How to use data to control food costs (Part 2)
In part 1 of this series, I touched on the impact human behavior has on the deployment of a restaurant back-office system. Two motivations to deploy a back-office system are: 1. improving process controls and 2. automating repetitive tasks. In this post, I'd like to describe some alternatives to traditional back-office systems. There are layers to this onion, and you do not need to view it as an "all-or-nothing" game.
Inventory and food cost
Back-office systems today are typically built around a perpetual inventory system that attempts to continually track your items on hand, the blended costing of your inventory and the theoretical food costs. To achieve this, you must configure the system with a lot of data, like your recipes, and maintain a lot of data in near real time (at least daily). And, those entries you make every day need to be near perfect if the summarized numbers are going to be meaningful at the Company level.
There are alternatives to controlling your food costs. They may not be as accurate as a well maintained perpetual inventory system, but if you believe in the Pareto Principal (80/20 rule), you may think that trading off a little accuracy to dramatically reduce costs and effort is a worthwhile exchange.
For example, instead of calculating theoretical food cost using the actual prices paid in each restaurant, you can calculate a "Plate Cost" for each menu item using an average price for the month across all your restaurants. Multiplying this plate cost by the quantity sold of each menu item in each restaurant each day gives you a reasonably accurate theoretical cost with a fraction of the effort. It also takes your managers out of the bookkeeping business and allows you to assign the task to someone better equipped for the job.
Tools for scheduling next week's labor are usually part of a back-office system. These tools require creating a forecast of sales or customer traffic, then applying a set of factors for each job that helps the manager determine the right number of people per shift per job. The end result is a labor schedule for every employee for every day/shift for the upcoming week.
The reality of labor management is that it depends more on good decision making in the moment than how good your plan was at the beginning of the week. If the traffic right now is much more or much less than the forecast, your labor costs for the week will be determined by how your managers react. Perhaps, you are better off coaching and training your managers on how to make good labor decisions instead of a great sales forecast.
For example, you can coach them on where and when they could have made better decisions using yesterday's data. A report commonly used by our clients maps the customer traffic to the labor hours so you can easily see when good and poor labor decisions were made. Coaching them and providing regular feedback on their labor decisions can assure you have enough labor to service your customers, but no more.
Another feature of many back-office systems is suggested ordering. The calculation of this suggestion is tightly linked to the work required to calculate theoretical food cost. The system calculates how much of each ingredient is being used based on the point of sale (POS) data. It might be very accurate, but at a cost of effort to keep all the data up to date and accurate.
For large restaurants with large liquor supplies and large upscale menus, there may not be a good alternative. For the rest of us, there is. We are going to apply the 80/20 rule again to focus on the items that make up the majority of the purchasing dollars with far less effort.
Many menu items have one ingredient that is the most expensive by far. The filet of sole, the double cheeseburger, the white truffle casserole, all have one ingredient that is the most expensive. By focusing on that one ingredient per menu item, and tracking the quantity sold of each menu item, you end up with a good estimate of the usage of that ingredient. This gives you the amount you need to order to replace the inventory you used since the last order.
Do you have a lot of preparation in your restaurants each day? Back-office systems usually create a prep schedule for these items using the extensive recipes you have loaded into it. As long as the recipes are up to date with the latest ingredients being shipped by your distributor, the prep schedules will be very accurate.
Let's apply a similar technique to the one I described above for purchasing to help you create a prep schedule. We are going to use the quantity sold of each menu item to create this prep schedule. Some menu items, like a baked potato are easy because there is a one to one relationship between the menu item and the item to be prepped. Count the quantity sold last Tuesday at lunch, and you have a good prep schedule for baked potatoes next Tuesday morning.
For prep items that are used in many menu items, the same approach is used, but you have to add up the quantity sold of all the menu items that use that prep item. Our clients know this approach as dimension maintenance, and they often apply this technique to create prep schedules.
Every restaurant company is different, and the best techniques for controlling the key costs of the business may reflect those differences. What you can afford to do in a location that generates $300,000 in sales each year is understandably different than if your restaurant generates $5 million. Good controls are predicated on one thing, and that is good data. It need not be perfect, it need not be precise to three decimal locations. The data just has to be good enough for the purpose you are applying it to. How much perfection can your company afford?
Dave Bennett / With prior experience with Dunkin’ Brands and IBM Global Services, Dave Bennett has led Mirus through its formative years and its current growth phase. Dave holds a B.S. in Business Administration and MBA from Northeastern University.