Mining Historical Data to Optimize Your Menu and Drive Sales
How do large restaurant brands with thousands of locations turn their menus on a dime, either when sales are slumping or a competitor sneaks in with a solid menu engineering strategy?
The price and value war is back in full swing thanks to the rollout of McDonald’s new $1, $2, $3 menu – a strategic approach that’s showing early signs of success in winning back the 500 million customer visits executives say they’ve lost since 2012. It also has fast food competitors quick on their feet, responding with their own bogos, promos, and value menu items. Dollar taco fries, anyone?
Which got us thinking … how do large restaurant brands with thousands of locations turn their menus on a dime either when sales are slumping or a competitor sneaks in with a solid menu engineering strategy? And what can other emerging and growing brands do to make similar business decisions on what to throw out, what to add, what to keep, and what menu items to put on promotional repeat?
The Future of Menu Engineering
At its core, menu engineering is trying to answer the question: are your inventory, recipes, menu design and pricing strategy driving the maximum profitability?
A well-engineered menu can drive traffic, increase sales and guest satisfaction; it also helps control the cost of food and waste. A poorly engineered menu can easily drive a restaurant out of business – as seen in many episodes of Kitchen Nightmares.
While menu engineering, food cost analysis, and calculating plate costs against commodity prices and marketing promotions are not new concepts, they are getting renewed attention because of sophisticated predictive analytics and the application of machine learning to the process.
The advanced modeling is also the secret sauce helping financial planning and analysis, culinary, marketing and procurement teams optimize their pricing decisions, deliver predictable promotions at the right time, and effectively manage stock levels to support overall predicted demand.
Menu & Recipe Modeling to Accurately Predict Future Success
Today, data needs to be analyzed in context to produce any meaningful insight. But not everyone has a staff of expert statisticians waiting in the wings to analyze your next move.
This is where machine learning, advanced forecasting and predictive analytics come into play. Connected data in an intelligent back office platform can analyze a whole host of historical data collected. Think KPIs like:
- Sales Per Head
- Total Sales
- Gross Average Check Per Cover
- Total Food Costs
- Food Cost Per Head
- Stock Value
- Stock turnover
- Actuals vs. Theoreticals
- Basket Analysis
- Food Costs %
- Trending promotions as a % of gross sales
- Trending total promotion amount
- Most expensive promotion item
- Commodity prices by region
Analyzing the ongoing collection of data produces predictable insights, like:
- The impact of ingredient costs
- The profit margin of a proposed menu price
- Your most profitable menu items for a particular region during a particular type of event
- You could even work backward from a sales goal and profitability target to assess the best ingredients and a recipe that increases average check sizes
These insights can help above-store planners make sweeping decisions to drive overall company profitability and revenue growth. When deployed down at the store level, intelligent back-office platforms can deliver on-time recommended insights for optimizing a sudden increase in traffic or an extra case of apples. (NOTE: this is far more applicable in organizations without centralized stock management.)
Did Starbucks Use Advanced Modeling to Determine a Price Increase?
When you’re making a decision to increase the price of a beverage at Starbucks, there’s a lot on the line. The reputation cost from media coverage alone would be an interesting data point to measure … but let’s stick to more scientific data points.
So how did Starbucks make that “venti” of a decision to raise the average price of their beverages by 1% in 2016 … and again in September 2017? Perhaps they modeled the impact of a price increase against a whole host of data points including the impact on customer loyalty market by market. What’s the threshold that would drive their most loyal and highest-paying customers away? What would a loss of 2% of their lowest paying and least frequent customers do to their sales? And as a premium brand with premium prices, they may have been willing to let some cost-sensitive customers go to a competitor while maintaining or even increasing revenue.
Starbucks also increased the price of their tall drink, a move that may have been intended to persuade customers to choose the Grande (more value for the cost) – but with a higher profit margin.
This is just one example of the many factors that go into menu and recipe decisions today. It’s both exciting to see the more dynamic modeling and more complex interlock that must take place across departments along with the sophisticated forecasting, menu engineering and cost, and price analysis technology. It’s a welcome future for restaurants who may not have the resources or proprietary tech to accurately predict the future with such precision.