Fail Fast, Fail Often

There was an article last week in the Wall Street Journal talking about an apparent change in the entrepreneurship/VC funding model.  Riya, Meebo, and others were cited as poster children for the new restraint.  The core idea was that entrepreneurs are taking advantage of the availability of capital to fund for long periods, often several years, rather than the traditional 12-18 months.  So why does this make sense? Why raise a bucket of money when a thimbleful will do?

The classic venture model has been to fund to milestones 12-18 months out.  In consumer web services, there are only two meaningful milestones --  (1) are you getting a lot of users and (2) have you figured out how to make money?  We use other metrics in other sectors (like management, product, etc.)  as proxies for real economic progress.  We also use them because (we believe) they would have residual value in an asset sale or merger. 

None of this is true in consumer web services. You're either hot or not. Second place generally sucks.

The problem is that it is hard for entrepreneurs and VCs to know a priori if something is going to be a hit. The only way to know is to try, and trying takes time and money.  So here's the real rationale for what it makes sense for these companies to raise "a lot of money" and not blow it.  They have to run lots of experiments.

By now we are all well-acquainted with the observation that software is cheaper than ever to produce.  But that is only half the story.  The other half is that it takes several iterations -- several trials -- to hit it big.   

Imagine you have a low-burn consumer internet company and you think you can do your next build for $2M (offshore, open source, etc.)  Imagine further that there is a 1 in 20 chance that you could be the next [insert fantasy outcome here]. Angels are lining up with $2M in hand.  VCs are waving $5-20M checks at you.  Everyone says this is a $10M pre-money company and you own 50% today.

Assume you have a 5% chance of Being Big on the $2M raise, and a 95% chance of nothing.  The chance of Being Big if you raise $4M is 9.75% (1-.95*.95).  This is because you can iterate twice at 5% probability each.  The chance of Being Big after raising $20M is 40.1%.

Of course, each $2M has a dilution to you as the Founder. As the graph below illustrates for this hypothetical example, the risk-adjusted ownership (diluted ownership x probability of success) increases as you raise more money.  (This conclusion is not universally true in all situations.)   

The key to this thinking is to resist the temptation to spend like a lottery winner. Raising the big VC round isn't winning the lottery; it is the purchase of a deck of weekly lottery tickets. 

This is how Munjal Shah described the move to Riya 2.0 in the WSJ article.  It was the realization that the first experiment, while a success by many measures, wasn't enough of a success relative to other options.

The larger-than-expected VC rounds in consumer internet deals are perfectly rational outcomes, for the entrepreneurs who understand the trials of consumer marketing.  Failure is baked into the calculus of the opportunity.  The key is to fail fast.  Set metrics ahead of time and be decisive. Because time is money -- literally.

Entrepreneurs who practice this discipline are just doing what VCs do every day.  Venture Capital is a hits business, too.  Companies often fail.  Time is money here, too.  Failure is part of the process.  We, too, are looking to fail fast and expect to fail often.  That's why funds are getting bigger, too.

Comments

[jim Forbes]

Fail fast fail often?
Why not just take the entire contents of a fund, wander over to a high stakes roulette table and put it all on double Zero?

heh heh

Jim

"what about respect the money"?

[nr]

thats EXACTLY the theory of why we had raised more funds than was necessary.

however, we had assumed (this is all theory of course) that the probability increases after every iteration i.e. it does not stay at 5% after the 2nd iteration because you have valuable learning from the first iteration...

the example, i always give is youtube... hotornot (1st iteration), the normal video uploading (2nd iteration) and then the embeddable player (the 3rd iteration)...

the key is to keep staying in the game...

[Matthew Markus]

Your analysis of the situation makes sense; however, if each launch has an independent probability of success then, given the time-value of money, doesn't it make sense for VCs to make 10 separate bets in parallel? In other words, parallel failures are far faster than serial failures. Furthermore, won't LPs be upset when they discover that their cash is earning market-rate interest in a single start-up's bank account instead of being put to use?

[Jeremy Fain]

This is definitely a great analysis, mixing common sense and quantitative proof. However, I don't believe having more cash to experiment increases the chance of becoming the next [insert fancy outcome]. Although I guess experimenting is the right way to operate from the entrepreneur view point, investing regular amounts of cash in more projects instead of a lot of cash in a few projects should be the way venture capitalists experiment with their investees.

Indeed, I guess the reason why start ups raise more cash are much simpler: i) there is a lot of VC money available since funds are getting bigger as you point out; ii) there are too few excellent projects since the very best projects are often founded by entrepreneurs who succeeded in a prior venture (often during the hype). The number of good entrepreneurs is definitely a bottleneck in the equation.

The law of supply and demand makes valuations increase and more cash than necessary is consequently raised by entrepreneurs.

[Vijay Chakravarthy]

Well, one has to be careful that this doesnt become a proxy for clear and rational thinking before embarking on executing a business plan. While like.com has some decent prospects of making money, meebo and riya.com should have asked themselves - how do we monetize this? What are our competitive advantages?
I still dont know/cant figure out how Meebo plans to make money, and how some technological innovation (ajaxian firewalls?) or moves by incumbents in the IM space will let them keep their existing user base.

[Steve Sarakas]

I disagree - here's why. A chief tenet of innovation is vision. Consider two extremes so I can make my point. In one case money burns in piles indefinitely because there is zero vision - no one knows what they're doing and development is chaos. At the other end development revolves around a vision which is entirely accurate and completion intersects the market perfectly. These are two extremes, of course. But between them, obviously, vision has to matter. That's my point - no allowance is given here to vision. Many consider vision black art. Others consider vision indeterminate, hence, development is at once "making vision into reality". Nonsense. Vision isn't trying to predict the future. Vision isn't taking chances, either.

1) Vision equals money every bit as much as time.
2) Knowledge plus emotion generates vision.

I may be trying to explain something in what is perhaps, an inappropriate place.

I remember in the early 90's, for example, when the "build consensus" mantra became the rage in management. I watched while meetings were conducted in which it was genuinely thought, that if you could get a room full of people together, none of whom understood a problem, you could somehow discuss it to death and thereby arrive at a consensus that was valuable. Zero times a million still equals zero. Invariably, the consensus was bad. It could have been how to treat an emergency stop on a piece of machinery, or what to put in a GUI. There was no vision. Instead, there was an attempt to purchase it with intellect. You cannot.

Projects fail all the time. Obviously those involved thought at one point there was enough money. Decisions were made all along the way, which at the time seemed correct. Imagine playing Kasparov in a game of chess. You can make every move as carefully as you like, but in the end you will lose. I see some of this here. Analyze it to death if you want. If that worked, then why so few good ideas? More than that, why do so few even recognize good ideas when they see them?

Any idea how many VC blogs out there in which a principle is laying out post analysis of the social networking phenomenon, as if they fashioned it themselves? Good grief. Here's one: "We're looking for the next YouTube, that's what we want. It has to already be a success, and you have to have a great team, and if there are a lot of other investors interested, then that's what we want. Call us, we're waiting". I hear skeletons rattling. The truth is, they don't have the foggiest idea what is coming.

You can "fail forward" in some reckless game of chance if you want, but I'll put my life, my time and my money, into vision any day.

[Andy Sack]

I really like your post. Good insights. THanks

[Alex Bernstein]

If you're going to go for a run in the dark you should have a solid idea which direction to start.

Seems that the trial and error process could be substantially improved with some smart branding and marketing work. If you understand your target customer.... I mean reeeeeaaaally understand what makes them tick, what they hope for, what they hate, where they hang out, and what motivates them, you'll have a foundation for trying to figure out how and more importantly why, when, and where your product may play a role in their life. (thus reducing the need for a blind trial and error cycle)

There is a science to discovering the intersection between the benefits of a product/service and the emotional and rational drivers of your target consumer. Unfortunately this science often isn't in the toolbox of most of the tech start up CEO's we've worked with.

Smart brand planning would put the entrepreneur in a position to take the smaller investment and still have an increased chance at success because the smart brand strategy work done up front will enable the business to move to market faster than the trial and error crowd bumping along in the dark.