Building successful stores starts with choosing a great location. Your site must draw new customers and provide a better, more convenient shopping experience than your competitors. That s where a good site selection process comes in. To do it right, you need to collect data about your current stores, competitors, and markets. With that information, you can determine the factors that drive sales and build a forecasting model to help you decide whether a site is a good choice. Models can produce highly accurate results, but they are only as good as the data that drives them. Bad or missing information = a failed forecast.
Minimizing bad data: is artificial intelligence (AI) the answer?
We always tell our clients that site selection is a combination of art and science, and we stand by that. An accurate forecast (to choose the right site) depends on good data, but some data is easier to collect than other data. The world is changing constantly, which does pose challenges when it comes to reliable data acquisition, but there s no question that the human element is an essential part of the equation.
Recently, forecasting using AI has become a hot topic and there is understandable excitement about the possibilities. Those of us in site selection AI look forward to the day we can automate the collection of data and measure a building s accessibility and visibility, parking adequacy, and traffic flow into a building s parking lot using Google StreetView and image recognition software. We re eager to embrace programs that scour the internet for facts about prospective customers, and build on that knowledge. It sounds great, doesn t it? What if you could simply command drones to collect real-time imagery of your sites and your competitors, track your consumers and translate their behaviors into a demand calculation for your concept?
While there have been great strides in the field of artificial intelligence and machine learning, it is a young technology. For now, robust data acquisition and accurate model development still requires human interaction and expertise. To ever replace manual data collection and expert human modelers, there are many challenges to overcome first:
- Data needed for site selection is often external to a company s databases. Yes, much of the data needed for analysis and modeling is stored in company databases and business intelligence systems, but there is other important information that you must collect from elsewhere.
- The best data sources still must be licensed, despite the proliferation of data on the internet, whether it be demographics, competitor data, or traffic generators
- Key site factors such as visibility, accessibility, parking, and signage must be manually collected. Someday technology will allow us to measure site attributes from afar, but for now it still requires human labor and judgment.
- Site selection is driven by local data, while artificial intelligence in retail site selection requires big data. The key to machine learning and AI is having a large training set of data that allows the system to learn and improve. Site selection is often constrained by a limited sample of data. For example, you might only have 100 stores and thus only 100 different site profiles to learn from.
- Every business must make certain assumptions. Many of the critical factors driving successful models come out of discussions with employees and customers.
AI does offer many advantages
Although we don t yet live in world dominated by artificial intelligence yet, there have been many exciting advancements in technology that is applicable to site selection today. Consider web scraping, wherein large amounts of data are extracted from websites and compiled into datasets (check out companies like AggData). Open-source or crowd-sourced data datasets maintained by volunteers and made available for free to the public are becoming more common (check out OpenStreetMap, which provides geographic market data for use in analysis and forecasting). Want to know where your trade area consumers live, work, and shop? Companies like UberRetail and Streetlight use data from mobile phone activity that can help businesses understand how and where consumer live, work, and shop.
Advancements in technology will continue to change how we do business. While technology definitely offers many advantages and we should adopt new practices when it makes sense to do so, it s important not to blindly accept that new is inevitably better than the proven methods that have worked in the past. Technology is not superior to the human mind. As a 2016 article published by Deloitte puts it, algorithmic forecasting has limits that machine learning-based AI methods cannot surpass; human judgment will not be automated away anytime soon.
