Tuesday, May 29, 2007

Decision Making and the Venture Capital Process

I recently read a fascinating scientific paper by Anthony Bastardi and Eldar Shafir titled On the Pursuit and Misuse of Useless Information.

The paper's thesis is that decision makers often delay decisions to pursue additional noninstrumental information - information that a priori will not affect the decision at hand - yet then proceed to make use of the information, thus making it instrumental, once it is obtained.

The key issue for executives and venture capitalists is the following: one needs to determine what information may prove critical to the decision at hand, and, therefore is worth waiting for, and what information is unlikely to affect (and thus need not delay) the decision at hand.

The authors note that "people often arrive at a decision problem not with well-established preferences and clearly ranked preferences, but rather with the need to determine their preference as a result of having to decide, and they often look for additional information in hopes that it may facilitate the choice."

The venture capital process is a class example of this phenomena.

Investors often do not have a priori preferences with respect to an investment decision and need to determine their preference to fund a company and do indeed, as all entrepreneurs know well, seek additional information with which to make their decision.

The central issue is clearly identify which information is instrumental, "information that can alter what decision is made," versus which information is noninstrumental, "information which will not impact the decision at hand if it were available."

Given people like to obtain information and base their decisions on compelling reasons for one option versus another, research finds that, given the option, people will wait for noninstrumental information.

Worse yet, once the noninstrumental information is gathered, people alter their choice based on this noninstrumental information. The cost is not only delayed decision making but also poor decision making as noninstrumental information impacts the final choice.

The key take-away is that all of us, when making a decision, need to carefully think through what we absolutely need to know in order to make a good decision, rather than delaying decision making and leaning on the crutch of more time to gather non-essiential data that may contribute to a poorer decision.


  1. Hi Will

    I think the common slowness in VC decision-making has an underlying (and perhaps unstated) purpose that probably is instrumental - or is at least simply part of doing a good job.

    If a VC, or set of VCs, do not have the sense that they are about to miss out on a deal they want to be part of, then drawing things out (possibly under the guise of gathering other information which is noninstrumental) is a good way to gather important information.

    There are many things that can be revealed by taking your time, particularly when dealing with entrepreneurs. Are the investees desperate? Is the product sound? Will it scale? Are they making progress? Is the team a reality or actually just a potential team contingent on a quick financing? Does the product scale? Will user adoption, clients, etc., continue to grow? How much money do they have in the bank? How will the investee CEO behave? Will they come back with a lower price? How will we feel about this investment if we put it aside and reconsider it in a couple of weeks? Can the deal be moved a little more in our favor if we let the natural attrition of time take effect for a while?

    There's a ton of important information that may come to light if you let things drag out a little. Yes, the company may die in the process as a result, but from the VC POV, if it was that fragile to start with, then maybe that's a good thing?

    Note that I don't think any of this is bad. Gathering important information and trying to get a good deal for yourselves and your limiteds is the job of the VC. Conversely, it's the job of the other side to go into the fund-raising process knowing as much as possible about the process, with as solid a proposal, fallback positions, security, etc. That's simply the job of the other side.

    The VC funding process has long seemed to me like a fascinating dance in which both players act like they don't really need the other side, play coy, talk themselves up, and play and maintain their corresponding roles and images with straight faces. All this while trying to drive the best deal possible, balanced at each step by the possibility (and, presumably, hope) that you're across the table from a future partner, with whom you'll be working closely over the next years.

    There definitely is a sensitive spot in the middle in which a company with a great product that would have made it (if funded) gets damaged, perhaps fatally, through spending precious time and money chasing funding and encountering VC delay. That's unfortunate. It's hugely frustrating, for example, to have a key team member need to take another job during that crucial window when you're looking for funding. But ultimately it's the responsibility of the entrepreneur to be aware of those possibilities and to mitigate them as best they can.

    In summary, I think common extended delays (often for other noninstrumental information) are part of a more subtle waiting game that it's the job of both the VC and the entrepreneur to know how to play. Of course it's the less experienced entrepreneurs who are typically less aware of and adept at the game, and so they take the most knocks. But that's life, and you hopefully do learn...

  2. Will,

    There are additional reasons for determining a priori whether information is instrumental.

    A limited amount of conceivable instrumental information might suggest an irrational belief or a poor decision. Specifically, if there is no conceivable information that would disprove a belief or overturn a decision, then the "principle of falsifiability" fails to hold, suggesting grounding in faith not reason.

    To illustrate, consider the statement, "Our software product has at least 1 bug!" The statement might easily be proved by instrumental information that the product has 2 bugs, but it is currently difficult to conceive of instrumental information that would disprove the statement by proving the product to be bug-free. So, a rational statement should have conceivable instrumental information that would prove and disprove it.

    Also, it is better to determine a priori which pieces of instrumental information weaken a belief and which ones strengthen it; likewise for decisions. A classic example of failing to do this is the gambler who tells himself, "I am almost out of money. The outcome of the next poker hand will be 'a sign' indicating what I should do next." He loses the hand and uses that information to conclude ironically, "My continued losing is a sign that my luck is about to turn around. I should keep on gambling!"