Monday, August 17, 2009

Engineers make better models of business.

The reason is simple. Engineers understand not only the idea of a "dependent variable" and an "independent variable", but its practical use.

For example, an engineering business model would undoubtedly have a page full of seemingly "independent" variables which control the actual income statement/cash flows/and so forth. How do I know this? Because I've made dozens of such models myself, and undeniably people are amazed and appreciative that on one page of the spreadsheet they can see all the variables that (when changed) cause an effect to the master model.

But engineers are (hopefully) wise enough to know that just because a "model" has seemingly independent variables... it is likely and almost certain that many of these variables are not independent at all.

A simple example I will use from a business model I crafted recently. By the way, if it doesn't roll into a full set of Income Statement, Cash Flows, and Balance Sheet, then its an incomplete model for Business (this is one of the things engineers often miss, and stop at the revenue/cost model).

Here is the example:
I want to model how I am going to acquire customers. For this, I assume that I will advertise. I assume that my cost to "expose" a customer to my product is A (e.g. Google is about 10cents). I then assume that it takes B exposures to gain an interested lead. I then assume that C% of interested leads become paying customers... yielding the following 'dependent' variable:

"Cost to acquire a customer" = A * B * C

Engineers are fine with this math, and then quickly fill out a line for "Marketing Expense" in the Income Statement, based on the number of customers I wish to acquire, and maybe even get so complex as to write a routine that "balances the budget" for 5% growth of customers... or some-such.

Of course, Engineers also know that this seeming dependency is only as good as the assumptions behind it... and that reality may be far from this. For example, is it true that a % of customers will convert? Is it true that the cost to gain interest are 'fixed'? In fact, having been in management, I can say undoubtedly this model is not at all close to 'reality'...

Nevertheless, it does give a good base starting point (one heck of a good one), especially because with this model, you can see the effect of any changing assumptions in the model. For example, what if it costs $1 to expose a customer!!! EEK!

So, engineers do make better models of business, hopefully know they are wrong... and experienced entrepreneurs know how and when to use these models.

Hint: Don't use these models to 'predict' customer growth, only to estimate Marketing Expense!!! Customer growth has FAR too many variables to rely just on advertising!


  1. does this example (picture) work for science too or for just engineers?

  2. Nothing enamors like the lights of NYC. Brooklyn Effects(TM) has made a module for Polished product Star X giving flashes and light holes adding enthusiasm to all recording. Is there the light choices, as well as advances for FCPX to easily change starting with one clasp then onto the next. FCPX Effects


Blog Archive