Let us repeat that: choosing winning sites requires both screening and forecasting.
These are not the same thing. Site screening is the practice of finding and either accepting or rejecting a site. Site forecasting is the practice of using a predictive model to estimate (or forecast) a site’s future performance (in revenue).
Even SiteSeer users confuse these steps or attempt to combine them into a single step. However, modern site selection is sophisticated, and done right, demands that the research team of a retail chain performs both of these steps.
The first step in making educated, smart site selection decisions is narrowing down the vast number of potential sites to those that fit your desired criteria.
Most chain businesses have specific rules or metrics that a site must meet for them to even consider it (and if yours does not, that’s an important prerequisite step to take). At a minimum, a site needs two things:
Let’s assume your business is Amy’s Diner, and you’ve determined over the eight years that you’ve been in business (with six locations) that your typical/target customer is consumers with children between the ages of 0 and 17 with household incomes over $75,000. You also have determined that your best-performing locations are in grocery-anchored shopping centers (vs. standalone buildings).
In SiteSeer, the site screening step happens with our Scorecards, Hot Spots and Prospects tools. A user can put all of their key performance indicators into SiteSeer and use these tools to evaluate and score locations against those metrics. From there, sites that meet or exceed the retail chain’s minimum threshold for success might move on for further analysis (or forecasting). Sites that fail those key metrics would not.
Once you’ve has identified a short list of potential sites for your next Amy’s Diner location using SiteSeer’s site screening tools, these locations can be evaluated further using site forecasting tools.
An analog model will help you compare a site to any already-operating locations in your network that are most similar. A more sophisticated machine learning model is specifically designed for sales forecasting, but these tend to work best with chain businesses that have sufficient history and data (and not for newer chains or those with just a few locations). In the site forecasting step, a “B” site according to the Scorecard might become a “C” site, or a “B” site might become an “A.”
Users are often confused when the screening and forecasting tools disagree. It is important to remember that these steps are independent of each other and not meant to provide competing results. They’re two separate benchmarks in a site evaluation process. It is not uncommon for a site to score highly on a Scorecard but return a mediocre forecast and vice versa. The initial site Scorecard might provide a list of promising sites, the deeper site forecasting process will reveal how sites fare after more comprehensive analysis.
As you read this, you might ask the same question that we hear from SiteSeer clients sometimes: can’t we just skip right to the site forecasting?
The answer is no, and here’s why. A two-step screening-forecasting process reduces the load on the more time-consuming and complex site forecasting process. Without pre-screening our locations, we risk making decisions based on bad forecasts.
Why? For most site forecasting techniques to work well, the sites we forecast in SiteSeer must be typical for that concept. A machine learning model relies on a training sample, which is usually comprised of most of the stores in a retailer’s store base. The model “learns” why certain stores perform well and others perform poorly and applies that knowledge to forecast a site’s performance. If the site is like the locations in the training file, then the model should be able to produce a relatively accurate forecast. But if the site has key differences from a typical store, the forecast will likely be inaccurate.
Large retailers have the resources to invest in data and research tools (and teams) to guide their site selection process, but what about those that are smaller but growing—and perhaps have had success with their current methods of choosing new locations? If this is your business, how essential is it that you invest in research tools like SiteSeer and data like that offered through our data partners?
Short answer: important. And here’s why:
SiteSeer is powerful for our large retail clients, with plenty of add-on tools, ample options for data, and professional services from our experienced research team, but it is also affordable for smaller retailers too.
Learn more about our powerful site selection software by taking a demo. We’ll share more about all of the tools in the platform, including those mentioned in this blog, and how we can help you expand the smart way.