Creating realistic trade areas with business analyst's included datasets

[AmandaGray]

Hi,

I'm working on a project to create trade areas based solely on information available to me through business analyst.  I'm using infousa's listings of business locations and annual sales (grocery stores, in this case), and Esri's "food at home" expenditures as input data.  Then, I'm creating trade areas around the groceries by several different methods.  I noted that the total food at home expenditures and grocery store sales were extremely close when viewed at the county level (so, a county that spends $X in food at home contains grocery stores with roughly the same dollar amount in sales).  This was great news.  However, when I try to look at a smaller scale (the city instead of the county, for example), I'm not seeing such nice alignment.

Has anyone had experience using the included datasets in business analyst to create trade areas?  If so, do you have any tips for creating accurate trade areas?  Drive time methods are getting me closest to accurate results, but I'm seeing food at home sales much, much greater than grocery store sales, still. 

Thanks in advance for any help.

[Jason_RobinsonRobinson]

Hi,

I'm working on a project to create trade areas based solely on information available to me through business analyst.  I'm using infousa's listings of business locations and annual sales (grocery stores, in this case), and Esri's "food at home" expenditures as input data.  Then, I'm creating trade areas around the groceries by several different methods.  I noted that the total food at home expenditures and grocery store sales were extremely close when viewed at the county level (so, a county that spends $X in food at home contains grocery stores with roughly the same dollar amount in sales).  This was great news.  However, when I try to look at a smaller scale (the city instead of the county, for example), I'm not seeing such nice alignment.

Has anyone had experience using the included datasets in business analyst to create trade areas?  If so, do you have any tips for creating accurate trade areas?  Drive time methods are getting me closest to accurate results, but I'm seeing food at home sales much, much greater than grocery store sales, still. 

Thanks in advance for any help.

Amanda,

A couple of clarifications are you using Append Data to summarize data to your trade areas around grocery store points? Are you comparing consumer spending variable with the sales volume data for the grocery stores?

Regards,
Jason R.

[HoustonRudy]

Amanda,

I doubt you will be able to ever get the data to match up.  First, grocery stores have a very large trade area.   They can pull customers from well out of the city limits and beyond.  Secondly, I don't believe the data is accurate for each individual store.  It's my understanding the sales are based on a combination of the number of employees and the sales data available to the public (Annual reports, etc.).

[SusanZwillinger · Accepted solution]

Creating accurate trade areas for business locations that are not your own (i.e. you don't have customer addresses and sales data) usually involves multiple analysis techniques.  The revenue estimates in the infoUSA data are not very accurate . . . so you can only use this as a rough guide of the "order of magnitude" difference in sales between different businesses.  However, there are some things that you can do to try to analyze the locations.  Below are some ideas:

1) Use the Business Analyst menu-->Trade Areas-->Data Driven Rings wizard to create rings based on the sales revenue data from infoUSA.  For example, you can have a 1 mile ring represent 10000 from the revenue field--which translates to a store with $10 million in sales.  (The infoUSA data reports sales data in thousands, $000)  You can then use this information to vary the size of the drive time areas . . . the assumption being that stores with larger revenue will have a larger trade area and longer drive time.  Obviously, that assumption could be wrong in some cases, but I would say that the assumption would hold in many cases.  If you have data from your own stores, you can use that information to guide how you choose to represent the size of the rings relative to revenue.  You also may need to separate your grocery store data into different types.  A super Walmart or a Super Target is going to have a very large trade area, but a local grocery store or specialty grocery store could have a much smaller footprint that serves only an immediate neighborhood.

2) Since many people tend to shop in their neighborhood for groceries  (i.e. they want to be able to get home before their ice cream melts), you might want to consider using one or more neighborhood boundaries as a trade area (for example, you can download a shapefile from Zillow if you happen to be working in an urban area).  You can translate these larger areas into a trade area by block group and then do manual customization of the trade area by selecting or removing block groups.  Use the "Create Trade Area From Sub-geography Layer" in the Trade Areas tools in the Business Analyst Tools toolbox in your ArcToolbox window.

3) Try the Huff Equal Probability Model in the Trade Area wizard if you can weight the importance or market share for each location.  This will help you understand the relative market share for each location, but it won't give you an easy way to compare one trade area to another; rings or grids are better for that type of analysis.

4) If you have time, you can use the Sales Potential Modeling under the BA menu (another form of the Huff Model) to create custom trade areas for each location based on an attractiveness factor for each of the grocery stores in the area.  Once you have the results of the analysis, you can translate the block groups to a trade area by exporting the block group IDs to a table and importing that using the Business Analyst menu-->Trade Areas-->Standard Geographies wizard.

5)  You can use the Territory Design toolbar to create trade areas (either automated or manually by block group) using the grocery stores as seed points and the "Food at Home" expenditure data for size parameters.

Regards,
-Susan-

[DavidLuke]

szwillinger you are really good with your suggestions. I like it.