Phone: (413) 727-8897 email:

Friday, March 03, 2006

Some Things Just Don't Work Well

When I first started reporting how great remote site feeding was to my best friend, The Murph, we got into long discussions regarding the various techniques we could use to analyze operating data. One of the top of the line MBA analytical techniques is multiple regression. Since some contracts required prices per manday with extra charges for special meals, we thought MR was the way to go. You could never get the general managers and operations directors too much food cost data. Serious craving!

Armed with our APL keyboards and an IBM mainframe, we started loading meal statistics into tables. I had weekly food cost numbers available by category and summary. A typical record would include weekly data as follows: Week, Breakfasts, Lunches, Dinners, Midnight Meals, Take Out Lunches, Mandays, Food Cost (in dollars). We included employee meal counts in the details and totals.

After many attempts to get a better fit using multiple regression analyses, I found simple linear regression always worked better. Food cost is highly variable. If you find a large fixed component, it is often a danger sign.

I recently ran regression statistics on some quarterly data available at the SEC's Edgar site. One steak house concept with predominantly company-owned units runs a 36% food cost. The regression line is actually 40% variable with a negative fixed component. Try to figure that out!

The notes to their financial statements are very clear and open. However, they do not address menu pricing strategy.

My best guess is they have a menu pricing strategy which adds a fixed dollar profit into every menu item to cover overhead. So to calculate a menu price, they multiply the standard recipe cost by 2.5 and add a factor to cover the overhead costs. Maybe this was a reaction to the increase in oil prices.

Click Here For More Information

No comments:

Restaurant Data Pros

web counter