A few years ago, I presented a session at the ICSC Research Connections conference on the topic of “Error and Noise in Trade Area Demographics.” Although my take on the topic was hardly groundbreaking, it remains one of my favorite presentations that I have ever given because I felt that I did a good job examining things that are always “there” but we don’t really think about.
Here was the basic idea of my presentation: There are two different phenomena that impact the usefulness of the trade area demographics that you and your system generates—error and noise.
What Is Error?
Error is easy to understand. If your system says there are 10,000 people within 5 miles of a site, but in reality there are only 8,000, this is error. Something went wrong in the process to produce the bad number. Three sources generally cause this type of error:
- Demographic data: If the demographic data is incorrect, the accuracy of your report is doomed. This is why using only high-quality demographic data is so important.
- Retrieval method: When you create that 5-mile ring or-10 minute drive-time polygon, you likely will split some block groups. Because most census data is only supplied at the block-group level or larger, you must determine what portion of the block group’s population is inside the trade area and what isn’t. If this calculation (which occurs behind the scenes in your software) is inaccurate, you are introducing error to the result. If you want to learn more about demographic retrieval, check out this blog.
- Drive-time engine: If your drive-time engine produces a polygon that is drastically different from the actual drive time in the market, this is another source of error. A low-quality drive time engine can produce inaccurate results. Even in high-quality drive-time engines, sometimes unscrupulous people set the parameters to really high travel speeds so that the polygons for any given drive time are much bigger than reality. I’m not sure if this should be categorized as error or not, but it surely is wrong.
Furthermore, once we talk about real-world trade areas for stores or shopping centers (as opposed to simply defining a ring or drive-time polygons), a third source of error emerges. If I say that a store’s 80 percent trade area is a 7-minute drive time but actually isn’t, this is a source of error—even if my demographics report captures the demographics of a 7-minute drive time with extreme accuracy, it doesn’t actually describe my actual 80 percent trade area.
What Is Noise?
Noise is a different story that may be harder to wrap your head around but is worth understanding. Any retail analyst who knows anything knows that when you create a trade area around a store, you are not shooting to capture 100 percent of customers. Different companies have different percentages, but usually the accepted range is between 60-80 percent. The reason for this is noise.
For sake of example, imagine a 1-, 3-, and 5-mile ring around a site. If you look at people within the 1-mile ring, a relatively high percentage of them would be customers. Extend out to the 3-mile ring and that percentage would be lower, and would drop even more for the 5-mile ring.
I define noise as the population in the trade area (in our example, the ring in question) divided by the number of customers in that area. When noise is lower, we have higher confidence that the trade area demographics resemble the customer demographics. When noise is higher, our confidence in this relationship is diminished. (Consult slides 32 and 33 in this PowerPoint presentation to learn more.)
So why not just use small trade areas so that our confidence is high? Inherently, small trade areas mean we are only capturing a small portion of the number of customers needed to support a store. If a trade area only encompasses 25 percent of expected customers to a potential site, how can we be sure we have enough customers available at this location to support a store? In other words, with small trade areas, we can be sure we have the right kind of people nearby but not the confidence there will be enough of them.
On the other end of the spectrum, we could try to capture 100 percent of the customers who would shop our store. However, this trade area would be huge, the noise ratio through the roof. The only thing we could be sure of is that most of these people are irrelevant to our site …
Finding the Happy Medium
As you see, there is a tradeoff in selecting the portion of your customer base to include in your trade area:
- Small trade areas: High in confidence in the type of people from the trade area that will shop your store; low confidence there are enough of them
- Large trade areas: High confidence that there are enough people to support your store; low confidence as to what kinds of people will shop the store
The middle ground is 60-80 percent trade areas. If the type of people in the trade area are crucial, you should aim for the lower (60 percent) end of this suggested range. If the quantity of people is more important, go with a goal on the higher end (80 percent).
I hope you found this topic interesting. I’d love to hear your thoughts and opinions; please weigh in below.