Seed vs. Growth Investing: A Reality Check

These days, you can’t walk into Silicon Valley hot spots like the Rosewood or read tech and venture blogs without hearing the deafening buzz about the most recent hot, seed investment. And with very visible 100x+ seed stage returns in companies like Zynga happening before our very eyes, every angel and VC firm is piling into the seed round of the Next Big Thing. Seed stage venture and super-angel firms are being formed left and right. Incubators are proliferating like never before.

Just last week, Fenwick and West’s recent seed stage survey captured what we all sense to be true: Seed stage valuations are growing rapidly, as are the number of seed stage investments being made — particularly in the consumer Internet and software space.

But no one is asking the fundamental question: Are the returns from seed stage investing worth the risk? And from a risk-adjusted returns perspective, is it better to be a seed or late stage venture investor?

To answer that question, I looked at the NVCA Benchmarking Database powered by Cambridge Associates — the same data source that Bruce Booth and I used last summer to uncover the little known fact that healthcare venture returns had outperformed IT venture returns over the decade of the 2000s.

At first glance, the returns numbers for realized seed stage investments from 1980 to Sept. 30, 2011, look quite good: nearly a 37% average gross IRR and 47% IRR for IT and Software (which includes Internet companies).

But look a little deeper and you’ll see that seed stage returns for the past decade — from 2002 through 2011 — look downright abysmal, with a paltry 5% gross IRR. As you can see, there has been a massive and unmitigated drop-off in realized IRR at the seed stage for software and IT companies since the 1990s.

When split out by sector, it’s clear that IT and software seed stage returns are bringing down the returns average for all of venture capital’s seed stage investing over the past decade quite dramatically. Healthcare seed investments realized a 14.1% gross IRR vs. a meager 4.8% from IT and software seed deals. Netting out for management fees, IT and software seed stage investments aren’t carrying their weight and clearly would have had a negative net IRR over the past 10 years.

The same data set shows that 71% of exited seed stage investments over the past decade have returned less than 1x. And just 4% of the returns of seed exits have yielded greater than 5x to their investors.

There is surely a lot to criticize in this data and approach — for one thing, it looks solely at the returns generated by the universe of over 1,300 venture firms in the Cambridge Associates data set. It also likely excludes some of the more prolific and successful angel investors. Further, it looks at a pooled average, and we all know that top-decile is really all that matters in venture returns. And, of course, it is a look in the rear-view mirror, excluding realizations from the recent crop of Internet winners. Past rarely predicts the future in investing, as we all know. But since these data are the best measure we have of performance in our asset class, it’s really hard to ignore these numbers.

In contrast to seed returns, the returns from late stage venture investing have been 4x better than seed stage returns since 2002 — that’s a dramatic difference in returns.

What seed investors surely must be counting on is that we are headed back to the go-go days of the ’90s — and that’s a pretty huge bet to be making today given that the exit dynamic of this era is thus far quite different than the ’90s. Today, the top end of returns are being accumulated by just a handful of companies and investors vs. the broad dispersion of returns and sky-rocketing IPOs in the last boom of the ’90s.

I’d argue that the seed stage returns data aren’t good enough to justify the amount of exuberance and valuation growth that we are seeing in the seed market, particularly in IT, Internet and software.

I’m all for new company creation — it is the lifeblood of our economy. And I’ve made a couple of seed investments in the past. But a decade of abysmal seed stage returns should tell the venture industry that it takes a ’90s-like boom to create the exit conditions that justify taking seed investment risk.

Betting on booms and bubbles may be inherent to IT VC — but it shouldn’t be, and can’t be a great strategy for limited partners — especially those with mandatory obligations to pensioners and university endowments.

Investors take massive risks with seed capital. Knowing that 70% of seed stage companies eventually have a negative outcome and less than 1 in 20 have a happy ending should cause lots of head scratching amongst those ramping up their seed investment allocation these days.

If the returns for the past decade have been four times better for later stage investments and the risk taken in later stages is significantly less, then what is the justification for the current seed boom?

Bijan Salehizadeh is a Managing Director at NaviMed Capital, a growth-stage healthcare firm based in Washington DC. He blogs at

Image credit: Dandelion photo courtesy of Shutterstock


  • Interesting piece on an important topic.

    Seed investing in healthcare is very much in vogue. And as you say, company creation is the life-blood of our industry, so the profitability of seed-financing really matters. The question is: does the data you present really lead to the conclusion you draw?

    The first point to make is that seed investments can take a long while to mature – that has two effects: (1) its hard to use ‘recent’ datasets, like 2002 to 2011, because it will be over-enriched with the failures (because you find failure a lot quicker than success). Because you use REALIZED gains as the outcome measure, many of the successes in the sample will still be unrealised with a median 5 years since founding. (2) Using IRR really penalises seed investing – because the money sits in the company for so long. Partly thats because most seed companies are by definition early early stage, but also because they initially have low burn rates and so can simmer for several years on next to no cash working out how to be successful, before bursting forth and flowering over a much shorter period.

    The other issue is that assessing proportions of successes and failures by number of companies rather than by capital at work rather disguises the key point that the failures cost next to nothing (as long as you recognise them as failures while still in the seed phase). By contrast, the winners can in their later stages get huge sums of capital to work. Getting a 5x return on a $10M investment pays for a lot of $250k total losses (160 of them to be precise).

    The absolute golden rule for seed investing though is that whatever is done with the seed money it MUST be able to tell you reliably if this is a winner or not. You have to be ruthless and kill the seed cos before they consume much capital if they haven’t crossed the threshold for further investment. If you do that, then you can make the maths work for seed investing.

    At least you can if you don’t use IRR to calculate the return on your seed investment! If you use an IRR on the total investment made into the company you put the seed investment in, then you should get a very different picture if the capital usage curve looks like most healthcare companies (small cash in early, for a long while, with big chunks in the last 18 months before exit to pay for the proof of concept clinical studies or pivotal trial or whatever is the big value driver).

    We NEED seed investment (whether from purely commercial minded investors, or from governments and charities it matters not). Without them, there will never be the ecosystem of later stage companies for us to invest in. I would hate to see this dataset used, wrongly in my view, to frighten commercial investors out of seed investing. Its not a case of “don’t do it”, but more a case of “Do it RIGHT”.

  • Great points, David.

    Bijan, would love to hear your response, particularly to point #3 about IRR.

  • David, you make some excellent points, some of which I agree with.

    However, with respect to using IRR, and specifically realized IRR as the metric by which to present returns, it is imperfect but there are precious few other metrics that measure actual returns to LPs that one can use in analyses like this.

    Looking at unrealized IRR would give credit for unrealized mark-ups – and that doesn’t seem to be a realistic way to measure since unrealized returns don’t equate to dollars in anyone’s pockets.

    By the way, looking at the all stages of venture-backed companies, realized IRR from 2002-2011 is 17.7% – so seed’s realized IRR of 5% still looks paltry in comparison.

    Finally, it could be right that looking at 2002-2011 is too close to present. That said, I took a decade long view to show the most broad set of data accounting for multiple cycles within that time period. We could also just keep looking at the 90s and praying for a return to those days.

    But alas hope is not a strategy…

  • Good work – so much of the press in the VC realm is barely disguised hype it is refreshing to see a piece that honestly lets some of the air out of the bubble…

Leave a Reply

PE HUB Community

Join the 12525 members of PE HUB to make connections, share your opinion, and follow your favorite authors.

Join the Community

Look Who’s Tweeting

PE HUB News Briefs

RSS Feed Widget