Venture Outcomes are Even More Skewed Than You Think

The typical “successful” venture portfolio is often described as having the following outcome:

  • 1/3 of companies fail
  • 1/3 of companies return capital (or make a small amount of money)
  • 1/3 of companies do well

Fred Wilson, for example, described this a few years ago:

I’ve said many times on this blog that our target batting average is “1/3, 1/3, 1/3” which means that we expect to lose our entire investment on 1/3 of our investments, we expect to get our money back (or maybe make a small return) on 1/3 of our investments, and we expect to generate the bulk of our returns on 1/3 of our investments.

It’s a generalization but one that’s pretty well accepted in venture circles and it’s how many VCs describe target fund distribution, myself included. But does this heuristic match reality?

Actually no.

Correlation Ventures just released a study that shows the distribution of outcomes across over 21,000 financings and spanning the years 20014-2013. For those of you that don’t know Correlation, they take a data driven approach to co-investing – essentially creating an algorithm that predicts the success of a company based on a number of factors that include both business trajectory as well as financing trajectory (we’re co-investors with Correlation in Distil, for example). The result is that they process a lot of data. Which leads to some pretty interested insights.

venture returns

Based on their data, a full 65% of financings fail to return 1x capital. And perhaps more interestingly, only 4% produce a return of 10x or more and only 10% produce a return of 5x or more. These data suggest that the heuristic I site above potentially presents a rosier picture of the venture industry than reality suggests is the case (there are some missing data here in that the vast majority of companies are in the 0-1x category but the data within that category weren’t released – but my suspicion is that within that category the distribution of outcomes follows a similar power curve).

This really underscores the challenge of creating a venture portfolio that produces reasonable returns.  If you were to actually construct a portfolio based on these averages, a $100M venture fund investing in 20 companies would produce a gross return of approximately $206M (that’s before fees and expenses). The resulting fund would have an IRR in the range of 10% (the exact IRR would depend on the timing of the cash flows, but I constructed a few models to approximate this and 10% was the average return).  That’s hardly something to write home about and underscores the challenge of being “average” in this industry.

Hidden in this exercise – and perhaps more important – is the challenge of finding companies at the right side of the distribution chart. In my hypothetical $100M fund with 20 investments, the total number of financings producing a return above 5x was 0.8 – producing almost $100M of proceeds. My theoretical fund actually didn’t find their purple unicorn, they found 4/5ths of that company. If they had missed it, they would have failed to return capital after fees.  Even if we doubled the number of portfolio companies in the hypothetical portfolio, a full quarter of the fund’s return comes from the roughly ½ of a company they invested in that generated 10x or above. Had they missed it, they would have produced a return that roughly approximated investing in bonds – not the kind of risk adjusted return they or their investors were looking for.

It’s important to note here that I’m extrapolating a bit – the Correlation data are based on financings, not companies (I asked – they didn’t have a sort at an entity level in this exercise). I thought about ways to normalize this but came to the conclusion that the best normalization was to use the raw data and caveat that it was financing level, not company level. I’m going to work with Correlation to get entity level detail the next time the do this exercise.

All of this math simply underscores how important winners are to venture returns and how difficult it is to find them.

Note: An obvious, but important, thank you to Correlation for allowing me to share these data as they were originally prepared as a private exercise for Correlation and their venture partners. As I mention above, we’re coinvestors with Correlation in Distil Networks. They have a bit of a unique model for co-investing which allows them to see a lot of data on a lot of companies to support their data driven investment thesis (which also allows them to reach fast investment decisions).

  • porteous

    Great piece Seth. This reality is poorly understood, particularly among entrepreneurs and I believe it explains a lot of the behavior of venture firms today. Thanks for sharing this.

  • thanks for posting Seth…this seems to be a consistent theme (that venture is even harder than everyone thinks, or don’t bet on the unicorns, or good luck getting returns, etc) but this analysis really spells it out.

  • jody s

    I think it also spells out the importance of identifying the “winners” and investing your pro rata when you can – helping skew your portfolio as far to the right as possible.

    • definitely. i’m sure many of the 0-1x were complete fails while many of the 3-5x were rounds where firms doubled down.

  • Rob Leathern

    Cool post, Seth. I guess a further analysis is that one could boil down some of the implications this kind of data would have for fund size, initial bite size/full allocated amount assuming pro-rata rights (crucial it appears) – not that it would be easy since there’s not really a normalized “funding size” that will garner enough information in most cases to know if there is a “there” there for a startup… fascinating stuff though. Related thread from a few years ago we had:

    • Thanks Rob. The “shorting a startup” concept is pretty amusing. I’m just about to post another data set with some additional information on this topic that’s worth taking a look at.

  • Seth – very interesting data. I think the heuristic that Fred Wilson highlights about 1/3, 1/3 and 1/3 is true for a small handful of elite, top decile funds (USV, Sequoia, etc).

    Of course, not everyone is above average or top decile. We highlighted some of that in an analysis of unicorn exits ($1B+) and what the data shows is that there are very very few funds that are able to invest in multiple billion dollar exits. In other words, not many are able to actually “pick winners”

    Here’s the data —

    • Thanks for the pointer to that data set Anand. And I think Fred and a few others really do beat the average significantly. Again the data suggest that picking an “average” VC is a losing strategy – LPs need to be in the very top performing funds to justify the risk.

    • mikey248

      From data I’ve seen, the behemoths have exited 16-22% of their portfolio companies, history-to-date.

      Assuming that the typical investment is 6 years for a large fund, that they’re on fund 7 on average, boost eventual exits from 19% exited to 23% in total.

      All this says NOTHING about quality of exit. But in any event, it’s far below 1/3 and far, far, far below 2/3.

      And that’s for the big funds, who get the better deal flow, and who enter later stage, and who can give companies a soft budget constraint to enable some exit, any exit.

  • Sdbb

    I’ve never heard of this 1/3 rule before. It’s always been the 10% of the portfolio making up for the 90% that doesn’t perform. 33% of your portfolio being successful? That’s not typical at all.

  • John Huntinghouse

    Great post and insights. I think the data sets presented here just place to the forefront what many LPs are starting to realize with their investments. Definitely something to mull over and take a deeper dive into.

    • Thanks John. Absolutely take a look at the post I just put up – more data (from an LP) about what works and what doesn’t work in venture.

  • benjamindblack

    Great post. One wrinkle that I would love to understand more has to do with follow-ons. My gut feeling when I worked in venture capital was that most of the losses came from bad decisions around follow-on investing. Its very difficult for a VC partner to admit they made a mistake and stop supporting a portfolio company. This loss avoidance bias, I think, is a key contributor to VC underperformance since piling as much money into your one or two winners is obvious and capped by the particular dynamics at the winner.

    • Totally agree. And I just put some more data up in a new post that suggests exactly this – larger funds do worse and my suspicion is that it’s due to exactly the behavior you’re describing above).

  • Since this is based on financings instead of companies, I wonder how these effects might be magnified by multiple venture rounds in successful companies?

    In the poorly performing companies, I would assume after one (or two) financing events the company is done. In the successful companies, there would be several financing events, where the earliest events have the highest multiples, and the latest events have multiples in the 1-5x range.

    Given that, it would imply that the distribution on a company basis is skewed even *more* right-ward?

    But returns would also be based on expectations… investors in Facebook’s last financing round probably weren’t expecting a 20-50x return, but by putting enough cash to work and getting a smaller (safer) return on a faster basis obviously would have done pretty well cash-on-cash and IRR-wise, right?

    Fascinating data, thanks for sharing!

    • I just posted some additional data (new post) that starts to answer these questions. Interestingly for larger funds what you describe is exactly right – I assume because they have a tendency to double and triple down on their losers. Smaller funds actually do better.

  • Seth, thanks for sharing these insights. I have been lead to believe the Angel community has a different investment approach to the larger VC funds and would be curious to know your thoughts.
    In Canada we have a much smaller VC pool and a more conservative investment attitude in general. My colleagues and I have been debating the merits of crowdfunding as an early financing strategy for pre-revenue start-ups.
    Equity crowdfunding seems to be just around the corner to complicate matters. We see an easy entry for equity ownership in start-ups, however, it is less clear what or where or when the exits will take place.

  • Chris Yeh

    The tough part about early stage investing is that until the sample size is large enough, it’s hard to tell the difference between good and lucky (and bad and unlucky). And it can be pretty expensive to make that sample size large enough!

  • Lou Covey

    Is it possible that there is less emphasis on finding disruptive investments and too much on following the herd? I read recently that 75% of all investment in the past 5 years has gone to technologies designed to make it easier to sell stuff to other people, rather than make stuff to sell. This decline in return seems to track pretty closely with the decline of traditional media and traditional advertising. While revenue for that has declined, it has apparently been replaced by venture investment in media technology… none of which actually seems to provide a return.

  • David Goldberg

    How much of this is a self-fulfilling prophecy, in that VC’s only care about finding the unicorns, and by doing so, invest in way too many high-risk ventures that fall in the 0-1x, as opposed to finding more consistent 5-20x?

    • WasatchHomer

      In early stage investing, I believe it is very hard, if not near impossible, to even know which companies will return 5-20x. As others point out below, the best strategy seems to be making your initial investments and then focusing on what you believe are the winners and abandoning the others. That is a very anti-venture/entrepreneurial way of approaching things, but probably correct from an investment returns perspective.

  • BostonBizPerson

    Did you put any floor on the size of a “financing”?

  • For those keeping track, if that’s a power law distribution, then the best fit alpha is 1.94, meaning that it has no mean: there is no average outcome for a venture investment. Makes it tough to analyze. (A power law with an alpha of less than 2 has no mean.)

    Kauffman’s Angel Investor Performance Project data seems to show an alpha of about 1.2, meaning (in some sense) that their investments were riskier than VC as a whole. This makes sense. My portfolio seems to be working out to an alpha of about 1.3, which also makes sense since I’m a seed investor.

    I’d be curious to know the alphas of other VC funds and which stage the invest in. It would be some interesting data both as (1) an answer to the question “can’t VCs just choose to do better” (if the returns follow some power-law behavior that is driven by some extrinsic mechanism, as a power law would suggest, then they can no more choose to do better than every GM in baseball can just choose to have a higher team batting average) and (2) some parameters around how that extrinsic mechanism works.

  • Mihail Lari

    Good to see this. Wondering if there are any studies done on the primary reasons that venture-backed companies fail…and if so, what venture investors do to help improve success rates? Is it primarily due to leadership gaps; or because it can be hard for an entrepreneur to tweak an idea (since investors may be be tied to the initial idea they invested in e.g. Twitter coming out of another company after most of the investment was returned), or something else?

  • I’d like to see the correlation between the 65% who returned 1X or less and their MARKETING STRATEGY.

    I’m convinced that ‘cargo cult’ or ‘fad king’ marketing and models—i.e. full on uncritical embrace of the ‘cult of disruption’ and the VCs advocacy of it inside their deals—kills a lot of startups as it did in the dotcom bubble.

    FYI, Google, Microsoft, VMware, Adobe, RedHat and many others all practiced upstream partnering with incumbent alpha players in their startup phase. They were all ‘non-disruptive’ innovators at the start with unconventional partner-advantaged marketing approaches. That’s how they grew into dominant superpowers with relatively low cost of sales and marketing.

    FYI: My recent series on Upstream 2.0 Platform Marketing (parts 1-4)


  • A major factor is that the Correlation data presents the distribution across individual financings and not portfolio companies. So it’s not an apples to apples comparison to make a reference to Fred Wilson’s portfolio distribution of 1/3, 1/3, 1/3 which refers to company level distributions of winners, Money Backs and Losers.

    In an attempt to justify the large number of low multiple returns i think are an inherent part of VC. In my experience, when you run a fund, you sometimes take hits in individual rounds from recaps etc, in order to get a better position in the next class of shares in a company that shows promise. If a team fails completely, it’s easy, you walk away.

    As a Fund manager, you always optimize the whole lifecycle of investments for every
    portfolio company, as well as allocations across companies. Some times a round wipes out your earlier holding in a re-cap, which nulls the initial (Seed) investment, but you still come out on top thanks to the follow-on rounds. And in some cases you do a financing with a 1.5-2X expectation in a single round because you believe that your total return for the holding will be 7-8X. Both scenarios create 0-2X multiples for single financings,
    which sounds alarming, but you can still come out with strong returns overall, both on company level and fund level. So, it’s non-trivial to derive clear conclusions from this data and how the numbers actually compare to Fred Wilson’s model portfolio.

    Another aspect is that the Correlation data seems to mix larger funds with small Seed Funds into the same bucket, which really makes little sense to me. If you’re doing a $60M Seed fund with a high volume strategy and you put into the same bucket as a $500M lifecycle fund or a $250M late stage (B/C/D-round) fund, it makes little sense if you want to assess how successful each strategy is and what the ideal fund/portfolio sizes are.

    The Seed fund could design a high percentage of losers into it’s model portfolio, typically 50%, which should still be fine due to the asymmetric capital allocations per company in a Seed fund. For the $250M late stage fund with a $10M allocation per company across 20 companies, a 50% failure rate would in most cases be a serious problem (due to the lower return multiples on a per company basis).

    And finally, the kicker in VC dynamics, nulling all of the above!
    It doesn’t matter what you do as long as you hit one Unicorn in each fund, then everything else is forgiven/forgotten. Great, we know what to do then.
    I think there is an opportunity in playing the mid field, aiming for $50-$250M exits, which is what we do in Scale.VC deal-by-deal syndicates.

    Great to get more transparency in VC data.
    BTW, Just had exchange with Brad and i’m looking forward to joining your AL Syndicate.

  • How many investments does it take then – if we invested in 5 average looking start ups would this work or do we need 25- 100 / year over XX years?

  • mikey248

    The fact that this data is at the financing level and not the entity level indicates that the outcomes are even worse as a pct of companies, since the successful companies will have more financings than unsuccessful ones, so are more likely to be counted in the right bars multiple times.

  • Greg Kuntz

    There is something to be said about ethics on the company’s side. Entrepreneurs should not be accepting checks their bodies can’t repay within their lifetime. This is called taking ownership.
    Entrepreneurs with boots on the ground domain experience should know what these limits are and how to plan accordingly within the startup.
    If you fail in a startup, you need to eventually pay the investor back plain and simple.

  • Thaddeus Diamond

    This was really useful! Is there a way to update this annually and repost easily? Would be curious to see how/if this trend sustained during the past few years. The high valuations and increase in VC capital makes me intuitively think that more VC-backed companies should reach higher multiples.

    • i’ll try to get this updated. or at least run it for our older fund (that might be a year from now).

  • Peter LePiane

    Awesome post! I’m playing around with my own Monte Carlo modelling using the above tiers. Wondering if you’d be willing to share the maximum multiplier found in the data. In other words, the last tier is 50x. What is the maximum data point in that tier? Also wondering what the minimum data point was. I assume it wasn’t zero but was some negative multiple. Perhaps that wasn’t the case as no follow rounds were invested.


    • Hi Peter. I wasn’t able to get the raw data from Correlation but I’m sure the max was quite high (seed deals that turned into unicorns). Not sure what you’d want to put in for modeling, but even for Series A investors it’s not unheard of to have 100x returns on a deal (we’ve had a few at Foundry).

      • Peter LePiane

        I put in 200x as a guess. Sound semi-reasonable?

        • Yup. How sensitive is the model to this particular data point?

          • Peter LePiane

            Not particularly. It’s monte carlo so it just needs to be a reasonable end point. Thanks!