress they were under was insane as shit would break unexpectedly all the time.
I'm sure this is not what engineers love, and what they tink they mean by Skunworks is cowboy coding and being paid handsomely to play like kindergarten kids in the sandbox with the latest shiny toys, while leaving work at 5 PM.
Nope. They want autonomy, mastery and purpose. They want to be able to apply what they know appropriately - not to be micromanaged or frustrated by instructions that make no sense. They want to be able to develop skills that make them valuable to their peers and their organisation - valuable enough to be secure and able to earn sufficient to protect their families. They want to know why they are doing the things that they are doing and to be able to believe that their efforts will contribute to something worthwhile.
If you can provide that all your engineers will both love you and jump into a bonfire for you.
Maybe if it’s a tiny bonfire.
Besides the reasons given by OP, an issue w/ skunkworks is that it requires two things that happen much less often in practice than people like: you need very strong, skilled team, and the need for high impact that management believe cannot be achieved any other way.
Similarly, at the heyday of xerox park, it sounds like Taylor was key to enable true, long term autonomy for his teams.
The team I worked in at AWS was like this and it felt a lot like startup teams I've worked with. That didn't mean we could take 'brand and legal risk', and if that's what you want then definitely go and start your own thing.
If you want more autonomy and higher velocity than is normal at corporates, then a 'startup within a big company' might actually work out for you.
The success criteria are different, and measured differently.
I personally got to the blue box - but then I gave up, when I realized that to get it to actually ship I had to sacrifice nights and weekends. Which would have been maybe fine - except that I knew that I didn't get to keep any upside. I had most of the risks of a startup, but very little of the payoff. Great learning process, but for me - it only managed to convince me that I shouldn't try within a corporation, if I want to try I need to actually do it by myself.
So they want you to prove out your idea with an allocation of one workday per week. I’d bet most people have to add their own nights and weekends to make it successful.
1: https://home.getkickbox.com/
That does partially answer what is not in the blue box.
That sounds super exploitative tbh. Using people’s passion and naivety to give you free moonshots.
You're an employee, you work for a guaranteed salary now in return for someone else shouldering the risk. That's the nature of the deal. If you want to share in the upside, you have to share the risk.
It's not like anyone's being forced to do this extra work (unlike, say, game dev, which is exploitative). It's just an option for people who want to tackle interesting projects.
> I realized that to get it to actually ship I had to sacrifice nights and weekends.
If you have to work nights and weekends, but get none of the benefits for it, you might as well work nights and weekends on your _own_ side project.
It’s paying your salary. Not having a salary, or having a lower and/or unreliable one, is part of the risk of a start-up.
> you might as well work nights and weekends on your _own_ side project
Nights and weekends plus working days. Not just nights and weekends. Start-up means all in. Hobby means just nights and week-ends.
It is exploitative by the company to expect you to work nights and weekends, but then all the possible benefits go to the company. If you're going to do work on your nights and weekends, then do work for yourself, not for a company that won't compensate you for it.
The above has nothing to do with whether or not it's like a startup; it's speaking only to the deal in question being exploitative.
(Side note: I don't have a problem working some nights and the occasional weekend for my job. But it's very rarely expected of me; it tends to happen when I make a commitment to delivering something in a certain timeframe and then it turns out my guess as to how long it would take was wrong. Because I'm flexible, so is my company; if I need to take a half day to help my daughter with something, nobody is going to push back on that. I just wanted to make it clear that I don't consider the occasional night/weekend exploitative automatically... just that the situation described by the OP appeared to be so.)
We’re disagreeing, not talking past each other. (Which is fine!)
Above-market salaries, bonuses and the promise of promotion (within the company or without) are fair compensation for aggressive work expectations. The upside is sharply capped in comparison with start-ups. But the downside has been practically eliminated. Nobody bemoans investment bankers, who have largely this compensation model.
If the person's job includes working nights and weekend and compensates fairly for it, and they have the option of working on an official side project _instead_ of their normal work, then I agree with your analysis of the situation.
If the person's job does not normally include nights and weekends, and they are compensated based on that, and adding the official side project to their work adds this time with no added benefit, then the company is being exploitative.
At the high level...
- if the side project can be done instead of your normal work, all good
- if the side project is done in addition to your normal work, with no additional compensation, then it's exploitative.
It's not, at all. You know very well what you get into, and have every option to get out at any point, with zero downside to your career. You get paid, get to learn a lot, get to experience building a product with basically zero risk for yourself. Why would the company also give you the upside? They put a lot of resources into this program.
Also, it appears that Mark enhanced the program in the meanwhile/ I was one of the early participants. There's more structure now, and there's a "goldbox" that suggests he figured a way to give employees some guaranteed upside.
Same at my company. They have similar programs but you also don’t get any stake in the outcome and are still controlled by executives who in the end get the credit.
Same for hackathons they tried to organize. The idea quickly turned from fun projects basically into overtime to check off Jura tickets quickly but with the addition of free pizza.
I think leadership in big companies is almost by definition very controlling. They simply can’t let go. It’s against the instincts that got them into their positions.
I dont know if psychopathy (or the whole dark triad?) can be quantitatively measured but I would not be surprised in the least if after the mean value, any % of incremental in psychopathy is way better than its equivalent in IQ to survive and thrive in the corporate world.
At startups and big companies
I think the point of the author is that, inside a big company, the fundamental risk/reward incentives that drive startup innovation are removed. Autonomy alone doesn't drive people to make crazy decisions. Risk/reward incentives do.
Google is never going to say to one of their internal employees:
"We're cutting your salary to $50k/yr, but you get 80% ownership of the business you create if you succeed. But there's a 95% chance you won't. And if you don't succeed you're fired. Good luck!"
That's the equivalent environment needed if you want to mimic the human decisions and behavior that drive innovation inside startups.
There is a possibility a better balance of risk/reward (say 5% of ownership and 150k salary) could well produce just as good or better outcomes.
Why waste time letting one of your clearly risk averse in-house employees (hence why they work at Google in the first place), fumble around for 5-10 years trying to build something? While your competitors might be doing the same thing and will actually succeed? And while the rest of your employees start complaining, "why can't we also play startup for 5-10 years on the Google gravy train like John?"
Google can just sit back and let the real startup ecosystem do its thing. Then buy whatever they see that has shown success in areas of strategic value to Google. No PR or legal risk while the startup does necessary but shady things to force itself into being. This is a much better model, hence why innovation is most often acquired, not incubated.
The simple act of having to do the rounds to repeatedly secure financing and letting that, or the actual performance of the business, gate growth or survival and being a cofounder or employee in this environment has a different emotional and operational strain on the business.
I worked in a company like this btw for about five years, we were started as a subsidiary of a successful medical device company by that ceo as a “what if”, to take the already developed dispensing hardware by our parent and adopt it to the general supply chain management industry. We were doing decent business ($30M ARR for a 40 person company) and were minimally profitable but eventually shuttered by our parent company after it became apparent the hockey-stick like growth was not coming.
I’ve also worked at startup that failed after about 6 years. The difference between how those two companies screeched to a halt was stark. In one case it came out of the blue and suddenly 1/3 of the employees were sucked up into the parent company and the rest got pink slips. In the other case it was a really wild final year with the writing clearly on the wall, multiple furloughs and downsizing for survival.
In some ways, a skunkworks is startup-like but in many critical ways it's way different.
There's a certain amount of institutional crud that's just crud, but most of it exists because once you are successful, you can lose what you have more easily than gaining something new.
Millions or billions of people use the major products FAANGs provide. They might get more users if they make these products better, and boy howdy are there a thousand ambitious devs, PMs, and managers trying, but they'll also lose millions of users if they make them worse. Hell, they'll lose millions of users if they make them different, regardless of whether it is better or worse.
90% of institutional crud is "how can we not piss off the billion people who are happily using our products and giving us money, either directly or indirectly?". Users hate it when our product breaks, so we're going to have excruciating review and approval processes to submit code changes, style guides and minimum test coverage and an entire campus devoted to monitoring for problems. Users hate it when their personal data gets leaked or abused, so we're going to set up razor wire and require devs to go on a quest for mystical artifacts before they can do anything with it. Users hate it when the site changes so we're going to require a dissertation in the form of a sonnet explaining why the color of the buttons should change, and then six months of live experiments to be sure.
90% of what you hate as a user about big companies is the result of them "acting like a startup" and radically redesigning UIs or changing the core functionality of a product or breaking existing functionality to add new or just developing a new product out of nowhere and presenting it to you like you should care.
I have been a (very happy) part of two such startup-within-a-big-corpo initiatives during my career in investment banking.
Although on different (in my opinion: better) terms, it does happen, although quite rare.
Or what we used to just call "An R&D department"
If you are interviewing for a "startup in a big company" job, one of the first questions you should ask is "how much oversight and process do we have from the larger company?" Make sure you ask a lot of questions about how decisions are made, who the product is being built for, how it is being sold, etc.
I've worked in two "startups" in two very different big companies. In both situations, the every major decision had to go through the larger org, which meant basically the whole thing was a waste. The only things that were approved were things the big company was already doing.
I'll provide a concrete example. I was involved in a product where our people had to work with the larger account execs of existing customers. The idea was the account execs could take the product to existing customers easily and grow very rapidly. What really happened was the account execs refused to put the product in front of customers unless it fit into the larger enterprise architecture strategy. They didn't need a startup - they needed a typical huge enterprise software team. So basically the product that got built was a completely half-baked POS that only looked good on slides. It was a market failure.
That's probably the only time I've seen that kind of startup-with-a-large-corporation work. I can think of other examples (like Amazon's google-competing search engine) which tried to copy that and failed pretty completely.
I suspect other successful things that happened after I was there were run as "startups" though (I'd guess Alexa?).
But in general every other time I've seen those happen at other companies they've been completely miserable failures.
Working in IT, usually those "startups" are full of Dunning Krugers who mostly go to war against the corporate IT and don't actually have very good business ideas (and honestly the corporate IT has been run like crap and deserved it, but that doesn't actually help launch your product). They're usually fed a bunch of ego-food about how they're the special children which will entirely transform the next generation of the company, and then they wind up fighting with everyone else.
AWS did indeed start out with E-commerce Services (2002) and Alexa Internet APIs (2004), but in mid-2003, when Andy Jassy took over AWS from Colin Bryar, he completely changed its charter to build an "Internet Operating System" instead. EC2 happened in South Africa in 2004 after Jassy and others had ear-marked compute as one of the key building blocks, along with Storage (S3) and Database (RDS / SimpleDB). In fact, S3 launched before EC2 did. SQS launched even before that, in 2004, though in limited beta.
https://en.wikipedia.org/wiki/AWS#History
Any time someone makes a categorical statement like these, you know it's not wholly true. There are more permutations to the real world than anyone can think of up front.
Being a startup in a large corp works if you are the "famous" manager inside the corp and get your personal playing ground (skunk works?) project and are building something new.
It works less, when you are being acquired and want to keep your independence. When acquiring the acquirer will look close at you to align the acquired product with the corporate interests.
From my understanding, the differerence is the companywide attitude to risk. A startup has a "grow at any cost, or maybe perish" attitude. If things go south, bankruptcy will take care of the leftover excess risk (barring criminal charges). A bigco cannot easily go bankrupt, even less a single department. There are tons and tons of capital to eat through if the accumulated risk is realized in cost. So a bigco has to do something about that risk somehow, because most of the company wants to keep what capital it has, only a small part of it really wants to risk things. So the only means to have a situation where you don't risk bigco for a single department is not making it a department. Make it a Ltd. in a holding or something.
I was thinking that as well. Can anyone explain why this isn't common practice?
If you squint a little bit, this is exactly what the whole premise of corporate VC is. Take the funds that you'd allocate to long shots and operate as a VC would, provide strategic distribution where you can add value as a bigco, etc. Problem here is that compared to pure VCs you risk portfolio conflict in a different way that may not be as attractive to founders, but at least it's viable.
A child company is not easily managed (and they shouldn't, but that's scary and risky by itself. The parent company need to have the know-how of an a serial startup builder investor)
Big companies try to standardize to allow C players to perform as Bs. This crushes As.
A small company I was working with had heard that IBM, who had a customer in common, were operating some sort of "garage" thing that the customer really liked. Rumours were flying around, someone had heard it was a video conferencing app, they were desperate to know more about this secret sauce.
Eventually I got to speak to an opposite number from them on a piece of work that joined with our own. They were calling any team of people a garage (because of the Amazon, Apple, etc stories). That's all it was, just a different word for team, like toddlers playing let's pretend.
We originally cargo-culted the Spotify model, but changed it to fit how we want to work, it seems OK.
Move on.
If you never heard of Allen Holub look him up on YouTube/Twitter. He’s one of the same voices when it comes to agile frameworks.
I.E. If you want to be more agile you'll want to adopt CI/CD, if you want micro services you'll want smaller more autonomous dev teams.
If you don't really own your company, you won't ever behave as if you owned it.
It's basically a less rigid version of a big corp job, not more, not less. Which isn't bad, big corps have to rejuvenate themselves in some way, but selling it people as a "win-win" because they get to work in a big corp AND do a startup is a simply a lie.
This doesn’t work all that well in my experience.
Employees focus in the stock which is not in line with the success of the product.
ESPPs aren’t bad. They’re seen as a bonus by most employees and some get excited watching the stock climb. Some may try to use that as a motivator.
But it’s more like watching your favorite team on TV that you’ve bet on with no control. That doesn’t inspire the right behavior.
Vesting stock based on some product metric would still be hindered by the futility of attempting to tie stock price to the actual success of the product.
If you answer no to either of those, there is your answer to why your "startup within a company" can't hit it as big as real startups.
Only when you own a startup and that means outside a 'company' can you ensure your efforts profit you primarily.
It's the capital holders who primarily profit from your work, whether you're an entrepreneur founder, or a salaried employee.
> In March 2014, media and internet conglomerate IAC increased its majority stake in Tinder
Also, IAC (Match Group) owns most of the top US dating sites, so they were already in the dating market.
https://en.wikipedia.org/wiki/Tinder_(app)
Johnson and Johnson I’m told handles the aquiring of businesses well (it’s a business conglomeration of hundreds of entities). I know someone who left after their division was sold from JnJ and everything started going south.
https://en.m.wikipedia.org/wiki/Johnson_%26_Johnson
The CEO (or other CxO) picks from a multitude of projects that already exist in different forms of completion - both internal and external.
For every conceivable project in a large company, there are already 5 different unofficial versions of the project - two spreadsheets on an analysts desk, one knocked up by a senior lead who needed a solution and 2 being hawked around by a MD who had some spare cash and let someone run with a side hustle. On top of which there are 3 SaaS options and Oracle probably has one to sell you, and McKinsey will do a demo next Thursday.
I will lay good money that when they bought Waze there are two projects in Google already that did the whole 'tell us what is on the road ahead' thing.
But the CEO picked an external buy - and those projects went the way of the Dodo.
By time you have big co sign off to do a 'internal' start up, you have basically hit Seed / series A level. Someone with money believes in you. You are past the major points of startup failure (I don't know what the stats are for failed before raising A and after but I bets its waaay lower)
I have been in both sides - the small scrappy start up, the funded start up and the getting something off the ground unofficially in a bigco.
And they all feel the same until you get 'blessed from above'
The Series A slowdown - this is point where all the crazy starts to slow - you actually have lawyers to read things, etc etc, someone starts to consider holiday pay and HR stuff. In a bigCo this hits all at once - all compliance needs to come in. People look over your shoulder. But its not that different to the Board suddenly asking new questions.
So, yes entrepreneur startups are different to 'intrapreneur' startups, but not that much. Its a fight to get anything off the ground, usually in spare time, and internal politics looks a lot like marketing plus who you know in the real world.
Finally - yes if google is giving out equity for free, then yes incentives are misaligned. This seems to be a google problem (one exacerbated by the fact that most previous tech giants had smaller tech giants after their lunch within a decade or two - we don't seem to see that in FAANG.)
Despite an effort to firewall us from the larger organization the culture and processes of the mothership always seemed to leak in and contaminate the organic startup flows. Additionally a large organization is extremely risk averse unlike a real startup and this manifests itself in a multitude of ways.
I agree there is no such thing as a start up within a big company.
But that's not the point.
The point is to innovate. And a lot of innovation can happen within a true start up. But a lot of innovation can happen inside big companies, too. I see it all the time. Most of it, though, is just incremental innovation, not massively disruptive innovation.
But that's true of most startups, too. Most startups (the ones that survive) are incremental in their impact: they make incremental improvements to some part of our world, and they get bought. Very few transform into the next Fortune 500 giant and have that level of impact on society.
Final thought: I have seen a few world-shaking ideas emerge from corporate America. The problem is it typically cannibalizes the core business, or is so outside the core business, and the C-suite doesn't know what to do with it.