Wednesday, November 4, 2009

Companies that measure, win: AKA Engineering Control Theory

Companies that do not wise up to the fact that measurement is MORE important than spending money are going to die. The ones with more cash (Apple, Microsoft) will die slowly. The ones with less cash on hand (start-ups, struggling businesses) will fail quicker. Marketing and advertising can be up to 40% of annual expenses, but very few companies measure their effectiveness. How can you 'correct' what you cannot measure? Why would you spend $10,000 on a banner ad, and then only measure how many 'clicks' it generated? Companies need to measure the effects of their activities in terms of the whole system, including sales. I believe so strongly in this, I started a company, Karmaback, to try to help businesses solve this problem.

Engineer-types call any measured system where the input is adjusted based on the output a 'control system'. Many people call this a feedback system. If only more engineer-types went into marketing! The major problem is that generating 'clicks' is easy, and generating sales is hard. There are SO MANY more variables than just the banner itself that goes into the sale. All of those variables need to be considered and adjusted based on the true measure of success: not clicks, sales. Finding the knobs that move sales, not just getting more 'clicks', is they key to success. In order to find those knobs, we must measure sales, and begin to create a hysteresis pattern leading to win.

Engineer-types understand the word hysteresis. The idea is simple: for any measured system, where the output is not 'easily' predicted by its input... under-correction and over-correction create a 'ringing' pattern called hysteresis. Anyone who drives a car understands this intuitively: it's the steering-wheel micro-adjustments you make while you drive. You see the car is going a tad too far right, you steer a tad to the left and vice versa. In business, we have many steering-wheels (knobs), and going the way you want is, hopefully, growing sales not clicks.

I leave you with 1 last thought. Any banner with a hot girl or an outrageous claim is going to get clicks. What knob, however, is going to get sales, and how are you going to measure it?

photo attributes:


  1. It seems my idol may disagree with me.. or not. Some of Seth's posts are a bit obscure.. this one:

    seems to say data driven decisions are flawed? WOW. How wrong can one person be!


Blog Archive