Big Data and Labor Costs

11 Nov

Labor is one of the two largest costs for a restaurant operator (the other being inventory). Yet it is one of the most mismanaged aspects of the restaurant business. Our experience has been that restaurant operators do not like to think of the expense side of the ledger. They focus much more on revenue, and chase revenue-generation schemes like Groupon while ignoring the most manageable part of their business. But, maybe this is why it is neglected: expenses are the objective score card for operations performance. While sales may fluctuate with macro events – those outside of your control – the operator must actually manage expenses. So the expense side of the ledger is actually a reflection of how well the operator can operate a business. Ironically better management of expenses results in pure profit, so it is curious why restaurant operators neglect its importance.

Typically, managers try to minimize labor costs to 20% – 25% of total sales. Too high and you’re burning money. Too low and you’re likely providing poor customer service. To do this, managers must make an estimate of the day’s sales. Sometimes they break the sales into hourly revenue. Sometimes they copy and paste the same schedule from the week prior. It just depends on the sophistication of the restaurant.

A very common methodology is a statistical mean average of same day last year, and same day last week. There are all types of assumptions baked into this methodology but, for an on-the-fly methodology, this is probably the best one out there. The net result is almost always an assured mismatch between customer demand cycles and labor expenses. Why? There are many factors that influence when customers turn up. Just a few…

- Weather

- Events

- Seasonality

- Road traffic

- Macro economy

- Micro economy

- Demographic shifts

- Promotions

It’s impossible for a restaurant manager to actually crunch all this data to build a sales forecast. He would have to procure data feeds, format the data, build an algorithm to weigh these external data sets, then produce a sales forecast for food and beverage using his POS data. But let’s just say he did. What would the end result look like?

We ran a pilot program with 60 restaurants to determine how much money an operator could save by using big data techniques to forecast labor needs. The results were not surprising. Across all cases we saw a minimum profit increase of 20%. This resulted from the tapering of overstaffing. We also saw an increase in customer service, attributed to fewer periods of understaffing. Over time better customer service may lead to increased revenue and repeat visits. This metric is hard to quantify exactly. A full WhatsBusy Wine Lounge Case Study is included in this post.

While the results are exactly what has been seen by Wal-Mart and other large retail businesses when turning to big data, it’s not something we expect restaurants to adopt. In fact all the restaurants in the pilot program, over the course of 3 months, attributed the profit increases to “black box luck”. You can lead a horse to water.

Leave a Reply