Everyone knows the value proposition for chain store real estate research:
“If you avoid closing one or two stores it pays for itself.”
So why are people so hesitant to invest in analysts, predictive analytics, and information systems to support better real estate planning and site selection?
I’ve been asking this question for nearly 20 years, and I think I’m finally beginning to see the answer.
It’s about measurement.
Here’s how the argument goes:
Summary (simplified): If we invest in better research, we can use a predictive model to estimate sales performance for proposed sites. If we know the population and income of the trade area, the distance to our competitors, and the quality of the site, we can use the weighting factors in the model to calculate the sales potential. We’ve heard that it’s possible to have a model where the estimate will be +/- 20% reliable at least 80% of the time. In the past we have only been +/- 20% correct 65% of the time.
Therefore, the avoidance of bad stores alone will more than cover the investment in people and tools to build and use the predictive model.
Great! We now have something that can be measured. We use the model to predict sales and we only approve deals where the predicted sales deliver an ROI greater than our minimum requirement (“hurdle rate”). After a year or two we can compare the performance of the stores opened before and after the model was used and see if the results were better. If they are better, we keep using the model. If not, we go back to the old way of estimating sales.
Enter the skeptics.
Legitimate questions can be raised that erode our confidence in the value of modeling:
- How do you define a trade area?
- What about the fact that competitors all have different effects based on their brand strength, management, and location quality?
- The quality of the store manager in our units can account for a difference of +/- 20% in sales. How do we measure that?
- The economy goes up and down in general and also in specific cities depending on how well the businesses are doing. Does the predictive sales model take into account these fluctuations?
These are just a few of the many questions that make it hard to trust predictive models. They all relate to the problem of measurement: determining the factors that drive sales and consistently quantifying their values.
The skeptic concludes that it’s a waste of time to create and use predictive models because they can’t accurately measure what’s going on in the real world.
The skeptic is throwing the baby out with the bath water.
If the sole objective of predictive modeling is a precise sales forecast, disappointment is certain. However, we are not trying to win a contest for counting how many jelly beans are in the jar at the county fair.
The objective of research-based real estate planning and site selection is profitable growth.
The value of analytics in other industries has been demonstrated beyond doubt. The management of investment portfolios, marketing campaigns, and many other complex systems can be greatly enhanced with decision-support systems that guide executives as they invest precious capital. But the value of analytics is not simply found in the accuracy of specific predictions; it comes from the discipline of “best practices” that include the people, the processes, and the tools that marry “art and science” and adapt to the changing landscape with winning strategies, tactics, and decisions.
Best practices can be described and measured. It may not always be possible to reduce the measurement to a scientific “cause and effect” relationship, but market leaders have the vision, commitment, and creativity to set high standards for themselves. In the chain store world, successful companies apply these standards to real estate planning and site selection. This only makes sense when you consider that real estate is by far the largest asset on the balance sheet of every chain store company (even franchisors have nothing if the franchisees have poor real estate).
Chain store real estate analytics is a topic that needs to move from the real estate department to the Executive Committee. A good or bad predictive model will never make or break a company if it isn’t used in the context of “best practices” that start with the CEO, the CFO, and the other senior executives whose jobs include hiring, training, organizing, and managing everyone in the real estate decision-making process.