Posts from August 2008

I’ve been here in Edinburgh for two days. I leave this afternoon for Leeds to give a talk at the Old Broadcasting House.

But the entire time I’ve been here, I’ve been admiring this structure in the center of town. We’ve been running around from show to show and haven’t had time to do any tourist stuff so I didn’t get to visit it up close.

Instead I took this photo from a couple blocks away on the way to dinner and sent it to flickr. By the time I got home, the answer was on Friend Feed:

Fred, that is the Scott monument, the figure sitting under the monument is Sir Walter Scott. – Bob

Thanks Bob!

I looked it up on Google and now I know a bit about it.

The Scott Monument was built between 1840-46 as a memorial to the writer Sir Walter Scott (1771 – 1832). Designed by architect George Kemp (1795-1844), who won a competition with his gothic design, the public can climb the 287 steps to the top of the monument for spectacular views of the city.

I hope to get some time to do that walk up the stairs today.

But this experience got me thinking about "shazam for photos" When you see something that you don’t know what it is, why not take a photo and have the technology do the rest?. If it’s fast enough, you could become a tour guide without having to do any work.

I love the lazy web.

The news that some Facebook employees are selling their stock privately naturally focuses on the price/valuation (a deep discount to the $15bn that Microsoft paid).

First of all, a deep discount makes sense. Most privately held venture backed companies now regularly value their common stock for the purposes of issuing stock options at fair value. These valuations are often done by third parties and are called 409a valuations. I see many 409a valuations every year and most value the common stock at somewhere between 20% and 50% of the last round’s preferred stock price.

So if Facbook employees are selling stock at company valuations between $3.75bn and $5bn, that makes sense. It’s 25% to 33% of the valuation Microsoft paid.

And many have said, including me, that Microsoft’s $15bn valuation was a premium to what Facebook would have gotten and should have gotten in a truly market based deal. Microsoft wanted a strategic relationship with Facebook and was prepared to pay a premium for it. Some have suggested it was even to Microsoft’s benefit to put a premium valuation on Facebook so it would not get bought cheap by someone else. I am not sure about that last part, but it’s clear to me that a financial investor (and the public markets) would value Facebook at somewhere around $7bn right now (maybe less).

So if $7bn is a better approximation of the market value in a financial transaction, then the $3.75bn to $5bn private sale prices make even more sense. They represent 50% to 75% of "fair value". And those kind of discounts are what buyers of secondary private shares usually demand. They are purchasing stock that cannot be resold easily and they are becoming shareholders in a company that they will have basically no ability to control or impact.

All of this said, I think this is a great thing. I have said before that we need a more active secondary market for founder shares and shares purchased by early investors in venture backed companies. As exits via IPO and M&A become more and more difficult, I sure hope the secondary market will step in to fill the gap.

Related articles by Zemanta

• More on the Facebook employee stock sale: "Dozens" selling, "investment firms" buying
• Lack of an IPO market makes it hard for stock option plans and weird things happen
• Facebook to let employees sell some stock options – at internal $4 billion valuation
• Report: Facebook letting employees unload stock options?

Seedcamp London is europe’s closest thing to YCombinator and similar programs here in the states (funny to say that because I am writing this in St Andrews University in Scotland).

The deadline for applying to this fall’s Seedcamp London program is this coming Sunday, August 10th. If you want to apply, go here.

Applicants are whittled down to 40 teams who come to interview in-person for a shot at the final 20 the third week in September. Union Square Ventures will be there that week to participate although unfortunately it does not look like I’ll be there due to the Web 2.0 keynote I am giving midweek in NYC and a few other unmoveable obligations.

Today is the deadline for a ‘fast track’ video pitch using disqus and summize. The winner of this gets right to the shortlist for interview day. Right now there aren’t that many submittals to this so take a shot at your best elevator pitch and go for it right now.

We are big fans of these programs. They give first time entrepreneurs much needed visibility and credibility that are hard to come by. So if you are doing a startup in europe and looking for a breakout opportunity, check out Seedcamp London. But don’t procrastinate because the deadline is approaching fast.

If you looked closely at the model and the assumptions in my last post on this topic, you’ll note that I assume an early stage venture fund will lose money on 1/3 of its investments, breakeven to make a little bit on 1/3 of its investments, and will make good money (5-10x) on only 1/3 of its investments.

What that means is that 1/3 of your investments will produce all of the proceeds as the breakeven to make a little bit deals simple go to cover the losers. But that is a hypothetical model. It’s actually even more stark than that.

Every really good venture fund I have been involved in or have witnessed has had one or more investments that paid off so large that one deal single handedly returned the entire fund.

Let’s take the hypothetical $100mm venture fund that I modeled out in the previous post. The average investment in that fund is $5.3mm. If the fund invested that much in one company over a number of years and owns 20% of the business and the business is sold for $500mm, then the fund’s 20% is worth $100mm. It’s a 20x multiple on the investment. Not a common occurrence, but it happens in this business.

When that $100mm is distributed, one deal has returned the entire fund. That is huge because then the other winners will typically collectively return from one times the fund’s value to three times the fund’s value. After carried interest fees, that gets you to the 1.5x to 3x NET to LPs I talked about in my first post on this topic.

When I look at a venture portfolio that is fully constructed, but not yet fully invested (like our 2004 fund is right now), I like to look for the deals that can return the fund. I think we have six or seven. That doesn’t mean that those six or seven are the best deals in our portfolio right now. A few of them could be complete busts. But we have six or seven deals that sitting here today I can honestly construct a scenario where they will return the entire fund when they are sold. I hope that one or two will actually do it.

Because if we can get that kind of hit, it will make the rest a lot easier. As many have pointed out in the comments to the first two posts on this topic, venture investing is a hard way to make money. We need some big wins to make the model work.

Some will read this and suggest that our business is all about swinging for the fences. But I don’t think so. There are hitters in baseball, the best hitters in fact, that hit balls out of the park when they are just trying to make good contact. That’s how you have to do it in the venture business. You try to make 20 great investments and you work with them closely in hopes that four years in you have six or seven that have home run potential, and after ten years, you maybe hit one or two out of the park. If you try to hit every one out of the park day one, you’ll strike out way too much and the fund won’t work out very well.

We’ve debated this endlessly over the years, Revolver vs Rubber Soul. My 15 year old daughter Emily is doing a program in Oxford this summer and was asked to write about her favorite Beatles album.

Revolver is a genius work of art. … It is terrifically well balanced and consistent,
giving the listener a taste of all the Beatles have to offer. Revolver
is by far the Beatles strongest and most impressive album.

There’s a lot more including the opportunity to leave a comment and join the endless debate.

Almost five years ago, I started this blog and had no idea where it would lead. It’s been a rousing success by most measures.  This morning, the AVC blog was #20 on Techmeme’s leaderboard, in the rarified company of journalists and bloggers I have read and looked up to for years. It’s still one of the top 1000 blogs worldwide as measured by Technorati in terms of incoming links. And it gets over 100,000 unique visitors to the web page and feed each month. But most importantly, it has built a community of regular readers and participants, many of whom I have never met but consider friends and advisors.

So I felt it was time AVC looked like a proper blog. When I kicked off this adventure in social media, I just picked one of the standard typepad templates and ran with it. The look has not changed much in the past five years. A couple years ago, with the help of the team at FM, I widened the center column and added dynamic width sizing and made it readable on most mobile browsers (something valleywag has still not managed to do). And I (without FM’s help) went on a widget binge that cluttered up the sidebars and led many to call this the most ugly blog out there. The result was apparently effective but very homegrown and not particularly polished.

There were two other problems. First, when I set this up, I had no idea it would become a serious thing. I picked a typepad URL (avc.blogs.com) and went with it. Big mistake. I’ve wanted my own domain for years but the one I wanted (avc.com) was not available. So I just keep going with the avc.blogs.com URL.

And, the page load times on this blog have been horrible. I’ve cleaned up my widget habit lately and that has improved things. And a while back, I set my center column to load first which has helped a lot. At least you didn’t have to wait for the content, which is what everyone comes here for. But the slow page loads hurt the loading of the comments which come from disqus. All in all, the page load time has become a serious issue. We tell our portfolio companies to focus on super fast page loads and yet my blog is the slowest one out there. That’s not right.

So this summer, I decided to do something about all of this. First, I noticed that avc.com was no longer active. So I engaged my friend (and blog reader) Marty Schwimmer to obtain it for me. Marty is really good at this stuff and he got avc.com for me last month. It was not cheap but it was absolutely worth it to me. From now on, when you want to visit this blog, just type avc.com into your browser and you’ll come here. It’s long overdue and I am very happy about it.

Second, I engaged Nathan Bowers, someone I met through this blog, to re-design it for me. We talked a bit about what I wanted and he sketched out a design. I liked it and he did the rest. It was really just one rev with a few minor tweaks.  I really like the result. I hope you do too.

We really wanted the look and feel to be familiar. The goal was to keep it similar enough to the old design that people would not feel like they landed in the wrong place. We got rid of the left sidebar completely. That messes up mobile browsers big time and I have come to dislike left sidebars. We also put mobile browser detection into the template and are rendering the page for specific mobile browsers when we detect them. We’ve tested the new AVC on iPhone, blackberry, windows mobile, and a few others. If it doesn’t work on your mobile browser, please let me know in the comments. It’s important to me that we get this right.

We’ve moved everything else (except one thing which I’ll get to next) into the right column. Search is at the top right (it’s where I always go to search in Firefox so it makes sense to me that it’s there). Navigation is next, then a bit about me, then the two ways to get this blog delivered to you (RSS and email). I really like the way Nathan did the RSS and email stuff. Very simple and elegant. Then some stats because I love stats and I am sure some of you do too.  And finally the widgets. We kept four of them; FM ads because they generate good money to charity, twitter because I blog a lot of short things you might care about there, mybloglog because this blog is a community and it’s important to me to know who has been here, and flickr because it’s always on your lists for the widget to keep and because it was the first widget to make this blog. That’s it and I hope to keep it that lean going forward.

There is one more “widget” on this blog in the footer. That black band at the bottom plays music. It replaces the tumblr widget that has been on the upper right of the blog. Click on it and you will hear my music posts on tumblr in reverse chronological order. We currently cannot pop this player off the page but that is coming soon. If you post music to tumblr and want this player on your blog, leave me a comment and I’ll get you the javascript you need.

That’s really it. Comments are still powered by disqus and you. The content is still powered by me and nobody else. It’s the new AVC, same as the old AVC.

The comments on my initial post on this topic went right at the VC’s compensation – management fees and carry – and their impact on returns. So at Ken Berger’s suggestion, I will change my planned post for today and address the issue head on.

The returns venture firms get on their investments are called "gross returns". I didn’t mention them in my post yesterday because I wanted to focus on the "net" returns to the LPs. I said 2x was the lowest attractive return on a venture fund and I meant net to the LPs. That means if you invest a dollar in the fund, you get two dollars back.

However, the fund has to get a lot more than $2 back on its investments to get its investors $2 back. That’s because before the investors get their money back, the fund takes a management fee. And if there are profits, the managers of the fund take a carried interest on the profits. Our fund takes 20% and that is the carried interest that most funds take. However, there are funds that charge 25% or even 30% carried interest fees. Some think the best funds charge the highest carried interest fees. Market theory would suggest that is true. But I am not sure that it is. I think we have a very good firm and we charge a standard carry. But that’s for another day, if at all. It’s a tricky subject to talk about.

The management fees don’t go directly to the fund managers. They pay for the costs of running the business. On small funds, that’s about all they pay for. On big funds, the management fees can get large enough to pay very significant salaries to the fund managers. Management fees are all over the map but range from 1.5% per year for large funds to 2.5% per year for smaller funds. And they typically tail off after the first five years to much lower percentages to reflect that the work of putting the fund to work is largely over.

The carried interest is only paid on gains. So if they fund makes no money, no carried interest is paid. But if the gains are large, the carry will be large too.

Back in 2003, when Brad and I started Union Square Ventures, we built a model of our fund to show how we thought the fund economics could work. At that time, we planned a $100mm fund. We ultimately raised a $125mm fund. But the model I am going to talk about is based on a $100mm fund.

Here are the base assumptions we made in our model:

I am not going to address all those assumptions in this post but if you have questions, post them in the comments and I will respond to them.

This model is basically our game plan. Some things changed in our execution of the fund, namely we raised $125mm, and we invested even less in the initial investment of "concept" investments, and we did more rounds for those really early stage companies. So we did 21 investments instead of 15 and we did more rounds of smaller amounts. But even though those seem like big changes, they did not effect the main drivers of fund capital allocation very much. It looks like the average investment amount will come in close to what we modeled and it looks like the capital allocation between seed, first round, and later stage investments will come in pretty close. Right now we are doing much better than 1/3, 1/3, 1/3 hit rate, but let’s wait until it’s all over to see how that comes out. You just never know in the venture business.

Here’s what this $100mm venture fund model produces:

Total Management Fees: $20mm

Total Invested Dollars: $80mm

Total Proceeds on Investments: $322mm

Total Gain on Investments: $242mm

GROSS Multiple: 4x ($322mm/$80mm)

GROSS IRR: 39.2%

Multiple Incl Mgmt Fees: 3.2x

Gain Incl Mgmt Fees: $222mm

IRR Incl Mgmt Fees: 32.9%

Carried Interest Fees: $44mm (20% of $222mm)

NET Multiple: 2.56x

NET IRR: 28.6%

So to make it really simple, a fund needs to get 4x (in this case $322mm on $80mm of invested captial) on its investments to generate 2.5x in distributions to its limited partners.

Paul Margolis of Longworth Ventures did a post a year and half ago on this topic and came up with similar numbers (although the difference between gross and net in his post are lower than mine).

The differences between gross returns and net returns are large in the venture and private equity business and it’s important to understand them and be clear about what numbers you are using when you talk about returns.

This is our first full weekend at the beach since memorial day. We picked a good one. We started it off on friday night with a beach bbq with guitars and singing around the fire under the stars. It doesn’t get much better than that. But if it does, it’s the annual sand castle building contest on Atlantic Beach in Amagansett.

You get all kinds of sand castles in this contest. I took photos of a few and they are in a set on flickr. I also reblogged Beach Obama on my tumblog.

The kids did their part with a bake sale to raise money for a good cause.

There’s nothing like to beach to ease the strains of a tough week.

When I write about venture fund returns, there are always comments and questions that lead me to believe that the economics of a venture fund are not well understood. And since most of the readers and commenters on this blog are people who work in the startup ecosystem, I think its important that the economics are better understood. So I am planning on some posts on this topic in the coming weeks.

The first thing I’d like to tackle is how returns are calculated and why the reported returns are often higher than people might think.

Unlike hedge funds or other investments you might be familiar with, venture funds do not call all of the committed capital upfront. If you commit $1mm to a venture fund, you will receive capital calls about once a quarter for anywhere from 3% of your commitment to 10% of your commitment. That is because it takes time for a venture firm to put the money to work and they’d rather leave it in your bank account than have in in their bank account.

And because venture investments are generally made in a number of rounds staged out over a three to six year period, even when the venture firm finds an investment, the amount they invest in the company upfront is a percentage of what they will eventually put to work.

So the money flows into venture funds slowly.

The money also flows out slowly. When a company is sold, the proceeds are distributed. There are some situations when the money is not distributed, but they are not that common and it’s not useful to dwell on them right now. Venture investments require long hold periods, typically five to seven years. So it’s common for a venture fund to have to wait five or six years to make its first distribution. Most venture funds have a ten year life and are often extended a few more years to get all the distributions out.

The low end of acceptable performance for a venture fund is to return two times invested capital to its limited partners (investors). If you put all the money in day one and waited ten years to get 2x back, that wouldn’t be a particularly interesting rate of return. It’s 8% to be exact, not the kind of return an investor would think is acceptable for a ten year illiquid investment.

But if you map out the cash flows that I described in this post, they look something like this:

So, if you invest $1mm into a ten year fund and get back $2mm, you will likely be earning something closer to 13% than 8% just because the $1mm wasn’t tied up in the fund for the entire ten years.

I’d like to make a couple points about this to be clear. First, as I said before, I think getting 2x invested capital back is the absolute low end of acceptable performance in a venture fund and I sure hope and expect we can do better for our investors. But I needed to use a simple number and so I went with 2x.

Second, these cash flows are conservative in my mind. Our 2004 fund has called about 65% of committed capital about four years into its history. This model shows 75% called after four years. So, money can often be called even more slowly than I showed. And waiting for years 8, 9, and 10 to get your distributions is also very conservative. Our 2004 fund returned about 40% of committed capital in its first three years which is likely to happen when you get some companies sold early on in their development. And we are seeing more of that kind of thing these days.

If you model capital being called over a slower pace and distributions coming back earlier, you could theoretically get to annual returns of over 40% for a fund that only delivers 2x on committed capital.

Of course, annual rates of return are not the only measure that investors look for in a fund. They want to get the highest absolute returns they can get. And 2x is just not that exciting. I think 3x or better is what it takes to deliver top tier performance in the venture capital business and that’s what we shoot for.

I’ll post tomorrow a bit more on the 2x vs 3x issue. It’s something I am constantly thinking about.