The No Stack Startup
There’s been a lot of discussion in recent years that the “full stack approach” is the future of startups. My friend Chris Dixon articulated the reasons for going “full stack” very well in this post from last year. But like many things, the best approaches are at both ends of the spectrum. Either go “full stack” or go “no stack.”
My partners Andy and Albert have been writing about the no stack approach this past week and it is the topic of the week at usv.com.
At USV we have never been excited by the full stack approach. It is well suited to investors who have unlimited amounts of capital to invest and a need to put all that cash to work. We aren’t that kind of investor. We like low capital requirements and low burn rates and extremely high rates of return on invested capital. So no stack seems like it will suit us well.
Our partner Brad said in an internal email about this today, “We need to think through defensibility, margin sustainability, and not having control of some infrastructure.” So that’s what we are doing now. And if anyone would like to weigh in on this, the comments here at AVC is a good place as is the usv.com topic of the week conversation.
Comments (Archived):
this is how lifestyle businesses are built. do they need funding however?
we decided a few weeks ago that we would not use the term “lifestyle businesses” at AVC anymore so i’m not going to do thatthis is how you can and should build a “personal cashflow business” but there are some that breakout into big scaleable businesses and we would love to invest in them
Adrian is 100% right: what’s the moat if you can build the product/biz to profitability over a weekend? Hint: you need VC capital to build the moat and scale, so it serves the VC well, eh? But if you’re totally reliant on third party APIs, that feels like a risky proposition. Lightweight? Sure. Defensible? Negative.
Defensibility is overrated.
Peter Thiel would disagree with you. On his book tour he constantly emphasizes how important it is to create…a monopoly.He provides Google and FB as examples.
See Fred’s comment about network effects. I’m just saying, the fear that someone can copy your tech implementation, is not a reason to not charge forward.
I agree with you both about network effects.It’s just then about whether you and competitors 1, 2, 3 get the scale ratios of the network effects better.Plus which part of the network you want.
Yesssss.
Peter Thiel would disagree with you. On his book tour he constantly emphasizes how important it is to create…a monopoly.I totally reject the notion that what’s good for the goose is good for the gander. Peter (or Warren Buffet or pick your icon) is operating a different level than 99.9% of the people (and I don’t mean intelligence wise just his circumstance) out there. Likewise the idea that the 99.9% of people who will not even come close to creating something large (or even profitable) or great listening to a thought like that (which could be right in some cases I will agree) makes no sense at all. It would be like someone coming into a high school music class and making all of the music students think they could be the next Elton John, Bono, Mick Jagger or Paul McCartney. It’s typical of the advice that is all over the internet which I will call “the answer”. As in “here is the answer” same thing that happens with weight loss and diets.
And so do I reject what’s good for the goose is good for the gander! :*).The thing about all the fantastic insights from Fred, Thiel, Musk…pick your icon…is that the founder and their team needs to know that THEY SHOULD HAVE THEIR OWN DOMAIN EXPERTISE/JUDGEMENT/LEADERSHIP on a whole range of issues and the investor/icon doesn’t necessarily know the secret keys that solve the problems in the same way as the founding team.There are so many nuances to navigate in founding startups and there is no precise blueprint because, if there were, there’d be dozens of Googles doing it exactly the same thing.Each founder like each musician is…unique. That uniqueness is the source of creativity and innovation.
No king or queen rules forever. Google and FB are monopolies to the same extent that Alta Vista and Aol were monopolies (or defensible) in the late 90s. I’m not saying it’s easy or imminent that these big companies lose their position. Just saying defensibility is certain for them.
Neither of them started out as full-stack or had network monopolies. Peter Thiel’s book just highlights the outcomes rather than the small details of how they built up their stack over time — either through acquisition or internal investment in R&D (that’s not something he wrote about, btw)Defensibility is, in part, any metric that proves the service is a habit (Google’s toothbrush test) and, in part, any metric that proves users are too meshed into the social binds of the platform that leaving it would feel wrong (if all your family & friends are on FB, you’re there too).
You are arguing Jim’s point. Monopolies don’t worry about defensibility, with the DoJ or Euro Commissions being exceptions.
I agree, all you have to do is count the # of Seed VCs who list Founder / Market fit in their top 3 investment criteria.If you can control the user experience and you can scale, you don’t need to worry about your end of the rink.
network effects can provide the moat even in a no stack startup
i’d love to hear more on this perspective, if anyone at USV cares to elaborate. are there any clear examples of this?
Anyone building a mobile app using AWS and the Ionic framework.You are coding and marketing. The ‘operating’ is 3P.
clear examples of what? i’m kind of lost in this thread
a clear example of a no stack startup as a compelling business opportunity. i think many of us get the example of a content producer — namely a youtube star — but are there are any specifics we can point to? perhaps there is a business that is successfully split across many platforms so as to not be dependent on one, still has powerful and defensible network effects, a sustainable revenue source with growth potential…….i feel like some specifics (if they are available) would help move the conversation forward.
Well Andy’s post which i linked to has a bunch of examples in it. Its unclear if you would call them compelling
ahh, thanks for pointing that out i overlooked it. with all of those i wonder how much they can grow until they need their own stack. textrex seems the most compelling to me, but if the market is there the competition will require innovation that will lead someone to build a stack to improve the user experience in my opinion
Fred,One of the examples of a no stack startup listed here http://bit.ly/1cvxpO6 is The Shade Room. Didn’t Instagram just shut it down?What happens when a big company you’re using as your stack decides they want to be in your business? Now I don’t know if that is the case with The Shade Room or not. But it certainly was the case with Meerkat and Twitter.
Rick, the key to a no stack tech play is to accurately assess the 3 to 4 critical value components.You have to control those components. In most situations, the likely answers are:- brand- UI- network of users- sales (debatable?)
I don’t think there’s such a thing as tech defensibility any more. Everything can be reverse engineered. It’s the ability to connect with and delight your customers now that makes the difference. The ability to inspire loyalty. We’re deep into Seth Godin territory now.But there is definitely a vulnerability that comes with relying completely on third-party services and technology. The ‘moat’ around that has to be Plan B, C, D, etc. in the event that a service dies or changes its policies in a way that cuts you out. Services like Twitter are constantly evolving their usage policies, and it can be painful, at best.
Yeah turning off the firehoses of intgration / data pulls can kill startups, as we’ve seen with Twitter’s recent moves. One thing that doesn’t get discussed enough is the effect of a platform that incentivizes community growth, that’s the stuff that Seth deals with in part. I don’t know how co’s like Box and Uber position with VCs, but you can bet that whether they use the word community or not it’s what drives a lot of the network effects. Heck, even something as simple as “refer a friend to get 5MB more folder storage” involves community in some aspect.
for the no stack – i think you want to see unique emergent properties arise from the system. This makes the experience unique and compelling – and cannot be easily copied
Owned & operated lost out?its the best acronym – OnO
It’s a topic I’ve been thinking about a lot as well, but I think the related theme is “Changing Stacks” not just no stacks.New stacks are replacing older ones, but in different ways and via open protocols too.What’s also important is how the stack intersects with a company’s value chain and that’s where the defensibility probably gets uncovered.
I could argue that my company is building a no-stack start up (i guess theirs a spectrum of how “no stack” you are?) but to your point about defensibility/margins/not having control of infrastructure – those all seem to the point the notion that if your no stack startup sees even a modicum of success, how do you make sure the competitors (the stack you’re plugged into) doesn’t copy and overtake you. It’s a fair question and I’ll use my beliefs about my business as points that can potentially translate to other businesses you see:Context about us:We’re building an predictive analytics software for marketers that does one thing very well and we integrate w/ every marketing channel (ESPs, FB, Push, Analytics Provider, Mobile App Trackers, etc.) for data and actionability.We don’t have code in websites or apps to collect customer data directly and we don’t provide the tool that sends messages to the customer. We hear plenty of times “why doesn’t X Company do what you do then?” but I have reasons to believe we can sustain the above value over time for the following reasons .1) Positioning (“we are unbiased player”) => this eliminates a lot of the of competition that can never make that claim (i.e. a Facebook PMD) b/c they already have a “dog in the race”2) We are also selling our Product to sit at the very CENTER of the marketing team where the Email person, the FB person, the Push person, the head of marketing, etc. build workflows around what we do. By controlling the workflow and having a team (not an individual) owning the product, you build stickiness.=> This is obviously an ideal situation where you are the central workflow but you can tell by the entrepreneur’s vision if their no-stack start up is more an add-on to the existing stack or focal point of the stack.3) Anyway your single aspect of the stack can leverage network effects in one way or another is obviously makes it hard for competition to copy4) The part of the stack you’re focused on his really really hard to replicate technicallyHope this helps!
“No stack” means you’re outsourcing and relying on each modular component of the stack provided by someone else not to fail (or to recover quickly during any failures because they have an ace tech team to troubleshoot it). You’re plug+playing into their system.It works really well for some types of startups and less well for others. An example of “no stack” is if you’re doing an Instant Messaging service which isn’t changing infrastructure via protocol innovations and all you need is AWS.”Full stack” is more necessary if you’re inventing from ground up, from UX-HCI on the front through to analytics and how the dashboard presents.
One step further: I’ve mastered the no startup startup. They’re highly scalable, high margin, defensible ideas that never leave their addressable market: my imagination.
A lot of people I know have mastered this startup. The real problem is that those startups dont have any revenue :PThis is nonetheless the way of life for few.
Are you starting by attacking the conscious or unconscious part of your market? I hear that somewhere between 80-90% of brain activity is unconscious so if we value your unconscious mind at $10bil (seems reasonable, no?) then even 1% of that is a strong growth track to aim for.
We all have a decacorn in our heads.
See my comment to Joe.Google can only monopolize 10% of our brains at most, haha.
Decacorn, anatomy porn! +10
Can we say then that our conscious mind is measurable in the action/behavior of clicking, reading, sharing, buying and all behavioral heuristics are only 10% of the whole potential market of our minds?In this case, Google can at most only dominate 10% — so much for Thiel’s monopoly theory and…That’s 90% of brain activity still wide open for innovation of tools and, potentially, next generation Googles?
We really do need someone to write a Philosophy of Clicks & Impressions…and they said it was a dead art..
Joe — it’s not obvious but it’s very much alive and in action. FB puts into practice its Philosophy of Clicks and Impressions in a very different way from Google’s.@fredwilson:disqus talks about “network effects”. Well, FB’s is rooted in social philosophy first whilst Google’s is in mathematical philosophy first, so their full-stacks are oriented towards poles apart imperatives.Read this Guardian article from 2008 which shines a light on Peter Thiel, the first investor in FB:”Thiel’s philosophical mentor is one René Girard of Stanford University, proponent of a theory of human behaviour called mimetic desire. Girard reckons that people are essentially sheep-like and will copy one another without much reflection. The theory would also seem to be proved correct in the case of Thiel’s virtual worlds: the desired object is irrelevant; all you need to know is that human beings will tend to move in flocks.”* http://www.theguardian.com/…Moreover, Zuckerberg’s mother is a Psychiatrist and he was majoring in Psychology with a minor in CS. When we read about FB’s emotions experiment, we gained an insight into their philosophy of clicks:* http://www.wired.com/2014/0…* https://www.youtube.com/wat…Also, add Sheryl Sandberg’s economics background to FB’s Philosophy of Clicks and Impressions when we read about their Atlas advertising exchange.Meanwhile, Google’s Philosophy of Clicks and Impressions is much more mathematical because of its Machine Intelligence and Search leaders like Peter Norvig, Amit Signal and Ray Kurzweil:* http://norvig.com/chomsky.html* http://www.slate.com/articl…* https://www.youtube.com/wat…Other teams will emerge whose leadership and investors have a different Philosophy of Purchase Intent than FB’s, Google’s, Twitter’s, Amazon’s et al.Notice I didn’t write Clicks and Impressions. There are inventions I’m aware of which take user interaction away from mouse+keyboard and even touchscreen swipes.
Interesting you should say that, I almost responded to your original comment with something satirical but then it made me think a bit more…I agree there are some very different approaches, and the way that you laid it out suggests something I’ve been thinking about, which is that there is room for other large players to reach attention / value / trust economy (as Seth Godin lays out much more elegantly than me). Google, Apple, and FB aren’t reaching nearly as much of the possible market as people think, if you admit that market is more about attention and trust than just total #s of people or impressions.
The market is more about how people SUBJECTIVELY QUALIFY anything.The clicks and impressions are external behavior outcomes.The subjectivity is part of the 90% in the conscious and unconscious parts of the brain not yet capable of being measured by modern tools.Which opens up opportunities for invention.
The internet is my stack.
Reposted from my reply to @gbattle over at the same discussion at the usv.com topic of the week conversation.Owning the last mile of experience isn’t simply the UX/delivery, it is the content itself. Ergo, the message is the medium for a No Stack startup, not the other way around.Agreed but this “No Stack Startup” is two mints in one as it fit just as well into the “Medium is the Message” paradigm !This “platform-as-medium” is also in reality the latest instantiation in an ongoing message stream about human organizational evolution, as were smoke-signals, the telegraph, radio, TV, the internet and now these democratizing platform-based remix-ponies make that universal remix-pony evolution-message more accessible than ever to the rest of us.Surely in the long run the magic-powers inherent in developing evermore simplified/accessible/universal remix-ponies can be liberated back into an internet of universally open-APIs as social utilities.Hopefully private ownership of our universal remix-pony’s magic-powers will be just a necessary experimental development-phase an not a permanent corporate lock-down constraint upon our new organic-level organizational remix opportunities?
Getting pretty deep there, but i think i get you :). What comes after the internet as we know it? And will that be allowed to happen by private ownership interests?
What happens after the Internet?Maybe this: brain is the ultimate interface* http://techcrunch.com/2015/…
Maybe it’d be helpful if Fred and community can provide examples we believe to be “full stack” and those we think are “no stack”?
andy’s post has some good no stack examples and chris dixon’s post has some good full stack examples
Andreessen Horowitz is a full-stack VC whilst USV is a no-stack VC.They started to build up EIRs, marketing people, PR people, product people etc as well as the investment partners to support their investments–Maybe around the time Chris Dixon made that post about full-stack because A16Z was themselves becoming full-stack.So the internal question for USV is not just whether there should be a migration towards investing in full-stack startups but also what hires would be necessary to the USV team to gain domain expertise on the parts of the stack USV hasn’t previously focused on?What are the value gains and opportunity costs?I will say that for the next wave of Machine Intelligence-related startups, the full-stack plumbing is started to be done so in 3-5 years time there’ll be no-stack opportunities.However, for now, the full-stack plumbing is being done by the Googles, Facebooks, Baidus, IBMs etc. and the team that founded Siri.
I would challenge the assertion that full stack and low burn rate are mutually exclusive.
Agreed.
we haven’t seen much evidence to suggest otherwise and we see a lot of companies and their financial models
The no stack startup facilitates product descriptions in the form of “this for that”, as has been discussed here. IOW, using 3rd party APIs and such, it makes it easy to describe what you’re doing as, say, “Uber for dogs” because you built it with Uber’s API and canine protocol (ArfTML).
I think there is at least a subtle difference in the use of “stack” by Chris and Albert.I think Chris is talking more about the business stack — i.e., should a company be vertically integrated? 15 years ago the business sages and consultants were telling companies to de-bundle, outsource non-strategic parts of the business, and focus on “core competencies”. This led to stupid decisions (IMO) like deciding customer care/service was not strategic and handing that over to a 3rd party who would be measured on SLAs, almost entirely around efficiency/cost. Now companies like Uber and Nest are realizing they need to build the business from soup-to-nuts to create the experiences they think will differentiate them in the market.Albert’s post – “No Stack Startups (the Technical Side”) – is, as the name implies, about the technology stack. What I took away as his big point was that with web services and APIs one can stitch together a business using publicly available components without having to build a back end in, say, Rails.While I think these are closely related (now) and Andy brings together the two ideas, it seems like the term “stack” is being used a bit different contextually by Chris and Albert.Would not it be possible to “build a complete, end-to-end product or service that bypasses existing companies” using existing technologies and APIs as Albert describes? I.e, a “full-stack” vertically integrated company that’s technical infrastructure is built almost entirely from existing platforms available via API? I take “existing companies” to mean market players especially in the distribution/route-to-market side, not technology platforms.
I agree, the subtle distinction in business and tech stack matters.Andreessen Horowitz is a full-stack VC as McKinsey is a full-stack management consultancy as Apple is a full-stack technology company.USV is a no-stack VC as Accenture is no-stack consultancy as Tweetdeck is a no-stack technology play.
You are doing that Twain thing again. Bang on.
Question is, can he now turn ‘that Twain thing’ into a no stack business on the back of disqus? 😉
Twain’s a she as JamesHRH knows because he’s seen me IRL as have Jim and Shana, :*).Disqus is a no stack startup. It taps into WordPress codex stack.All commentators generate valuable content. So the content’s the no stack business.Maybe I should start randomly doing brand and product promotions like the Kim Kardashians of the world do in their tweets, haha.
Sorry for confusion, I immediately thought “Mark” when I read “Twain”. Disqus as no stack business of course! Q. Now is – who or what will you promote??
CH-OC-O-LATE!!! That and Vodka and super-cars.No, in seriousness, I have 5 pet subjects:(1.) Human+Machine Intelligence.(2.) Data, data and data.(3.) Corporate strategy.(4.) Contributions to global society.(5.) Inclusion and integration in Tech.
And here I thought my bugatti powered by chocolate-flavoured vodka would never catch on ;). Nice list. I’m in Africa and contributions to global society seems much needed but tricky to implement.
Maserati Birdcage over Bugatti Veyron just as in silver birch over palm trees.
Big – our Twain is as original as the first one who made a Mark for the name!
I’ll take your word for it. I’m seriously intrigued by her comments for sure. If a developing world perspective is wanted for her list of interests, I’m your man 🙂
Thanks, that’s one of the kindest compliments anyone’s ever given me!I’ll endeavor to be more worthy of it!
I would love chat sometime on 1,2, and 4, which is the nexus of what we do.
Super, thanks! Now there’s a “Twain thing”, :*). Nothing’s a “thing” until someone else spots it!!!Recently Macys tweeted an “Honorable Mention” of my team’s IoT app based on my mad ideas. An A-HA moment dropped and one of my team sent an email, “Now we’re a THING!!!” Lol.It’s no-stack and taps into Twitter Fabric API so I get what @fredwilson:disqus is saying and agree wrt network effects and ability to start with very little financial capital (factoring in, of course, that in terms of people and IP capital all successful startups kick off with a $1+ million base of skills talent and intangible assets).Macy’s Director of Digital said he’d “never seen anything like this anywhere before” and we got a handful of testimonials, including from Twitter’s Developer Advocate. In fact, Macys sent their tech team reps round 3(!!!) times because they couldn’t believe what they saw(*).They told us to enter our name for their hackathon prize — even though we hadn’t used their API! And then gave us a shout out on Twitter.That hack’s now Startup #2. (*) There’s something in our pipeline that literally is “Can’t believe what I’m seeing!” and you won’t need a Hololens or any other VR headset to experience it.Startup #1 is full-stack and a LOT harder to do. I agree with David Semeria, though. Full-stack and low burn-rate aren’t mutually exclusive.If someone else tried to build Startup #1, it would cost them $ millions to hire the skills needed.However, doing the “Twain thing” means it can be built for a fraction of that. The fun of hacking and integrating all the nodes in your own mind! Lol.
i like it when technical and business architectures align
Stack alignment between business & tech is a “Holy Grail”. It’s what I think about and work towards every day.The full-stack vs no-stack decision isn’t a trivial one.Originally, I too wanted to aim for low-hanging fruit of no-stack for the same reasons as other founders (low capital requirement, faster to implement, network effects).In 2010, a number of the startups had began to release APIs. That Oct, a friend and I tried to mash-up a movies & music recommendation engine that pulled in data from Chartbeat, FB profiles, Hunch, Yahoo BOSS and IMDB.There were so many issues with their APIs back then! Hunch’s CTO sat with us and we flagged up all the callback and JSON file problems. Chartbeat we couldn’t do anything with since their Dev Advocates had left before midnight so weren’t available to help us troubleshoot their issues.Six months later, in March 2011, Chris Dixon released Forage which was…music recommendations using the Hunch API, Twitter and YouTube.Between that and an incubator telling me to build a commenting system similar to Disqus (also no-stack) and any number of API mash-ups I’ve seen but not been inspired to copy (I’m not a “me too” type of founder: I like to make original stuff)…I realized the options were:(1.) Do no-stack and “flip” the startup in 2-3 years;(2.) Do full-stack with a view to solving a seriously hard problem that would use everything I’d ever learnt how to do and that would be a LIFELONG adventure.(3.) Leave tech, get married, have kids and re-train as an interior designer since everyone assumes I’m a creative when they meet me, anyway.I decided to solve the seriously hard problem despite all the “pain in the ass” experiences. It’s very character-building when we push the envelops of products and ourselves, haha.It’s so bleeding edge it will take Stanford’s AI Lab folks another 5+ years to get to where I am — even just in terms of frame working what the roots of the problems in Machine Intelligence and data intelligence are.Fei Fei Li the director of Stanford AI Lab said in a PBS interview last week: “We are very, very far from an intelligent system, not only the sensory intelligence, but cognition, reasoning, emotion, compassion, empathy. That whole full spectrum, we’re nowhere near that.”(http://www.pbs.org/newshour…The spectrum part NEEDS the technology I’ve invented and I know this — even if the market doesn’t (yet). Neuroscience research is vindicating the bet I made a few years ago not to solve the problem with legacy paradigms and tools but to do it differently.It’s all good. I can invent and code no-stack products and startups and get married+have kids until the market’s ready.And I know from experience that full-stack isn’t necessarily cash-intensive.It does, though, need intelligence intensity and integration.
The no stack approach is brilliant for starting up from scratch, building something and finding product market fit. But I think that eventually, as the company grows and scales it will have to start bringing some stacks in-house.You can see this in how Google, Facebook and Apple ended up building their own datacenters…
I have a friend on Instagram who runs her high fashion business purely by positioning her camera in the right place and building a network around the image and lifestyle she shares online. People pay to be mentioned on her instagram account. I don’t know of examples which take this concept to another level of scale though. Enabling the networks creates scale. The networks themselves are limited by their niche interest groups no?
Exactly wrt network effect limitations (niche interest and it can only grow so big) as I noted to Jim in comment further down.It depends on which part of the network you want.
In my experience, it was important to start out as a ‘full-stack’ company at first in order to understand where our real value was, where we were the experts, where we competed best, and where we wanted to build the long-term defensible foundations of the company, before being able to transition into more of a scalable ‘no-stack’ company.We provide a back-office API for building online securities/crowdfunding/p2p marketplace platforms. At the fullest height of our stack we were involved in developing specifications, design, branding, strategy, front-end build, back-end implementation, customer support, and hosting.Providing all of these services ourselves was important in the early days for:(1) generating revenues before the long-term business model kicked in; (2) understanding the requirements for the end-to-end service and how each part would work together; (3) testing out different product and service ideas in order to understand where we would be able to take the best position.Then, having gone through that process, we were able to identify that the company’s real long-term value is all tied up in one narrow area – the API and its back-end business logic – which is now our exclusive ‘no-stack’ focus, and we brought on expert partners to handle the rest.
Great comment, and agree. There are some exceptions but generally in early stage the phrase “full stack” could be exchanged for “ecosystem” and trying to scale / grow the snot out of something before you understand it and where it plays in larger ecosystem is typically unwise.
Very interesting, thanks.
In case someone is wondering what full stack means, here is a helpful link: http://www.laurencegellert….
As an early stage/self-funded entrepreneur, you spend so much time testing out your assumptions that it’s really nice to live in a world that you can build things “No Stack”.There are so many tools out there that get you to a place where you can launch something without building a heavy product (Twilio, Stripe, Typeform- to name a few). We live in an exciting world where people appreciate solutions built into things they use in their daily lives (Instagram, text messages, etc).For us, (early stage entrepreneurs), that allows us to tap into users existing work streams and offer solutions that branch off of that. It’s an exciting place to be!
Well as Chris says:The good news is that if you can pull this off, it is very hard for competitors to replicate so many interlocking pieces.That’s a big “if”.A matter of money, time to market, logistics, having the right people and a host of other things. No hard and fast rule and I’m not sure how anyone can advocate for one or the other concept unless the specifics are fully understood. This is kind of like the classic business school learning which doesn’t take into account any nuance or specifics. (Not a rant on Chris by the way.)Chris gives an example of Harry’s (I’m guessing because they bought the factory that makes the blades). While on the surface that would appear to give you more control (and should) there are cases where you might actually suffer lower quality and have less control. Why? Because now the factory manager (as only one example) is your employee and you have to be the bad guy in firing him and have the aggravation of finding a replacement. [1] You can’t lean on someone else to do various dirty deeds, losses and aggravation that would come if you were using a vendor (and out of their pocket) it now come from your pocket. If the factory is inefficient it’s your factory so either you upgrade it [1] (and can afford to) or you are stuck. If you are using someone else’s factory in theory you could just “find a new factory” that already had the modern equipment. And so on. Or force your will on the vendor where he gets the sleepless nights.[1] Take the issue of “upgrade”. If you are an important customer of the factory you can shift the risk of buying a new machine on the factory owner vs. you taking the risk. If things don’t work out he is stuck with the cost of the new machine, not you. An example of this might be Walmart requiring all of it’s vendors to buy bar code machinery and bar code label all products. Didn’t cost Walmart any money, all born by the vendors.
full stack = Cold War era; US Military mixed stack Nigerian army and South Africa merceranies ; no stack Executive Outcomes in Equitor New Guinea
interesting analogy
I believe there is a huge value gained in understanding how the “stacks” are built, and essentially having a overview understanding of how you structure to scale. To give a simple example, we’ve moved our company from Living room > Accelerator > Offices in different countries, we bought a network based storage device and WiFi router so that wherever we worked we would have fast access to our files, and as we’ve scaled up it’s been no effort to transition the team to the larger network storage as people are used to keeping their stuff there.I know there is cloud, but from a simple cost point of view, for the TB’s we have, it’s way more effective to build a bit of an IT stack etc.
I’ll take the bait.no stack startup is a non sequitur. it doesn’t exist. great branding yet no basis in reality.there is a stack, you just don’t know what it is. there is a stack you just don’t have any control over it.there is a stack it just doesn’t belong to you.more accurate name is the leveraged/opaque/crazy-man stack startup.for those who’ve been burnt before building on others infrastructure only to be told one day “hey kid, get off my lawn!!” – the idea of a leveraged stack startup is as appealing as getting poked in the eye with a rusty fork. Twice.for a VC firm that coined the phrase “don’t be anyone’s bitch” this is exactly what building a leveraged stack startup makes you.for a VC firm steeped in encouraging the values of open source and public protocols, building on a black box foundation is an oxymoron.for a VC firm whose war cry is: “Large networks of engaged users, differentiated through user experience, and defensible through network effects, leveraged stack startup makes you impotent.no access/ownership of the large networks of engaged users. no control of the user experience and no means of defensibility as the network effects are not yours.the old mantra was be the gorilla. the new mantra seems to be, be the fly hovering on the gorillas back.makes perfect sense. until the gorilla kills you.#Bah #Humbug.
harsh. but hey, the truth hurts.no stack is appealing as a low cost trojan horse entry strategy of sorts. once you have something to protect, though, you’ll need to develop your own stack (even if that’s not what you call it).
I agree with a lot of this and have been kicked off the lawn, as well.What I think I’ve learned is that when you build something that leverages other services, you’d better make sure that you add value for those services as well as for your users. And, have a Plan B.
The more value you add the quicker you rouse your hosts jealousy/anger/apetite, the quicker you get whacked.Whole endeavour is self defeating
I’m thinking Zynga.
Fair point. It will be interesting to see how Meerkat does. They claim to have Plan B, I think.
to be fair, Fred did essentially make the same point when he said ” We like low capital requirements and low burn rates and extremely high rates of return on invested capital.so the no-stack-startup is an investor utopia/ideal. Maybe the no-stack-startup should be how we lay out a null hypothesis but not necessarily the actual product vision?
That’s a bit extreme Liad. Twitter and FB, for example, never set out be in the “stack supply” business. On the other hand, companies like Twilio, Dropbox, etc make this explicit in their TOS. In other words, the data is yours and as long as you pay on time, they will never kick you off.
TOS are not irrevocable. They can always kick you off
Actually that’s not totally correct — a party cannot unilaterally change a contract. This is why, for example, Google shut down its first version of Translate (which was free) and created a new version (which did exactly the same thing) with a payment-focused TOS.But it’s irrelevant anyway. My point is that if you’re explicitly in the stack supply business, you’d have no reason to kick paying customers off.
I was just thinking about Amazon who both hosts (stack supply) and also delivers various content and utilities (consumes successful clients).
“stack supply” businessIsn’t that the inevitable evolutionary trajectory here ?Evermore simplified/accessible stack-remix-ponies at every level of automated organizational assemblage.The history of computing/network-effects is to hide the variables until even idiots like myself may be able to master many once complex organizational assemblage processes.Desk-top-pulishing —> Spreadsheet Construction —> Website construction —> etc.
Yes, Albert calls this IFTTT programming. But I think it’ll be mainly limited to prototypes. People have built some really cool stuff with Lego, but a commercial product is another beast entirely.
yup, that’s a lot of where our thinking is around how to invest in this category. we’ve already been making a lot of bets in the “stack supply” business and it’s a good sector
correct
Is a no-stack startup *dependent* on someone else’s platform? Is that what the consensus has decided?
That’s one of my takeaways from the examples given and the fact if you’re not building it yourself you must be using someone else’s.
As I re-read Andy’s post, he seems to assert the same thing. But I’m starting to disagree, quite heavily, that this assertion is a good thing.Let’s take a “no-stack” business that was given as an example: a fruit stand that sells via Instagram. The business should be defined by what it sells, not where it sells it (a theme I’ve trumpeted on AVC for some time now). If you sell fruit, you sell fruit. If Instagram kicks you off, you can go to Facebook to sell your fruit. If the whole Internet abandons you, you can sell on the street. But at the end of the day, if you’re building a solid business, you’re going to build your brand and loyal customer base around having good quality fruit. If you build your brand around where you sell the fruit, you weaken yourself. You want people loyal to you, not where you are — because if you move, you want them to follow.So if a no-stack startup is defined by where the product is found, vs. the service or goods the product offers, then I want no parts of that — nor should anyone else. As you’ve already pointed out, that’s a dark path full of terrors.
I take your point but I still hear echoes of that old sales clicheLocation – Location – Location
Right on. Except that the platform you choose is quite important to define and deliver your customer experience. Yes, the fruits need to be perfect with good quality. Nevertheless the platform you used will determine a big part of your customer experience. It’s a long way from me selling fruits to you eating them.
I think it all depends on where, how, and why you start…I think NSS gives you the quickest, lowest risk, lowest cost way of trying a lot of things out and giving you the freedom to fail A LOT with as little pain as possible ( http://www.brainyquote.com/… )So if you are just looking to build an income, NSS is probably a perfect start. If you are looking to build a business, it’s probably at least worth looking into.However, if you are looking to build a ‘startup’, then I probably agree more with you in that it’s going to be *really* hard to be disruptive without doing something special in the stack…though it could still be leveraged for rapid prototyping and the proof-of-market and proof-of-concept phases as you build your early converts….just my two cents (now available at half the cost and double the value)…
If a core audience can be built while squatting on other businesses stacks, and that core audience will then follow the brand elsewhere it’s a great marketing hack. But it’s not a business.I’m trying to think of any enduring success story for a business which has thrived and gone big with no control of its delivery. Maybe Zynga?
Yup, I wonder what happens when the stack below you stops supporting you or pulls the rug from under you.Also, it might get “expensive” to make a lot of API calls continuously if that’s what it takes.
yes if you are building on facebook or twitterno if you are building on twilio or stripetake a quick look at Albert’s deck. i think he makes this point wellhttp://www.slideshare.net/A…
Deck was insightful. Conversation here seems to have missed your point.
On last slide: 1-person 1-week 1-million user startupThe other day Chris Dixon tweeted about 1 billion user startup with a solo entrepreneur.Now wondering if he thinks that will require a full-stack or no-stack.
Then, it gets a little more complicated because there could be a hardware stack or software stack, or a hybrid one. Key Q on my mind is – How do you “replace” what a stack does. You still need functionality.
How about answer this question first: is “not providing functionality” a key aspect of the no-stack startup? Or is it, rather, about providing more narrow, targeted functionality?
It’s all a matter of preference. Does the company want to own and control every aspect of their creation or do they just want to license their idea/product and sit back and reap the rewards? There is nothing wrong with either-or.
Each individual scenario has a context, and this broad conversation isn’t really getting deep enough to evaluate that — as you’ve pointed out.Define “control every aspect.”Does Microsoft Word not “control every aspect” of what it does? It would be easy to claim “yes,” if you consider “what it does” to be word processing. But you’d say “no” is “what it does” is provide office utility — and you’d say that its the no-stack to Microsoft Office’s full-stack.Semantics matter here.
I just want to talk about this from an engineering perspective for a moment. There’s a couple of different concepts here to talk about. Firstly, “reinventing the wheel” and “glue code”.Starting with reinventing the wheel. Do you remember when open source wasn’t really a thing? When there was no GitHub and hell no one ever really looked are sourceforge. It was also a time when maintaining competitive edge meant writing code in secrecy and having your own proprietary technology. Times are changing. We’re seeing more and more of the foundational layers of technology become available either as services or as open source code. It means that we’re not having to do that work again, we can build upon the layers of abstraction.Starting at the bottom of the stack. I don’t have to go rack a server in a datacenter anymore. Now I can spin up thousands of VMs on AWS using an API. Heck AWS Lambda allows me to write a function that’s executed on some event. I have the choice between spinning up the equivalent of a machine or merely thinking about logical execution of code. Soon the levels of abstraction will be about containers or unikernels and we’ll no longer be worrying about machine management, but luckily we’ll be able to govern largely intricate and complex systems using microservices architectures within this world. If you want to go all the way to the edge there’s Google App Engine which takes away everything but really makes it difficult to build SOA style architecture.Building upon that. We have the notion of “glue code”. Because we’re no longer having to “reinvent the wheel” or write those fundamental building blocks from scratch, we can essentially just write the specifics of the code that glues together the necessary applications/services we care about. At a high level you can use existing APIs for things like notifications, sms, emails, etc. Companies like Mashape provide easy integrations for these things. And at the lower level if you’re looking to compose something like a microservices architecture based on containers and distributed systems you can essentially leverage systems like kubernetes for container orchestration and cassandra for a distributed datastore while writing your glue code to handle discovery and client/server communication. An example is a microservices library I’ve written called go-micro, https://github.com/myodc/go…, essentially glue code with pluggable interfaces providing a way to write and manage microservices.We’re in a new era where the focus is primarily on higher level abstractions and delivering value beyond what we’ve done for the past decade or more. The notion of “full stack” will continue to evolve over time. There was a day when owning your own datacenters was part of being considered full stack. Uber is more than likely running on AWS. Are they full stack? Well from a business perspective sure, from a technology perspective, perhaps but only if we’re reclassified the term full stack.
As I read the comments, I find many of them, as well as the concept of no-stack in general, to be quite interesting. But part of me finds the conversation, as a whole, futile — because it’s wrapped around the new buzzword du jour.I wonder if the positives that come out of having this conversation are negated by the issues that come from the constant force shifting our collective weight from buzzword to buzzword every quarter or so.
#truth
@fredwilson:disqus I am not sure you proofread this post to make sure it had a suitable jargon to information ratio 😉
i’m sorry about that. i plead guilty as charged
Low burn rates, not to confuse with lean..If there is a certain cost to providing support for the full stack and no platform or service exists yet (because of this very reason) then the opportunity exists, and the opportunity is very large then because of this same fact.
Spoke with a startup today, and we had that conversation. Currently they are full stack, but aren’t opposed to outsourcing parts of their business so they can focus on what they do best, taking care of customers.
Anyway, yes – it takes more time to be provide the full stack (and more time equates to more money for runway and planning) – though the power of once you have succeeded can easily wipe out all competitors – you simply can’t compete on offerings.
The way I see the solution to this dilemma is: small business/one man company is more than welcomed on platforms (no stack required), win-win scenario; unicorns and 100 years companies need to be full stack. Otherwise, one day…http://www.bothsidesofthetable.com…In other words, I don’t think Huffington Post could have thrived on the back of, say, Facebook.And this topic made me think about all the innovation in the payments industry on top of cards/banking platforms. My humble opinion is that the existing successful start ups may reach a day (read market share/revenue) when Godzilla wakes up and shakes its tail.
Hi Fred, I have noticed on other blogs that it is possible to publish comments in one place from different sources e.g. your blog and the USV blog, Sorry I don’t know how it is technically done, perhaps someone could explain if this is something you would like to do e.g. with this post.
you use the same disqus comment thread ID
Looks like the “no stack” startup is the chromebook of startups 🙂
i love chromebooks
Some of this is semantics. Every startup needs some proprietary-ness that is their secret sauce, be it workflow, marketplace, distribution, algorithm or slice of the technology stack.I guess my point is that people get into these false dichotomies of full stack v. no stack, when it’s really about differentiation. How much of what stack do you need to build to create a differentiated, defensible position.
I like the idea of a ‘no-stack’ company but I think one of the biggest questions I have on the success of one is competition. Since it is fairly easy now to build on top of other platforms, there’s little barrier to entry. You could argue network effects might tell who the winners or losers are but if the only real difference is the number I send an SMS to, there’s very low friction in moving to another service. Good for consumers but bad for the business. The reliance on another platform is also concerning, as facebook/twitter have shown in the past, these platforms you build on top of could cut off your access tomorrow without warning upending your entire business. These are some reasons I still think the full-stack business has greater long-term business potential then the no-stack.
This should be herehttp://www.programmableweb….
Fascinating trend observation. Interesting how VCs are watching this trend, because I can’t see them make big returns from investments in no stack startups, because they seem small. Or am I getting this wrong?
We can also flip this to look at app stores and YouTube as partial-stack media companies without the content production portion of their stack.
I understand why “full stack” is not part of USV thesis, but I disagree with you that it’s only for deep pocketed investors. Actually in my view “full stack” is quite suited for bootstrapping entrepreneurs, it’s often cash generating from day 1 and can be sold as service while you build a more scalable productised version. I am thinking here of a full stack restaurant, law firm, head hunter etc.. many of the examples made by Chris (or especially by Balaji in follow up https://storify.com/nikcub/…
I don’t think there is an easy answer to this. With each new generation of Internet development comes a new herd of “infrastructure players”. They are usually quite effective in avoiding the limitations suffered by their predecessors and build new rules of engagement that resonate with their values and visions. What it means is that they produce a better “ecosystem”. The real challenge is whether they can build a business model that both satisfies their hunger for dominance and at the same time enables others to grow (the very definition of an ecosystem). I think at the core of it is the infrastructure players’ ability to sustain innovation and continue to power the ecosystem that allows them to “do no evil”. Once they start trailing, they will very likely look sideways for profit growth opportunities and end up hurting other players in the same ecosystem. In short, it is what they decide not to “own” that determines the habitat of the ecosystem. What it also means is that they’d better get really good at what they believe they need to “own”.
Fred, if you read this I would be very interested if you could dedicate a post around entrepreneurs who hire a developer company to build an app for them. Typically, it was established companies hiring developers to make an app for their existing business. As you know its a relatively newer startup model, that aligns entrepreneurs with a core team including; front end/back end engineers, a mobile programmer, & creative direction. Once the MVP is built, if the app can gain traction and you can begin building the product & company to scale. Are you seeing more companies lately that have started with this approach? Do you believe that the core founders model have/can change? Can these types of models open up doors to people with no coding background but a strong idea/business sense to be aligned with great programmers?
I dont like this model. I think i posted about that before. Will try to find it
That would be greatly appreciated. There is little available information, and I have yet to hear of a successful company that began with this model. It seems sound providing the developers provide all necessary resources, and alleviates the process of having to find several cofounders to put forth time & effort in very early stages.
I believe that the “full stack” approach was very relevant in years that passed. However, in 2015 there are more mobile apps, (ultimately consumer choices) than ever before. It would be unlikely to think each company competing for the barrier to entry can succeed only using a full stack approach. “No stack” companies have the ability to be lean and focused which in today’s generation can prove to be a significant advantage over others.