Startup Churn

We encourage all of our portfolio companies to measure their churn rates by cohort. It is very revealing.

I saw this tweet by Liad this morning that shows startup churn by cohort.

I don’t know the source, but the data is sobering.

Some of the churn is companies getting sold. Some of the churn is companies getting profitable. But most of the churn is companies failing.

We have looked at our portfolio this way and our portfolio has performed much better than this. Some of that is selection. Some of that is support. And some of that is tenacity of the founders.

But, as Liad says in his tweet, startups are no cake walk.


Comments (Archived):

  1. William Mougayar

    Same will apply to blockchain ICOs who are painting a rosy picture for eternity. What matters is how the CEO keeps steering the boat despite the churn.

  2. jason wright

    “% of startups reaching next funding round” – i would like to see % numbers that describe “sold”, “profitable”, and “most”.D = Dilution?

    1. Rob Sobers

      I was thinking the same. Just because they don’t raise another round doesn’t mean they’re dead or dying.

  3. JimHirshfield

    It’s a funnel… Like a sales funnel. Lots in the top, few come out the bottom.

    1. pointsnfigures

      Of course, if a sales funnel looked like that most VCs wouldn’t invest.

      1. JimHirshfield

        Most VCs don’t invest (from the Entrepreneur’s perspective)

    2. ShanaC

      that might just be life.

  4. heuristocrat

    It’s a little weird without the source. It also would paint a very different picture if the declines are due to profitability or being bought out. Of course for a VC looking to do additional rounds as the company develops I guess it’s not good at all.

  5. Tom Labus

    Are we at the end of a cycle with less viable companies being created now?

  6. creative group

    LIAD:This is earnings season and for those of us with skin in the game very important.”Sobering Data” Fred quote:The medium you used (Twitter) reported a 5% sales decline on 575M. The more important information is that the company reported no growth in monthly active users for the last three months. About half of the years gains has been lost on that news.How about that for churn.Again Jack Dorsey is an excellent visionary. That vision is being lost at his instance to run two companies.Twitter stakeholders appear to be going along to get along in not having Jack concentrate on Square only.DISCLOSURE: NO TWTR HOLDINGS

  7. Richard

    Keep the chart going into performance in the post IPO stage and the trend seems to get worse.

  8. Gregory Magarshak

    The USV portfolio always struck me as having an outsize share of well-known companies. I wonder if it’s partly because you guys are the best known name on the east coast!It would be nice to see a breakdown of how much the comapanies actually made in revenue, though, and in what years. Twitter famously held off for years. Meetup from what I remember Heif telling me had one year where they bit the bullet and traded “virality” for profitability.

  9. Matt Saunders

    the chart doesn’t show # of companies per year – thus hard to gauge whether the top of the funnel increased significantly, (more co’s thus lower % reach next round). In addition what was the trend of in M&A activity were there more exits before co’s reached C and D rounds?

  10. LE

    I don’t know the source, but the data is sobering.Sorry but my first thought was how can you put any validity into something when you don’t know the source and/or the scope? (God knows I question everything even if the source is considered ‘of record’ or reliable in my cynical and skeptical way.)Then I see from the link that @liad:disqus provided:This report covers venture capital funding in Europe and Israel (including Russia and Turkey). Includes: angel & seed rounds, VC series, growth equity, converts

  11. creative group

    CONTRIBUTORS:OFF TOPIC ALERT:BITCOIN NEWSRussian Alexander Vinnik was arrested in Greece on suspicion of being the Key person in laundering over 4 Billion via crypto-currency BTCe exchange. Also accused of being involved in the MT Gox theft.

  12. sigmaalgebra

    > startups are no cake walk.For my startup, I’m not seeing that. My startup looks like a cake walk to me.For a given startup, the probability of project success estimated just by taking a simple average from all observable startups as in the data here means essentially nothing. Instead what is meaningful is the conditional probability of success given (in probability theory) events, i.e., in more common language, circumstances, data, etc. that apply for the startup in question.In simplest terms, does the startup team know enough about what the heck they are doing and/or can learn that quickly and well enough to be able to do the work with high quality, successfully, and routinely, i.e., as a “cake walk”. That is, pick a big problem where the first good or a much better solution will be of enough interest to enough people to generate a big business and then get the first good or a much better solution. A tiny fraction of potentially very valuable information technology startups and teams do know; nearly all the rest do not.Violinist Nathan Milstein had a really nice performance of a certain passage in one of the Bach pieces for solo violin. Others said that what he was doing was difficult. His response was, “It’s not difficult. Either you can do it or you can’t.” He went on to explain that he could show a good violinist how to do it and then they could learn it in a week.For my startup, from when I first looked at the problem, I had, from my history of applied math and computer based problem solving, no doubt that the problem was big, that I could have a very good solution, the first good solution, that such a solution really should be nearly all the basis for a very valuable new business, and that I could do the needed applied math for the crucial core secret sauce enabling technology and write the production quality code. I thought I could.So far, all the work has gone just as planned, fast, fun, and easy, a “cake walk”. No problems.I saw the problem, quickly cooked up a solution from 50,000 feet, derived some original applied math for the crucial core of the solution, typed the math, complete with theorems and proofs, into D. Knuth’s TeX, worked out all the fine details at least at a level for early production at revenue of, say, $1 million a month, designed, wrote, and ran the corresponding software. No problems. Routine. Fast, fun, and easy.And now I’ve done all that. For the code, I wrote production quality code as the first version, with no prototype. And it is my first Web site code. Yes, the code is very well commented with good external documentation. No, the code doesn’t need refactoring.So far all the planned software is written and on some initial test data appears to run correctly. I have some tweaks I want to do to the software and the user interface, and I want to check for a third time in fine detail some tricky indexing in some of the crucial core code.Then the software will be ready for, say, the $1 million a month. Sure, for $10 million a month, will need to do a more on sharding for scaling (already in the architecture and much of it in the running code), processing the log files, data handling, etc.To go live, I need to add more initial data and develop some simple tools for routine manipulations of such data.Then going live will need only very routine work done by 1+ million Web sites in the world — they’ve done it, and so can I.I have not been able to spend full time on the work, and I’ve had some independent, unexpected external interruptions, that would have been the case no matter what I was doing, but the work on the startup itself has gone fine — fast, fun, and easy, a “cake walk”.So, right, I’m a sole, solo founder, janitor to CIO to CEO, and 100% owner.As inSam Gerstenzang, “The Happy Demise of the 10X Engineer”,athttp://themusingsofthebigre…with This is the new normal: fewer engineers and dollars to ship code to more users than ever before. The potential impact of the lone software engineer is soaring. How long before we have a billion-dollar acquisition offer for a one-engineer startup? How long before the role of an engineer, artisanally crafting custom solutions, vanishes altogether? Yup! So, I’m a “one-engineer startup”, but I’m aiming at much more than $1 billion.Athttp://themusingsofthebigre…I just posted some of the topics I used in writing the code and some thoughts on training when I have to hire.The results for the users require some data manipulations. For really good results, the data manipulations have to be powerful, too powerful for intuitive heuristics, routine software, computer science, AI, ML, etc.So, for the data manipulations I used some original applied math. It’s math and not computer science, AI, ML, etc.First cut, it would appear that I would need more data and disk drives than, say, the NSA. But, since I know some good pure/applied math, I was able to see how to reduce the volume of needed data down by some astronomical (literally) amount, down to something quite reasonable. Indeed, so far my back of the envelope arithmetic suggests that I could serve the world for about 150 TB, that is, one 2U server with 10 high end solid state disk drives. And the astronomical reduction in data size, sure, means a similar reduction in central processor time.Ah, applied math, with theorems and proofs, for secret sauce.Cake walk!

    1. cavepainting

      Hi, good that you feel it is a cake walk, but the reality is until the dogs eat the dog food (as in customers you are planning to sell to are actually engaging with the web site, using it, and driving revenue), it is all a hypothesis that is yet to be proven out.Hubris is at the root of all failure.

      1. sigmaalgebra

        Basically you are just saying that accurate planning is impossible. Lots of rock solid evidence shows that accurate planning really is possible.The planning for my project has been beautifully accurate to date, and the rest of the planning is nearly as solid.You missed the initial point. Again, once again, yet again, over again, one more time — the probability of success means NOTHING. Similarly for the probability of the struggles you mentioned. Instead, what is important is the conditional probability of success. etc. given particular circumstances. As I explained, a tiny fraction of startups know what they are doing, and the rest mostly don’t.Planning such as I’ve done to be successful REALLY IS ENTIRELY POSSIBLE, REASONABLE, AND COMMON, common if know where to look, and that is NOT in the outrageously sloppy and silly work of Sand Hill Road.Your view is old, is in “The Little Red Hen” in Mother Goose.Since the children’s view in Mother Goose, the world, especially the US, especially the US DoD and a lot of the work of the NSF and NIH, are just awash in astounding contradictions of your view.E.g., the first version of GPS worked as planned, right away. If the US DoD charged for it, then they would have a license to print money.E.g., similarly forhttp://iliketowastemytime.c…Mach 3+, 80.000+ feet, 2000+ miles without refueling, never shot down, with great pictures of the USSR, as planned.E.g., Keyhole — a Hubble but aimed at the earth — worked fine, right away, as planned.E.g., every person who rode a rocket into orbit put great trust in planning.E.g., Bell Labs knew early on that vacuum tubes were too large, expensive, noisy, and unreliable and that a better amplifier was needed. They had a solid state rectifier so made an informed guess that a solid state amplifier would be possible. They had Shockley, Bardeen, and Brattain working on that before WWII. The war interrupted the work, but soon after the war they were done — had the first transistor. And Bell Labs knew just what they had, knew its value (change civilization), and, thus, gave the rights to the world for free.Product-market fit? Sure, that may be the main, difficult, important issue for some new sugar coated breakfast cereal to make more dentists rich, but it’s mostly silly, baby talk for serious work on serious problems.Accurate planning can work for projects in applied math, physical science, medical science, engineering, and practical computing. And for product-market fit, such planning can also work there.Yes, traditionally the big challenge was getting product-market fit for a new product for a new market. But, presto, bingo, that was done by IBM’s System 360, IBM’s DB/2 and SQL, the first Xerox copier (“a license to print money”), the IBM PC, Windows, each version of Windows to the present (except for Vista and 8), Microsoft’s Office, the daisy wheel printer, the laser printer, packet switched data communications and TCP/IP, Knuth’s TeX (the unique world standard, right away), Google, Facebook, Cisco, and each of Apple’s products, solid state disk drives, and more.Knuth’s TeX: I was one of the first members of the user’s group; I saw right away that Knuth had a great solution for a really nasty problem. It was certainly a really nasty problem for me and, really, for everyone in the world trying to write in more mathematical fields. That was a plenty large market for Knuth to sell lots of his The TeXBook. And, indeed, TeX became the unchallenged international standard right away and has remained so. Right from the beginning, there was no doubt in my mind and likely not in that of Knuth.These are all elementary, long well covered points here at AVC.My work is for a huge old market, still rapidly growing, where about 2/3rds of the market is served at best poorly, and I have in nearly all quite simple, solid running software and solid applied math an excellent and the first good solution for the 2/3rds.Yes, FedEx has been a big success and, really, with a lot of struggles and no cake walk. E.g., I personally saved the business from going out of business twice. For a business with such complex operations, some struggles would be expected, but too many of the struggles were from poor planning.But for my project, the technical and operational parts are now solid. So, essentially the question you are raising is the old product-market fit. So, sure, for FedEx, some people might have doubted that people would want to use the service. But founder F. Smith was sure, and he was fully correct, right and for the right reasons.He had some early evidence that he would be right. And I have some early evidence that I’m right.My project has been a cake walk so far. After I go live and get publicity, people will like it or they won’t, and I won’t have the operational struggles (planes, package handling, pilots, jet fuel, airports, Teamsters, scheduling, really an NP-complete problem, BoD arguing, fighting people down the hall goal subordination, etc.) FedEx did.For my project, whether or not people like the service, either way, the work itself will be no big strain or struggle. If people like my work at all, then just simple arithmetic shows that I can reach revenue of $1 million a year easily and, then, just from virality, $1 million a month. At that point there is large enough sample to extrapolate.Be wildly skeptical. Don’t count the revenue before it is in the bank. But with that view, no one would ever do anything new.

        1. cavepainting

          I don’t know the market opportunity you are addressing. Unless you share more details on what it is, it might not be possible to have a meaningful conversation.In the absence of specifics, hard to debate it. You might well be correct and I do wish that it is a cake walk for you or any founder who is passionate about what he/she does.

          1. sigmaalgebra

            Market: Internet content search, recommendation, and/or discovery for each person’s interest, of the moment, highly personalized.Key point: Make some, enough, progress with the computer science Holy Grail problem of meaning of the content and, then, give each user for their interest at the moment the content with the meaning they have in mind.Google, Bing, etc. do really well with search based keywords/phrases. So, what they have is still a lot like what Google, from a computer-library project at Stanford, started with — for Internet content based on text or at least with a lot of relevant text, a computer version of an old library subject index card catalog with results sorted in descending order based on a gross measure of popularity.IMHO that work does really well for about 1/3rd of Internet content, searches (recommendations, discoveries, etc.) people want to do, and content they want to find. The other 2/3rds is served at best poorly.So, the approach of Google, etc. is search by keywords/phrases. Already decades ago (e.g., a friend of my father’s, the friend working on information retrieval at Battelle) the field of information retrieval knew well in clear terms that search via keywords/phrases was just a small part of the problem.In simple terms, to search by keywords/phrases requires a user to know (1) what they are looking for, (2) that it exists, and (3) keywords/phrases that accurately characterize what they are looking for and requires (4) the search engine to have such keywords/phrases for all the content. That’s asking too much, far too much, of both the users and the search engine. Again with a lot of really good C++ coding at Google, this keyword/phrase approach works really well for about 1/3rd of search.In particular, of crucial importance is the meaning of the content, and using keywords/phrases to accurately characterize meaning is impossible, absurd. Meaning is challenging to handle, and keywords/phrases are not even a good start.The other 2/3rds includes still images, music recordings, video clips, old movies, old TV shows, and more. There doing well with meaning is just crucial.In particular, even for content based on text, where meaning is also important, keywords/phrases are not even a start on characterizing meaning.There are various approaches to meaning including language semantics, natural language understanding, e.g., the work of P. Norvig at Google.Note: Our work is for “safe for work” Internet content.So, I saw the problem, take 2/3rds of Internet search around the world (if do that then get a company worth, ballpark, $1 T), worked out how to do that, where the crucial core is some original applied math I derived based on some advanced pure/applied math prerequisites, designed the software and server farm architecture (scalable, reliable), designed, wrote, and ran the software, etc.To a user, my work looks like just a simple, easy to use Web site. Each Web page is simple, dirt simple, is 800 pixels wide, loads for at most 400,000 bits (that is, darned small and very fast), needs a Web browser up to date as of 10 or so years ago, with both vertical and horizontal scroll bars works in a screen window as narrow as 300 pixels, has high contrast colors and large fonts, has no pull-downs, pop-ups, roll-overs, or icons, little or no use of JavaScript (Microsoft’s ASP.NET writes a little JavaScript for me to support some feature in ASP.NET, maybe cursor positioning, but I didn’t write a single line of JavaScript) works fine with JavaScript disabled, has no user logins or passwords, makes no use of cookies, has some of the best protections of user privacy, etc.So, will users like it? Well, they like Google and Bing, both a lot. But users quickly realize some of problems with Google and Bing. Really the problems are as I explained above.So, for Internet search, 2/3rds of it is served at best poorly. I have a terrific, the first good, solution for the 2/3rds.Am I in competition with Google and Bing? Not really. They have their 1/3rd, and my 2/3rds is quite different. Heck, I will continue to use Google for their 1/3rd! And there is a sense in which my work will generate more traffic for Google.Yes, there is a sense in which my site can do some relatively good ad targeting.Q. What about equity funding?A. Ah, there’s not a well known venture firm in Silicon Valley, Boston, New York, Baltimore, or anywhere else the least bit interested in my work. So, I’m self-funded, but the expenses for my business are tiny and will always be just a tiny fraction of any significant revenue.It may be that when my revenue reaches $1 million a month I will get calls from venture firms suggesting that they can do a lot to help my business grow to the next level, fund the big, go to market roll out, fund the big staff expansion, help with recruiting all my C-level people, help with business development and recruiting some high end business development people, supply advice from their business acumen, know some computer science professors who could give me help, suggest a really good startup law firm they know well, know lots of investment bankers and can help arrange an exit via M&A or IPO, etc.One concern I would have would be, say, getting a budget approved with a project for some good ad targeting. So, I’d give the BoD a talk covering some of the ad targeting applied math, original with me, but would be afraid that most of the Board would soon soil themselves, the furniture, the carpet, and leave a trail on their way to the rest rooms, meet at the closest bar, and vote to fire me, with none of my stock vested — they would take my company from me and leave me with 0.00. No thanks. I’d rather remain 100% owner.It is true that some equity funding might have helped some of the work, but by the time I have $1 million a month in revenue and less than $1000 a month in expenses a seed round or a Series A round, say, for $10 million for 1/3rd of my company, the time will have passed; indeed, the time has passed now.During such a phone call, if it comes, I will try not to laugh and at the end tell them the dates when I sent them e-mail on my project, e-mail they ignored.Really, as a sole, solo, technical founder who did all the work, where my business is just dirt cheap to start and run, I see no chance that an equity funding firm and myself could reach across a table and shake hands on a deal. In simple terms, the first time they want to write a check will be long after the last time I would accept one. Besides, I want to work only with good people, and the fraction of venture partners I so regard is tiny.Sure, I’m being unusual, but the success I’m seeking is exceptional and for that being unusual is nearly necessary.

          2. cavepainting

            Hi, thanks for explaining in more detail.I am sure you are familiar with the history of semantic search and meaning based search and the number of companies who have ended up as dead corpses.Leaving aside questions related to how good the product is, the big question is distribution. i.e. who is the initial target segment and how will they find the site? An early stage startup is unlikely to have the cash to buy search or social ads to drive traffic; word-of-mouth is an answer that does not add up in 99% of cases unless the product delivers a transformative experience relative to current options.So, a key question to consider is: what is your go-to-market strategy? How are people going to find your site?Also, have you done user tests on various search terms to know for a fact that the results from your engine will be substantially better than Google or bing? Do the users (not you) perceive the difference in a very tangible and visceral way?

          3. sigmaalgebra

            … the big question is distribution. i.e. who is the initial target segment and how will they find the site? So, a key question to consider is: what is your go-to-market strategy? How are people going to find your site? I have some publicity approaches lined up. For one of the approaches, it’s still a little as in Lawrence of Arabia where I want my story told and some tech industry writers very much wants a good story to tell.The initial market segment is a good one, for virality and ad targeting, the set with emphasis on some or all of: specialized professional career, classical music, opera, ballet, art collecting, classic old movies, fancy food, high end interior decorating and architecture, travel, scenery, urban, affluent, well educated, thinking, curious, early adopter, politically astute and active, unhappy with McDonald’s, McPaper, and one size fits all and wants deeper and more specialized information and meaning, etc.unless the product delivers a transformative experience relative to current options. It does, very much so; the UX, that is, user experience, is very much “transformative”, engaging, involving, interactive, for some “addictive” or “game like”, etc. Typically one search will take much longer than at Google or Bing except in the cases in my 2/3rds where Google and Bing are at best poor and a poor user flips through 10,000 Google results, the first 10,000 of Google’s 20 million or so.For about 2/3rds, Google is at best poor, and my work is much better.I am sure you are familiar with the history of semantic search and meaning based search and the number of companies who have ended up as dead corpses. One way and another, I know a lot; I can’t say I have expert, comprehensive knowledge.From everything I’ve seen for anything like any competition, they are in an airplane with pedal power and I’m going by at Mach 2 in an F-22.Really, information technology startups and their VCs flatly just don’t know how to take a real problem, invent some new, powerful applied math for the first good or a much better solution, implement that math in software, and have the software be attractive to people. Now that Chris Sacca retired, and even he didn’t finish his math Ph.D. (I did finish mine), it’s looking for hen’s teeth to find a VC who could review some original, advanced math and, then, write a check based on that review. That means that VCs won’t look at my math, and, for competition, won’t look at their math, if they have any, either. So, any competitor and I have to do the same: Get to good traction, usage, revenue, etc. before any VC interest. Net, since I haven’t seen any competition, I doubt that there is any.Here’s one important reason for no competition: The overwhelming assumption is that since the subject is information technology and since computer software is required, the academic foundation just must be in computer science, and now, AI or ML.Wrong. Has been wrong. Badly wrong. Will remain badly wrong. Instead, flatly, the best, really the only promising, academic foundation is math.As I discovered getting a paper reviewed (“neither I nor anyone on my board of editors has the mathematical prerequisites to review your paper”, essentially the same feedback from more than one of the best computer science journals), so far the computer science profs don’t know that math (sure J. Schwartz at Courant was a rare exception; sure, and von Neumann if want to count him, but he was a bit too early for current computing). So, the computer science students don’t know that math. So, the Silicon Valley information technology startup entrepreneurs don’t know that math and can’t compete with me.Sorry, computer guys, AI, ML, intuitive heuristics, lots more C++ code, and some elementary linear algebra and statistics just won’t cut it!On linear algebra, take my word for it: I effortlessly, without trying, totally blew away everyone else in an advanced linear algebra course by a world class expert, and it was my first linear algebra class ever (uh, in my career I’d worked with a stack of some of the best books, carefully studied several of them, made various applications, had written my own notes in linear algebra, enough for a book, from memory a night before the course started wrote out from memory about 80% of the course material).Very much, people just do NOT want to major in math.But working with “meaning” is super difficult; as I mentioned, it’s a computer science Holy Grail problem. I have a lot of good reasons — from math and more — seriously to doubt that anyone else has done anything like what I’ve done.As I’ve mentioned, I have confidence in the crucial core secret sauce from my original applied math with advanced pure/applied math prerequisites.Also, have you done user tests on various search terms to know for a fact that the results from your engine will be substantially better than Google or bing? Impossible. Google and my search engine are not directly comparable at all. These two are very different tools for very different parts of search.You are asking if a cargo ship is better than a freight train — the two solve very different problems.Again, I’m not in competition with Google. My work does not replace Google. When Google works well, and for about 1/3rd of search it does, use Google. E.g., I will continue to use Google for their 1/3rd.In particular, sure, Google has “search terms”; my search engine doesn’t.No wonder: As I’ve explained, search terms, that is, keywords/phrases, are hopelessly bad at characterizing meaning. Again, my search engine concentrates on meaning; thus, there is no way a search engine based on search terms or keywords/phrases can be comparable with my work.Again, Google has their 1/3rd where they are good and is at best poor in my 2/3rds, and I have a very good, the first good, solution for my 2/3rds.Where Google is good, usually I’m not better or even as good.Again, I’m for the other 2/3rds, not for Google’s 1/3rd.Do the users (not you) perceive the difference in a very tangible and visceral way? Very much so. Part of my handling meaning is necessarily handling artistic taste, wonder, amazement, what is enthralling, entertaining, exciting, engaging, “visceral”, etc.Tough to describe artistic taste with keywords/phrases.Or, there’s a lot of content on the Internet. Google is good for about 1/3rd of it, and that leaves out a LOT of content, leaves a lot of content users have no good way to find. I’m good for the other 2/3rds. But to do that, the work is different, very, very, really totally, different.People don’t think about the 2/3rds because, like death and taxes, they don’t see any possibility of a solution, of anything better.It’s an old story, as Henry Ford said: “If I had asked people what they wanted, then they would have said a faster horse.” Well, then the people didn’t know how to ask for a car, and now the people don’t know how to ask for a better search engine for the 2/3rds of search served at best poorly. And the information technology entrepreneurs don’t know how to build a better search engine — it’s more elusive than what Ford did when he did it.E.g., people just don’t dream that there could be a good way to search for images; well, doing well on images is part of what I’ve done. E.g., my test data is based on images, and the search engine worked great, right away!Just what it is about by far the best search engine for 2/3rds of Internet search, recommendation, discovery, etc. currently served at best poorly that has to be regarded as small potatoes?

          4. sigmaalgebra

            Your main concern here is, when the Web site for my startup goes live, will people come?Well, that is a special case of something quite general:Can I get the attention and/or money from other people?One way is to be an employee. My father did that, and I long did that.I assumed that there is a career there, especially if have a Master’s or Ph.D. in a STEM field.Well, in fact, no, not really.College professor? Have to go 10+ years living a lot like a graduate student before get tenure and rank and income enough to buy a house and start a family — unless spouse has a good job (my spouse didn’t).Generally, in a career, can get quite specialized and then get fired: Employer runs short on money; lose out in internal politics, say, being attacked by gossip from a clique.Once in a specialized field and fired, can be in deep financial trouble.So, in a job, still have to work to get an employer to give you money.So, in a career as an employee or as a startup founder, in both have to work to get people to give you money, or in a startup get the attention of people and, then, get advertisers to give you money.Either way, career employee or startup founder, just MUST work to get people to give you money.Between these two, which is better?Point 1: How difficult is it to get the money?Point 2: How much money can you get?Really, on both of these, really, if think and be a little creative, being a startup founder is MUCH better than being a career employee.That’s just a fact, Jack.So, sure, as a startup founder, I have to work to get people to give me money, but that’s not new, different, surprising, or avoidable since I’d have to do the same in a career as an employee.So, net, I should be a startup founder.

          5. cavepainting

            I think there are two considerations or concerns. Objective proof of transformative value for the user. That is, you have done enough piloting and testing with potential users to know that they see enough value to use the product and will keep coming back. It is not what you say or think but what they do. A model for distribution that is viable. ( in your case, figuring out how you will be found on the internet organically). This is the frontier where most consumer internet companies fail. It is a bitch of a problem to solve.

          6. sigmaalgebra

            No, neither consideration is operational, that is, they are not practical. Why? By the time have the solid evidence you want, have already done the development work and gone live. But the main reason for the questions was just, should I do the development work and go live? So, your answer to that question requires that I do the development work and go live so is no help in answering your questions about transformative and publicity.The solution is the same as common in nearly all good projects: Good planning.E.g., for F. Smith to invest much of his family fortune in FedEx was one way to answer the question, really also yours, “Will customers like it?” He did a lot of planning and had good reason to make the investment and was successful, but according to your criterion to answer the question he had to make the investment — thus, again, learning from making the investment was useless for knowing before the investment if should make the investment.Sure, in the planning, need to pay attention to publicity and “transformative value”, and I’ve done that. But now you are suggesting that I test transformative value before going live as a way to know if I should go live, but testing transformative value basically requires that I go live so is no help in knowing now if I should go live.For publicity, some things catch on and some don’t. Some things people like, and some they don’t. Or, how many ads on Google did the Google founders run as they first went live on Google? Sure, none. So, Google got heavy usage without the approach of ads on Google.With a better mousetrap, will users beat an Internet path to my Web site? Maybe; maybe not.But, again, as below:Just what it is about by far the best search engine for 2/3rds of Internet search, recommendation, discovery, etc. currently served at best poorly that has to be regarded as small potatoes?You are evaluating my project with such long rods that you don’t see the reality, the value; all you see is just a cardboard cutout that reads “Internet startup.” At this stage, to go live is at best an informed judgment call, but the judgment has to look carefully at both the real problem and my solution.Or, bluntly, the Internet is a big thing; providing good tools to let people find content on the Internet is a big thing; currently ballpark 2/3rds of Internet search to let people find things is served at best poorly, and that is a big problem; I’ve got the first good, a really good, solution for that big problem. In part a “big problem” means that there are ballpark 3 billion people who have the problem and will very much like a solution if they can get one.Such a solution won’t please a lot of people, 3 billion or so? Something that is in line to please 3 billion people won’t get publicity, virality, or usage?Your concerns seem to be from, again, looking at nearly all the Internet startups and applying what obstacles they faced — lack of transformative value and lack of enough publicity and virality.Uh, transformative value is close to guaranteeing both publicity and virality.The problem with looking at all those struggling startups, that were “no cakewalks”, is that those startups weren’t very good.Net, to evaluate my startup, have to look closely at my startup and use what know about the Internet, Internet search, and the details of my startup.The next time to evaluate is how long after going live it takes me to get to $200,000 a month in revenue. Then, sure, to $1 million a month, …, $1 billion a month.But, by $200,000 a month, there is no question I have a good business, and your questions are largely moot.Again, with your criterion, no one could justify doing anything new, no matter how intense was the need.

          7. cavepainting

            You either know the space cold and know what exactly you are doing or you are drinking your own kool-aid too much and are not sufficiently challenging your assumptions. I have limited information to know which is more true in this case.If you think you are the former while you are actually the latter, it usually does not end well. Awareness of what you know and do not know and the stuff in between is probably the most under rated quality in the startups business.Either way, my best wishes and I look forward to your launch and seeing the product.

    2. The Art of Football- Mo

      Hi, you must be the newbie!Mike Tyson said it best, everyone has a plan til they get punched in the mouth.You clearly have not yet been punched in the mouth.See, you keep using absolute terms. Cake walk! However, you admit your project has yet to go live. Ask any expert or leader in anything, and one thing you will constantly see, is that their understanding is nuanced and intricate.This is like saying, I have yet to ride on the skateboard, BUT, I have figured out mathematically how to land a 900 just like Tony Hawk. All the computer diagrams and formulas in the world will not prepare you for your first broken bone.I don’t know if Fred answers his own comments section, but how many times have you been drawn to a startup founder because he was a true businessman as opposed to a kid in his first rodeo? I think I remember Fred mentioning his going bankrupt twice and coming back from it. Those who have truly owned their field do it with retrospective humility, because they realize how many mess ups they’ve had and how many strokes of good fortune outside of their control, pushed them along on their path. All the business acumen in the world can’t save you from a bad partner, or a downturn in the industry, or your business getting broken into, or what have you. A developed leader with a healthy self-image will readily admit their shortcomings alongside their successes.I will stop now because I also admit this post is a brazen attempt at self promotion and outside links aside from a bunch of drivel that vacillates between hollow and flippant, but it’s more for the reader who has little experience and can be easily seduced by those who claim they know the shortcuts to the top. There are no shortcuts. Just the fundamentals, and the work to master them that takes a lifetime.

      1. sigmaalgebra

        Nonsense.I won’t look to Mike Tyson for how to plan applied math projects.More details in…

      2. ShanaC

        he does answer the comment section occasionally.

  13. cavepainting

    In terms of broad trends, the conclusions are directionally right.But startup funding metrics do not reflect eventual success or failure. Not getting funding might mean one of the following: a) the company might have gotten self sufficient with cash flow, b) raised debt, c) sold the company, or d) run out of cash and shut down. While d) is likely 70+% of the cases, the others cannot be dismissed as insignificant. And of course, getting funding does not mean eventual success either.Wonder if mattermark, crunchbase, etc. might have any reports on actual startup outcomes for the parties involved. (i.e. cash-on-cash return for each investor, actual exit valuation, etc.). Of course,some of this is confidential, but there is a lot of public information these days and some of it can be crowd-sourced.

    1. Twain Twain

      Anu Hariharan, a former partner of A16Z and now co-lead of YC Continuity Fund (approx $1 billion fund) said that to raise an A, startups need to achieve at least 15% month over month of organic growth. She noted that this is also the reason why YC graduate, Airbnb, took 2-3 years to raise a Series A.

      1. cavepainting

        That’s true. Still.. even growth can be an illusion for it could be a flash in the pan that flames out (or) be a growth hack powered by advertising without an underlying sticky product. Yik Yak, second screen apps, Secret, Viddy, etc. all experienced profound growth only to eventually fall by the wayside.

  14. Ronnie Rendel

    I think the roles of VCs has changes considerably since 2009 (or rather 2004) to 2017. I feel like when I started reading this blog, VCs were looking for breakthrough ideas and helped outstanding founders build successful companies around.Today’s it seems (I could be wrong) VCs (which there are more of now?) are more like mortgage brokers in 2007…

  15. JaredMermey

    @fred – when do you consider yourself seed versus A round? Is this about your check size or if first in?The subjectivity to that criterion would affect where on the funnel your investment starts relative to what @LIAD charted. Seed gone from 1/3 to 1/5 making next round and A has gone from 1/2 to 2/5 making next round.

  16. Frank W. Miller

    Looks like the numbers get worse as time goes on. Does that mean that startups have been having more trouble getting to the next round as the last five years have unfolded or is it just that they are younger and less likely to have needed to go on to a next round yet?

    1. ShanaC

      excellent question.

  17. Pete Griffiths

    From the perspective of the system, startups are outsourced risk, nicely contained with founders making unnatural efforts to succeed against all odds.If it was easy, it wouldn’t be outsourced, )

  18. Yoram Wijngaarde

    We did this chart at inspired by Mattermark. We also compare Europe vs. USA churn rates here… (USA is doing a bit better than Europe)