In 1999, advertising fundamentally changed. It was the first time advertisers stopped targeting their potential customers either by the location (website) or by their basic demographics (Women, married, household income >$100,000) and started targeting them by their behavior and intent.
The pay per click advertising model was born and allowed advertisers to show ads based on potential customers interests. So that instead of blanketing the sports section of Yahoo! to sell baseball gloves to potential customers, you could now only show them ads to your business when they searched for things (baseball equipment, baseball gloves, catcher's mitt, etc).
Take a look at advertising options today, and you can target ads to people who have actually visited your website or liked a competitor's website that sells baseball equipment. You can show ads to people who live in a particular zip code, who only drive Nissan Muranos and are fans of Chicago Cubs. Advertisers are tapping into the world of big data to get a much better understanding of who their audience is, what they like and buy and all their psychographic and purchase preferences.
If Advertising Has Evolved, Why Hasn't Most of the Retail Site Selection Process Evolved?
There are two parts to deciding on a retail brick and mortar location: the science and the art. Some retailers, like Starbucks employ more science than smaller operations that can't afford to deploy science, or, if not constrained by capital, definitely constrained by expertise. Or, in some laughable cases, retailers do a "follow the Starbucks" strategy -- which is worse because they are assuming a Starbucks customer is their ideal customer.
The "tip" of the science iceberg is setting minimum demographic requirements before a retailer will even consider a location. The problem is that this is where their science ends for the majority of retail brands. If a location hits a demographic threshold, they then typically jump right into the art side of choosing a location.
Here are a few popular brands and their general site selection demographic criteria.
Manchu Wok -- looks for: "Average HH Income: $30,000-$80,000. Per Capita Income: $15,000-$40,000. Daytime Population: 10,000."
Manchu Wok signed up with a market-leading service that has been helping companies pick locations for years. Using the basic demographic information similar to Manchu Wok's they ended up with a location that looked like they had the right demos but only produced 20% of expected sales when many of their other locations were thriving. Why?
IdealSpot, a commercial real estate analysis company, found that only a tiny percentage of people living and working in the neighborhood they moved to actually ate Asian food in the past six months. That one variable was a better predictor of success than all the drive by traffic, household income and population data you could get.
Jamba Juice -- looks for: "Average age less than 38 years. Daytime employment greater than 15,000 in 2 mile radius. Strong vehicular and pedestrian traffic counts. Average Household income of $50,000 - $75,000." A similar Juice and smoothie retailer - "Median Household income $75,000+. Population in 3 mile radius-150,000+. Median age: 18-45 yrs."
IdealSpot analyzed their data and found that none of these demographic factors are true predictors of a success location. They would not reveal what their analysis found since they treat their clients' data with utmost confidentiality.
John Prior, IdealSpot's VP of Data Science told me, "Most of the analysts and business owners trying to evaluate new locations "the old fashioned way" (spreadsheets overflowing with old data and preconceived notions) seem to have a limited view of the factors that might affect the success or failure of a particular brick-and-mortar location. Like, how visible is it to traffic? How accessible? What times of day? How much disposable income do nearby residents have?
Those are all good questions to know the answers to, but are they enough? Do analysts compare their initial data points to the actual success or failure of a business? And why are they only looking at 10 or 20 inputs? This is where most analyses "hit the wall" -- people will spend lots of time and money collecting piles of data, but then have no way to relate it to success or failure of their business. This is how businesses end up with locations that can fall far short of their expectations."
IdealSpot analyzes over 15,000+ inputs for each and every location they evaluate. In addition to the "traditional" inputs (such as demographics, retail sales, buying trends of residents lying within a X-minute drive time), they have access to social media trends and internet search patterns.
Is this the future of retail analysis? Probably, it certainly makes sense. It also makes sense to use that data to determine if an area around an existing retail store is changing. That might give the retailer early notice that a location change is in order.