From the 2005 Letter to shareholders:

If we take a simple example of a single store, then a comparison of SSS from year to year is fairly straightforward. If a store does $1 million in sales at a 10% operating margin this year, generating $100,000 in operating profit, and does $1.1 million in sales next year at the same operating margin of 10% generating $110,000 in operating profit, it will report a 10% increase in SSS. Now, let’s add another dimension. Imagine that this same store spent $500,000 to improve the store experience during that year. The 10% increase in SSS generated an additional $10,000 in profit. Whether the $500,000 investment makes sense or not in hindsight will depend on the future performance of the store. Obviously, if the store only improves by the $10,000 in profit, the $500,000 investment doesn’t make sense. I believe that companies that pursue SSS growth at any cost often fall victim to these traps.

In reality, the calculation of SSS becomes even more difficult. Individual retailers are opening, closing, and remodeling stores all the time. In this context, the simple comparison of a single store breaks down. Let me explain. Imagine that a new store opens on January 1, 2006. In the first year of operation, this store would be excluded from a company’s calculation of SSS because most calculations only include stores that have been open at least a year. A retail store matures over time and the first year of sales is often at a level that is a fraction of its potential. If we assume that a store opens at 60% of potential and matures to potential over four years, we know that this store will grow by 67% over that period of time (from $6 million to $10 million, let’s say). On that $10 million-in-sales store that opens at $6 million in year 1, the SSS increase over the next three years will average 18.6% per year, with the higher growth rates occurring in years 2 and 3 rather than year 4.

At the end of that period of time, the $10 million store may be at a relative steady-state, and let’s say it is earning at a 10% operating profit, or $1 million per year. The key question is not how well the store did from a SSS standpoint but rather how much money was invested to generate the $1 million profit. If the store cost $5 million to build, a $1 million profit represents a 20% pre-tax return on investment, which is attractive. However, if the store cost $20 million to build, the 5% return on that investment would not be attractive at all. Nevertheless, regardless of cost, the store would still have reported 18.6% compounded growth in SSS.

Complicating things further and bringing things even closer to reality, the more stores that are opened relative to the outstanding base of stores, the higher the SSS metric a company can produce, regardless of whether the new store openings make economic sense or not. If the mature stores (i.e., those that are over four years old) grow at a 1% rate and the new stores grow at the 18.6% per year rate (remember, it is likely that in years 2 and 3 the rates are materially higher than the 18.6%), then mathematically it is simple to show that the more new stores that are opened, the higher the SSS calculation. Only after a period of years will one know whether the new store investments actually made sense and actually contributed to the creation of value.