Now let me give you an analogy to help make clear of what I mean by this model.

Let's imagine that you're trying to make money mining for gold.

In that analogy, value is the money you make from the mine,

I, the idea, is like the location of the mine.

If there's no ore in the ground,

you're not going to be able to make any money mining for gold.

D is your efficiency and effectiveness at extracting the ore from the ground and

turning it into gold bullion.

And E is the price of gold, it's a factor that's outside your control,

but that really has a very big influence on whether you make any money.

So the analogy here is that the location of the mine is like the idea.

What, your effectiveness at mining the ore is like your

effectiveness in developing the idea.

And the exogenous factors are like the price of gold,

factors that are outside your control that can determine your success.

Now of course, all three factors matter.

A good idea is important, your ability to develop that idea is important, and

what happens to the factors outside your control are also important.

So the real questions is, how important, relatively speaking,

is the idea in determining success?

Now what you'd really want to have, to answer that question,

is you'd like to have a very large sample of new ventures, you'd like to look at

the outcomes of those new ventures, and then you' like to go back in time and

look at the raw idea that the entrepreneur began with.

And you'd like to understand how much of the variance in outcomes is

explained by variance in the quality of the raw ideas.

In statistical terms, this is equivalent to asking, how much of the variance in

the outcomes, if you think about all of the variability that would be expressed in

the outcomes of these new ventures, how much of that variance is explained by

the variance in the quality of the raw ideas those entrepreneurs started with?

Now that question is really impossible to ask in a statistically

rigorous way across all new ventures, but I did some research that

looked at a very narrow domain in which we could get some data and

actually start to understand the explanatory power of the raw idea.

And I'd like to just take a few minutes and tell you about that study.

My collaborator, Laura Cornish, and I studied over 100 products

that were created by the crowdsourcing web platform quirky.com.

Quirky developed products such as these,

an articulated power strip, a cable management system,

a divided water bottle, and yes, a tofu press.

These are examples of four products that Quirky developed.

So the question we asked is, if we look back at the ideas as they were

originally expressed, how much of the variance in outcomes for

these commercial products could be explain by looking back at

the variance in the quality of the raw idea as originally expressed?

We'll how do we actually measure the quality of the raw idea?

What we did was to take a large sample of consumers,

show them a concept description of the original idea, and

use a simple purchase intent survey, which asked them to select one of five options,

from I would definitely not buy to I would definitely would buy.

This is called a five-box purchase intent survey, and

it's one of the standard techniques in market research for

assessing the quality of an idea and the eventual sales of a new product.

What we found was quite interesting.

The first thing that we found was that experts are actually not so

good at predicting success.

We found that four random consumers, just randomly selected consumers,

actually provide a better estimate of idea quality than seven experts.

That is,

consumers themselves are the best single indicator of the quality of an idea,

and it's the best single predictor of whether the product will sell or not.

And then, in fact, four consumers provide a better estimate than even a larger

number of experts than even seven experts.

If you're interested in the details of that study, I've provided

the citation to the original paper here, and you can look at the details.

Let me show you how the ideas were originally expressed.

This is idea 133, the tofu press, and it basically shows a little picture,

and says that, when you're trying to cook tofu,

wouldn't it be great if you could squeeze the water out of the tofu, and

wouldn't it be useful to have a tofu press to be able to do that?

So what we did is we took ideas expressed this way, we showed them to consumers,

we asked them to indicate on that five-box scale how likely they would be to

purchase a product if it were developed around that idea.

We did that for more than 100 of the products that were developed by Quirky,

and then we did some statistical analysis of how well

the consumer purchase intent predicts the eventual sales rate.

I won't go through all the statistical details with you, but

I will say that the model explains about 6% of the variance in the sales rate.

And that is if you look at the variance in purchase intent for

all those different product ideas,

that explains about 6% of the variance that you see in the sales rate.

Now, as a technical note,

let me just say that we actually take the log of the sales rate.

And we have to do that, because the sales rates vary from a few thousand units per

year to over 10 million units per year for the most successful products.

If you take the log of the sales rate, and you use that in this statistical model,

then the raw purchase intent, that is the purchase intent as expressed by consumers,

explains about 6% of the variance in the sales rate.

Now on the one hand that seems like not very much of the variance, only 6%.

On the other hand, if you look at the fact that it's the log of the sales rate,

it's actually quite a big number so that a one standard deviation better idea,

that is a one sigma better idea, as measured by consumer purchase intent,

corresponds to about a 75% higher sales rate for that product.

There's good news and there's bad news here.

The good news is that some of the variance in outcomes can be explained by

the quality of the raw idea.

The bad news is that there's a lot of variance in the outcome that's

unexplained, that is that's likely explained by D or E, and not by I.

Let me give you an additional caveat about idea quality and

how important the idea is.

So while on the one hand the idea is important,

on the other hand, there's a lot of unexplained variance.

The other point I want to make is that many ideas are readily available publicly,

and so they aren't really a source of unique advantage.

This is actually an image, taken from a slide, from a student project in

my 2009 Wharton workshop on web-based products and services.

The idea is for a company called CabStalker, which would allow you

to take out a smartphone and hail a local transportation solution.

In effect,

what this team was proposing is almost identical to the functionality of Uber.