The title of this post is a line in a Damon Gough (aka Badly Drawn Boy) song that I heard on my bike ride this morning. You can click on the black banner at the bottom of the screen and hear it play while you read this.

And it took me back to an email exchange that I've been having this week with some friends about "breakage" in a venture capital portfolio. Tim (Connors I assume) from USVP left this comment on my blog the other day:

i was at the ycombinator event yesterday and i think PG said they have had 102 companies through and 14 so far have been series A funded by VCs

I was surprised that the number of VC funded YC companies was so low and forwarded the comment via email to some friends in the VC business with the following comment:

I would have thought it would be much higher

To which I got back a bunch of comments about "breakage", meaning companies that don't make it. Early stage venture portfolios should have a decent rate of failure, and the earlier the portfolio, the higher the rate should be.

I've said a bunch of times on this blog that I think an early stage venture portfolio should have 1/3 failures, 1/3 money back situations, and 1/3 that deliver the returns the VC expected when the investment was made.

But that hasn't been the case in the USV 2004 portfolio so far. We are done putting new names in that portfolio and have made a total of 21 investments. We've sold three companies to date, leaving 18 active portfolio companies. And to date, we have not written off a single investment. That realization prompted me to make this turn in the email discussion:

No writeoffs yet after four years

But then, that was true for flatiron from 1996 to early 2000

And then we had breakage non-stop for two years

Flatiron had 59 portfolio companies and we eventually wrote off 20 of them without getting anything material out of them. That's one of the places I get the 1/3 failure rate from but not the only one.

But the thing of it is, we had made every single one of those 59 investments before we wrote off a single investment. From 1996 to early 2000, we had a run where we had 17 exits and no writeoffs. We went into the market meltdown with a portfolio of 42 companies (we'd exited 17) and over the next two years we wrote off 20 of them. The remaining 22 companies are almost all realized now and about half of them have been money back situations and about half have been big winners.

The 1996 to 2008 time period is not a totally normal period to be making any conclusions from, but it's interesting to go back and look at this data anyway.

One thing is clear to me and it was stated by my friend John Borthwick in his reply to my email from above:

From what i see the VC model doesn't offer a lot of visibility into failure until there is an external forcing event, the tendency to get someone else to invest and plug an existing investment w/ their dollars can blur what is in fact failure

That is true. The forcing function is usually a bad market when nobody wants to write a check. Then the existing investors are forced to look hard at each other and decide if they want to keep investing. And then, if the company is really not making good progress, the answer is usually no.

I don't know if we are getting to that point yet in this cycle, but my bet is we are getting closer. It will be interesting to revist this post in a year.

SAI contributor Fred Wilson is a partner at Union Square Ventures. He writes the influential A VC, where this post was originally published.