Free weekly demonstrations. Register for a demo

Marrying Traffic Data and Retail Analytics

March 23, 2022

Much like the retail industry, the people counting industry has advanced leaps and bounds in the past decade. An era of break beam senso has given way to advanced stereo video technologies, Wi-Fi analytics, and a host of offshoot applications. But possibly the most welcome change is that people counting seems to make sense for almost every kind of brick-and-mortar retailer these days. Let’s figure out how traffic data plays into retail analytics and how we got here.

I’m tempted to jump straight to the ever-decreasing cost of a people counting system, but let’s start somewhere else. Over the past 5 years in particular we’ve noticed a material change in the accuracy of the sensors on the market.

For a long time, it seemed like 95% accuracy was the going rate for an overhead mounted sensor compared to a properly performed visual validation. While this is a strong number, if you’re on the fence about the value of people counting, a few percentage points may push you in the direction of doing a more limited traffic study or attacking the problem using some kind of statistical methods.

Simply put, today’s sensors measure the reality that takes places underneath the sensor.

There are still environmental and traffic flow challenges to contend with, but the accuracy rates produced by modern stereo video sensors provide a much more compelling case to begin counting.

The second part of the accuracy equation that has also improved vastly over the past few years is the ability of the solution to determine who to count and often more importantly, who not to count.

Excluding employee traffic, estimating child traffic, and grouping shoppers into a single buying opportunity allow retailers to refine their counts to answer the question at hand; how many opportunities to sell did we have today. These classification and filtering features are standard in most people counting solutions, and standard in a way that can actually be implemented to retail analytics.

Traffic Data for Retail Analytics

The unquantifiable benefit of quantifying visitors to your store

We prepare applications for our customers all the time ranging from simple conversion analysis to advanced applications like engagement measurement and queue management, but often the initial value customers see in our solution is knowing the simple fact of how many people walked through the door yesterday.

To be honest we don’t always understand why. We do know that shopper count serves as a great denominator to almost any retail analytics metric you can think up.

Measuring Conversion

Conversion rate is a great apples-to-apples performance comparison for stores across a brand and vital in a retail analytics report.

Not knowing this number is similar to praising a baseball pitcher for throwing six strikeouts without knowing if it was within two innings or nine.

Sure, Store 123 consistently out-sells Store 456, but they also receive triple the amount of store traffic.

Accounting for the rate at which visitors become purchasers levels the playing field for store comparisons.

Applying Conversion to the Revenue Equation

With the addition of conversion in the revenue equation, retailers gain another factor to move the revenue needle.

Traffic x Conversion Rate x Average Sale = Average Revenue

Traditional tactics such as promotions and up-selling aim to boost the traffic and average sale portions of the equation. Focusing on customer experience through staff training and store layout increases conversions. All three ultimately impact the bottom line.

Forecasting and Planning

As sensor accuracy has matured over the past decade so have forecasting algorithms.

This means it’s possible to have a clear picture not only of what store traffic looked like today, but what it might look like in the days to come.

This has an enormous benefit for planning store operations and making staffing decisions.

Learn more about Vea’s traffic forecasting tool >>

System Cost

I haven’t searched much but I’m convinced that there is some derivative of Moore’s “Law” that suggests that the more specialized a technology is the faster its cost comes down.

I’ll avoid specifics here, but I would say if you considered counting some years ago but were driven away by system costs, it’s probably time to look again.

It’s not just the hardware; as systems have improved and become more resilient to environmental and traffic flow changes and vendors have focused on scaling implementation capabilities the overall cost of ownership should be much more palatable for those looking to start.