Full stack startups

Many of today’s most exciting startups were tried before in a different form.

Suppose you develop a new technology that is valuable to some industry. The old approach was to sell or license your technology to the existing companies in that industry. The new approach is to build a complete, end-to-end product or service that bypasses existing companies.

Prominent examples of this “full stack” approach include Tesla, Warby Parker, Uber, Harry’s, Nest, Buzzfeed, and Netflix. Most of these companies had “partial stack” antecedents that either failed or ended up being relatively small businesses. The problems with the partial stack approach include:

  • Bad product experience. Nest is great because of deep, Apple-like integration between software, hardware, design, services, etc, something they couldn’t have achieved licensing to Honeywell etc.
  • Cultural resistance to new technologies. The media industry is notoriously slow to adopt new technologies, so Buzzfeed and Netflix are (mostly) bypassing them.
  • Unfavorable economics. Your slice of the stack might be quite valuable but without control of the end customer it’s very hard to get paid accordingly.

The full stack approach lets you bypass industry incumbents, completely control the customer experience, and capture a greater portion of the economic benefits you provide.

The challenge with the full stack approach is you need to get good at many different things: software, hardware, design, consumer marketing, supply chain management, sales, partnerships, regulation, etc. The good news is that if you can pull this off, it is very hard for competitors to replicate so many interlocking pieces.

My guess is we are still at the very beginning of the full stack movement. Many large industries remain relatively untouched by the information technology revolution. That will likely change now that startups have figured out the right approach.

72 thoughts on “Full stack startups

  1. I don’t think it’s vertically integrated in the classic sense. This isn’t oil company buying supplier. It’s a tech company building the complete experience including non tech parts. The end result looks more like a company that we wouldn’t call vertically integrated. E.g. Uber competes with taxi companies but I don’t think we’d call taxi companies vertically integrated.

  2. It’s related but also a new phenomenon. This is tech companies wrapping “non tech” functions around the tech to go after existing companies. If there is an existing word for this it is probably Geoffrey Moore’s “whole product.”

  3. poke says:

    Electrification is probably a closer antecedent. Power plant, power lines, light bulbs, motors, etc, all conceived of and delivered as a single end-to-end product.

  4. anshublog says:

    This is also why “this time its different” is true. For the first time (as far as I can tell) in the life of silicon valley we are no longer competing with each other for the same $1 Trillion Dollar IT budget – Microsoft vs Oracle vs IBM vs Juniper vs AppDynamics vs Hadoop vs New Relic vs whatever. We are competing with Starwood Group vs AirBnB, Uber vs Yellow Cab, etc. In the past, only ecommerce companies did that. This expands the TAM for Silicon Valley (or “tech” whatever that means) from $1Trillion to N Trillion.

    Whenever we have done that in the past – Google is “full stack” ad company competing with CBS and NBC, Amazon is “full stack” retail company competing with Target, we have had mega unicorns.

    This time its different. And bigger!

  5. Terry A Davis says:

    The CIA is dirty. I cannot play. I figured-out that the designation as God’s official temple is all that matters. I cannot lose.

  6. Terry A Davis says:

    APIC? No, I don’t mess with APIC. Here’s the plan…. “God says, make it work hardware people.”

  7. Is it really “end-to-end” though? I think it’s worth examining why going “full stack” is even possible today.

    I take it as a sign of how well the hardware/software ecosystem has matured. Uber could not have gone “full stack” without ubiquitous GPS-support in smartphones. Netflix could not have gone “full stack” without the Internet and fast broadband speeds. Nest would not be possible without cheap LCDs and ubiquitous in-home Wi-Fi.

    These companies aren’t taking some ambitious leap forward in doing everything in-house, it’s simply become much easier for them to take advantage of new infrastructure. I’m sure we’ll see this trend continue as the tools available to both consumers and developers improve.

  8. Anshu, Chris, You’re both spot on. The deployment phase for the internet is just getting started.

    The current terminology “Internet of Things” probably understates the potential impact. Just about every industry is set to be disrupted. The availability of realtime analytics, machine learning and cloud infrastructure, all at extreme scale and low cost, will allow the creation of completely new forms of “robotics and automation” that will in turn create all kinds of new full stack businesses. Robotic cars, ships, trains, factories, machines, homes,…. Robots on land and in the air (drones). In 10 years time, my guess is that Google will be mainly viewed as a full stack robotics company, not an ad company.

    The term “robot” has, of course, so much legacy, we probably need a new name for this new era, and for the “mega unicorns” that are going to be created.

  9. Great post – now these are real businesses. Being full-stack will also come with costs: in the past one of the advantages of internet companies was the lack of being tied to the speed and friction of the ‘real world’. Relevance will vary by industry. For example facebook could only control the whole experience when it would own distribution – network and spectrum which is rare and cost quite a lot.

    It’s probably only practical to integrate (extend stack) until the first universally accessible infrastructure element – eg (almost) no one needs to build power lines or roads anymore. This is missing in two important industries though: financial services and telecoms. For the second, this is why net neutrality matters.

  10. I thought about using “vertical integration” but I think it’s an overloaded term that implies something else. Here we are talking about tech companies incorporating “non tech” functions.

  11. I agree there are lots of enabling technlogies like GPS. But it’s still a new phenomenon to bundle so many “non-tech” functions into the tech company product offering.

  12. SixDoors says:

    This is true, and in line with this article: http://tomtunguz.com/the-new-marketplaces/

    Consumers expect a great customer experience, and having a mastery of the customer experience is key. The founder of eBay could not do eBay today in a purely decentralized way today with no built in payment system, package tracking, etc….

  13. Michael Toomim says:

    Is this different than the Apple approach of controlling and developing the best versions of hardware, software, and services?

  14. Putting Buzzfeed and Netflix in the same sentence regarding “Full Stack startups” is odd. Netflix produces and aggregate content legitimately. Buzzfeed looks for ways to bypass compensating content creators. This is not good. The future of work for many will depend on creative content because humans do that so much better than computers and robots. It took a lawsuit for Buzzfeed to change their copyright infringing practices, now they wisely use creative commons work.

  15. No, I think Apple has been the main inspiration for a lot of these companies. Microsoft was the ultimate “partial stack” company and inspiration for the prior generation.

  16. A sentence from item #3 states: “We take responsibility for the technology, the advertising, and the content ” Content, he included content in the stack they will be responsible for in item #3.
    Yet they have a history of being irresponsible with content. Imagine if it was software ip they had infringed. A developer they hired used backend he pulled from another company instead of building from scratch.

    Most Silicon Valley consumer startups manipulate content producers. Yet startups intuitively understand that content is typically the most important layer of their stack.

    To illustrate what would you say is the most important item in the console stack ?

    A.) The OS
    B.) The Console
    C.) The Controller
    D.) the Games

    It’s obvious you need to get everything right but in the consoles wars the games are the true difference makers the tech stack almost feels like a commodity.

    Silicon Valley consumer tech has been exploiting content creators. They see this strategy as a strength. It is not, not anymore. Their strategy has devalue consumer web tech, consumers expect their web tech to be free or close to free. Therefore making the content layer of the stack even more valuable. This is a severe point of weakness in the stack ripe for disruption, given the oppurtunity I will.

  17. David Yasnoff says:

    I’d be interested to hear your thoughts on how you think this is going to change the timeline/funding/headcount of startups.

  18. Agreed completely.

    To really grasp how important “Internet of Things” is, one has to step back and realize that servers, desktops, and smart phones were really just the first three major “things” on the Internet of Things.

  19. ak102 says:

    Great insight, I’ve always felt this but never could quite put it into words. This is precisely why at post.fm we chose to develop a new email service and not just a client, going against Paul Graham’s advice – wisely or foolishly, because the task is more than twice as hard. But the value is definitely in the “full stack” due to the enhanced experience. I think Apple led the way there – showing everyone how tight integration can be so much better than the traditional client/server architecture for the average consumer.

  20. Tarun Anand says:

    So what do you think about Android – is that a partial stack company (I use the word company broadly here) What about the ecosystem that these companies create around them? Is it harder to create compared to “partial stack” companies because then do so much in-house?

  21. I think many industries are going exactly the opposite way.

    Think Java “Write once run anywhere” as a portrayal of a poor attempt at “eating a business problem”, and you see it emerge that some functional value is more about the ease of use or the interface than the internal functional value.

    So consider business intelligence – yes there is huge work going into BI as a platform, but the real value is generated where you tap into domain expertise.

    This leads to many companies becoming “modules” that handle a well-define vertical problem. Salesforce is not full stack sales, any more than Moz is full stack web-dev, or even Twitter is full-stack communications or (dare I say it) Facebook is full-stack friendship

    So long as we use generic platforms – the smartphone, laptop, watch, screen, keyboard, memory module, which are built from components that can be plumbed into full stack solutions by integrators, the disruption offered by new entrants offering full-stacks is usually only due to their clean set-up with absence of cash cows and legacy relationships that must be managed.

    So yes these startups have been tried before in a different form – and this is exactly why they are only sustaining innovation – but there is often only any “new market innovation” on the very borders of existing use cases.

    A unicorn is not a full-stack horse dressed up in wings and a horn – but something far less mundane. (etymology : of this world) – it arrives where no one expects it – not conveniently as a mere extension of well trodden path.

  22. I would add any modem connected device – weather satellites, transponders, telemetry devices. But the commoditisation of the solution so that “everything” is included is what is freaky. Will my jar of dill pickles “know” it is empty and wants refilling or recycling – Will trash self-sort and book itself a ride to the producer that needs the raw materials for repurposing?

  23. Matthew Perle says:

    Since the role of Android is roughly equivalent to Windows in that generation, do you see Google getting back into building smartphones at some point? Or are they content with a less integrated experience as long as it provides a plausible alternative to iOS?

  24. ✌TJ Parker says:

    Harry’s certainly comes close to classic vertical integration given they purchased their supplier for $100M+

  25. Chris, brilliant piece. Over at Redfin, which has both our own real estate search and our own real estate agents, we talk about this as an end-to-end customer experience:

    https://www.linkedin.com/today/post/article/20130408160410-5434591-the-next-50-ideas-for-billion-dollar-software-companies?trk=mp-author-card

    It has worked out well for us.

    Greylock’s James Slavet made the same point about other companies too:

    https://www.linkedin.com/today/post/article/20131114195950-58666-airbnb-redfin-uber-and-the-rise-of-the-e2e-movement

    And then we just did a plug for it at SXSW:

    http://www.inc.com/welcome.html?destination=http://www.inc.com/nicole-carter/redfin-ceo-glennkelman-talks-about-50-billion-business-not-online.html

    And Tedx:

    I really like your emphasis on deeper differentiation, and delivering more customer value, which ultimately leads to more revenue.

  26. fl1nty says:

    Also to add to your point wouldn’t the classical definition of vertically integrated for uber mean making their own cars?

  27. Mohan R says:

    I think where are with MOOCs and online education is one of these as well. Still at a Technology Trigger, but so much more to figure out then just posting video lessons on the internet. What happened to teaching according to different learning styles. I think who ever can take the content and translate it to learning styles will win, more so then who has the most content.

  28. ORE Agency says:

    Brilliant article – in our industry, it’s crucial to understand just how big (and fast) the changes are…when helping our clients to form their narrative and shape their brands, there is really no way the models of old can still be used. Branding 101 is dead…

  29. TA says:

    “Many large industries remain relatively untouched by the information technology revolution.”

    Can you give some examples? I’m having a hard time thinking of any.

  30. Dmitri Cherniak says:

    Hey David, I’m on working on something along the lines you’ve mentioned: http://www.beaconreader.com

    Had same insight as you after building a consumer app for story creators. Looked at ads but with a $5 cpm, most people would make peanuts even though same 1,000 or so readers would come back time after time. Looked into how publishing worked after that and realized almost the entire system is a house of cards. News orgs are paying for views on social channels to meet their page view goals for advertisers.

    Either way, our approach appears to be working quite well and Silicon Valley is taking note. We’re actually in the current batch of YC.

  31. Mike K. says:

    Case in point, I have developed a new technology that can generically animate morphs between any two shapes. It’s artful, flexible, unique and patents are possible. I’ve been trying to build a full stack app around the technology (in the iCanMorph app for the iPad) that lets users do dozens of unique things with morphing that they have never been able to do elsewhere. I’m trying to build a user base so the app is free for now with monetization later. See the technology here : https://itunes.apple.com/us/app/icanmorph/id718045966?mt=8&ign-mpt=uo%3D4 and see the start up support site here : http://logicenginedesigns.com. This is an interesting case because not only am I pursuing a full stack approach (with a 40+ modes of unique user experiences including 2 unique games already developed), but I am also pursuing the old approach of licensing the technology with existing companies where they could use these new morphing techniques to add fun splash screens, unique gaming animations, and unique social media experiences based on morphing. Any advice or commentary on how to implement a successful full stack approach would be welcome. Please direct commentary here : http://logicenginedesigns.com/contact-us.html

  32. I need to think about this a little more Chris because at the same time you also have emerging and interconnected systems to allow a company to figure out what part of a value they want to be providing, and use other services for the rest. The AWS-ization of service chains, if you will, in media, hardware, science, other disciplines.

  33. Tommy says:

    Gotcha – so vertical integration can be a part of this, but it’s not sufficient for the whole. It seems to me that a lot of ideas in the past tried this in what is commonly referred to as the “Dot Com Bubble”, but since the Internet wasn’t really at scale yet (only 40-50 million users, no smartphones, low bandwidth/connectivity) it didn’t work. WebVan seems like one – there are now successful iterations on that product using small tweaks, but largely just because they can get scale (relayfoods.com is one)

  34. uncleMasta says:

    True. To say it another way is that the pieces of the stack are getting bigger. And we can build bigger stacks now. Template engines, load balancers, etc. don’t need to be rebuilt (or bought) anymore. So, we just move on to bigger things.

  35. Well, I mean compared to companies I saw 5-10 years ago that were trying to sell dispatch software to limo/taxi companies. It’s a matter of degree.

  36. I agree. I think layers are always changing from commoditized/valuable etc. The big change I’m trying to call out here is tech companies doing lots of “non tech” things to bypass incumbents.

  37. Zev L says:

    Uber is moving into full stack territory and controlling the customer experience more and more. A vehicle leasing deal, integrated GPS, and driver “city knowledge tests” are examples of Uber getting involved in the customer experience through non-mobile channels. Uber’s product is more than just an app, it’s the complete experience, which is started and completed with the app – everything else is “full stack” territory.

  38. The two concepts are not in conflict with each other. Very different, and in fact quite complimentary, types of companies. I would argue the full stack (aka end-to-end customer experience) company would not be possibly without the critical enabling infrastructure provided by the AWS-ization of service chains.

  39. It is using technology to cut out the fat. To create efficiencies. Creating a simpler experience for the operators of the value flow. It is boiling process down to the “human element” and leveraging technology to carry the load of standardized processes. We live in exciting times!

  40. iOS and Android are freely distributed technologies. Similar to the internet. They are provided infrastructure. Thats the point. Todays technological infrastructure allows todays starts-ups to wipe out entire industries of yesteryear.

  41. To start just about any consumer service industry that distributes workers to do work. i.e. uber vs taxi companies. In todays age individual people will be empowered to take pride in what they do and their reputation will be held online for all to know and this knowledge will give power to a new type of expectation from consumers. In the near future consumers will expect to see ratings of the individual, who is the service provider and the larger institution will act more as a government of that industry rather than a company.

  42. Sunil Pande says:

    That would be Bill Davidow’s “whole product” as in Marketing High Technology – An Insider’s View (1986)

  43. Sunil Pande says:

    First, what is the difference between Vertically Integrated and Full Stack? Second, ref. the statement “Suppose you develop a new technology that is valuable to some industry. The OLD approach was to sell or license your technology to the existing companies in that industry. The NEW approach is to build a complete, end-to-end product or service that bypasses existing companies” Why is this an OLD and NEW thing? Both were options in the past and both remain options today. Not everyone will license their technology, nor will everyone be “Full Stack”. Both types of companies have roles and I can’t imagine it can ever be an either or. Or am I completely missing it?

  44. Interesting but get hung up on idea of ‘full’ stack. In complex new products there isn’t really a finite stack so full is hard to benchmark.

    Seems the examples mentioned were successful because they focused on end-to-end customer *experiences* (or the Christensen Job-to-be-Done idea).

    Interesting to note, most of those companies started off bundling enabling components:

    Tesla started by building on the Lotus chassis
    Uber is possible because of the iPhone hardware and software
    Netflix bundled content and technology, first DVDs then devices and internet distribution

    At the end of the day, the stack is a function of the end goal, which is delivering great customer experiences. So it reasons that as experiences are reimagined and redefined, stacks will be as well.

    That could be startups’ greatest challenge, and also their greatest competitive advantage: they’re starting from scratch but also free to redefine and reinvent the stacks.

  45. Is one of the core reasons to why “Full stack Start-ups” are killing it in so many verticals that software is becoming the core competitive advantage in many industries? This combined with the fact that the speed of innovation in the software industry is so much faster enables companies with software technology as their core competence to leap frog existing actors and create superior user experiences and competitive advantages. Tech based full stack companies are able to integrate status quo parts of the their vertical and innovate where possible through software.

    Uber dominate the taxi industry through continuous improvements of the technology to optimize and automate everything from user experience, support, operations to marketing and sales. The same is true for Amazon, Netflix, Airbnb, and with a slight variation Tesla. Electrification is a huge shift to the automotive industry and Tesla built its core competence around it and created the innovations needed to create an amazing user experience. I do think Tesla is more of an example of that an entrepreneur can change the world rather than being part of a larger trend.

    Changing the core competence of a company has time and time again been proven to be somewhere between very hard and impossible. Purchasing software based solutions to compete is really hard because of the need to continuously stay ahead of the game and the pace of innovation. SaaS has somewhat enabled this, but in order to compete with a full stack tech company you probably need a multitude of SaaS solutions and nailing such an ecosystem demands, well, software as a core competence.

    Are we now seeing a new breed of companies that disrupt industries based on the ability to build, implement and scale software based solutions rather than competing based on a core innovation?

  46. I agree this is a trend and one I’ve thought a lot about as I’ve built my own local startups, Redbeacon and MyTime (http:/www.mytime.com).

    I think a lot of entrepreneurs who start full stack companies end up biting off more than they can chew. They didn’t start out to become owners of housekeeping firms, car washing firms, and handyman firms. Like most of us reading your blog, they’re tech entrepreneurs and would probably be more comfortable building a marketplace platform like AirBnB than training housekeepers to make sure to dust baseboards. Imagine if Yelp had said we’re going to make food that people can rate vs. rating food that other people make. It probably wouldn’t have scaled nearly as well.

    Proof in point, I’ve been consistently disappointed with most of the startups offering full stack services (I’m trying not to name names, but you can probably guess some of the ones I’m thinking about). I do think Uber and Lyft are the exception, and not the rule here.

  47. Drew Patterson says:

    Really provocative argument. But not all these examples seem alike. A core part of the Uber/Airbnb models (at least as I see it), is the fact that they are leveraging someone else’s underused assets/capital (their cars/apartments).

    To me, what you are describing is a reimagined franchise model. Thanks to continuous, realtime feedback loops with both active (eg user ratings on a driver or host) and passive (eg tracking driver location, monitoring frequency of inventory updates) inputs, a brand like Uber or Airbnb can consistently deliver a great experience to the end user without having to directly own/manage many of the intermediate steps or the direct delivery of the service.

    I also think we’ll see some of the same tools/opportunities within an enterprise (vs. in a direct to consumer brand). In your podcast, you mentioned some verticals where this hasn’t yet happened (eg transportation). I believe that’s a result of both the capex required to play in these spaces and the current consolidation of the industry. Livery service and apartment rentals are obvious places to start, since (nearly) everyone has a car or apartment they could use to contribute to the market. To me, the question is how you create the sense of urgency for change within a sector. At CheckMate.io, we’re trying to use guest demand for a mobile hotel check-in experience (like an airline experience) to drive industry adoption, but it’s an open question of how quickly change happens.

    It’s exciting to think about what other

  48. Fully agree with your post. Thought, I think we can go even further.
    We became so technology rich that in certain cases it is not even necessary to have new critical complex essence to build great value and amazing user experience.

    Just the fact that in past there was preference toward partial stack, speaks about abundance of technologies that are ready to become next generation services. Actually I think that there is a big gap between what we could and what we are able to do now.

    In certain areas there is a clutter that is holding on massive adoption of obviously great technological advances.

    Also there was a huge shifts in market segmentation, created by holders of the potential that they are unable to use productively because of misalignment of technology and usability.
    What should be delivery of promise that technology will improve our life experience became rather even more managing of to many unfinished single stack solutions.

    In this regard, I think Full Stack Startup is not one model, but any model that will deliver best possible value based format of intended solution.

  49. Excellent pieces. Keep writing such kind of information on your page.

    Im really impressed by your site.
    Hey there, You have performed a great job. I will certainly digg it and personally suggest to
    my friends. I am confident they will be benefited from this web site.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s