When I was interviewing candidates for analyst positions at my former employer, I would often engage in a friendly exercise of “cognitive estimation.”  We all use cognitive estimation  to answer a question, when the exact answer is unknowable, or difficult to know without expensive or difficult measurement.  Clinical neuropsychologists use normed tests of cognitive estimation to evaluate the impact of brain injury, dementia,  and Alzheimer’s disease in patients.  Back at the market research company, I made up my own examples to assess the “common sense” and thinking process of applicants. Here’s an example: 

“How many gallons of ice cream were sold in Massachusetts in the month of July, 2007?”  

You could spend a lot of money with research houses, point-of-sale tracking systems, or field observers, and come up with some rather precise numbers.  Or you could take the total number of Massachusetts residents, multiply by some reasonable estimate of daily ice cream consumption on warm days (something more than a spoonful and less than a quart, I’d guess) and multiply that by 31 days, and get to a good-enough, back-of-the-napkin answer.  This back-of-the-napkin analysis is exactly what seems to be missing in some startups, and it’s exactly the kind of analysis that marketing needs to be doing before product development engages too much engineering talent.

Today, I visited a very interesting company, Intellivid, which develops software that aids in the analysis of output from video surveillance cameras.  I’m interested in companies that are participating in the video surveillance market, and I focused on Intellivid, because, in my estimation, they were focused on the right segment of the opportunity, which is retail.  “Why retail?” you may ask.  “Haven’t you seen all the cameras in the airports?  What about the thousands of cameras in the city of London?”  Well, let’s see.  According to the folks at Intellivid, the top 100 retailers in the U.S. operate 150,000 stores, generating $3 trillion in sales and suffering about $40 billion in “shrinkage,” the industry’s term for “Someone walked off with the goods without paying.”  That’s about 1.3% of sales, which doesn’t seem bad, until you stack that up against the operating margin of retailers.  I checked Collective Brands, which owns Payless ShoeSource and Stride Rite. Their 2006 operating margin was just under 6%.  What would the retailers pay to reduce shrinkage  by 25%?  Enough to make a market, I’d wager.

Let’s get more specific.  What would a Payless ShoeSource store be willing to pay to reduce their shrinkage?  Collective Brands did about $2.8 billion in sales in 2006 and had about 4,800 stores, which works out to just under $600,000 in sales per store. The company’s operating margin of about $166 million works out to just under $35,000 per store.  If their shrinkage is near industry norms, that means their stores lose about $8,000 a year to shrinkage.  Again, that may not seem like a lot, but compare that to the $35,000 operating margin per store, and if shrinkage could be eliminated, they get almost a 20% increase in operating margin.  What do you think that would do to the stock price.   OK, what if it was only a 5% increase?  The stock would still jump. So what’s the cost design point for a Payless ShoeSource security system? You figure it out, and then you can put your developers to work.  I think that the folks at Intellivid will be focusing initially on opportunities where the total per-store shrinkage is greater than at PayLess ShoeSource and the cost per event is greater than a pair of shoes.