Without accurate site models, you can’t generate reliable sales forecasts or make consistent, quality real estate decisions. But site models are only accurate if they reflect the current state of your business, the markets you serve, and the opportunities you’re considering.
Tango Predictive Analytics lets you incorporate every variable that matters into your site models, and our experts work with yours to craft a true representation of your business and your locations. But your business, your customers, and the real estate market aren’t static—so your site selection model shouldn’t be either.
You don’t have to start the site modeling process over from scratch every time you remodel a store or add a location, but some changes will require us to either refit or recalibrate your site model to ensure it reflects the current state of your business and the opportunities before you.
- A model refit lets you adjust the market data that you’re applying your model to.
- A model recalibration refreshes the model itself—reprioritizing variables, adding new criteria, etc.
In this guide, we’ll explore how each of these updates helps you make more informed decisions, and when to use each type.
When to do a model refit
You’ll generally want to do a model refit once per year. This service is less involved than a recalibration, and comes at a lower cost. During a model refit, we review and adjust your site selection model to incorporate new data points and sales performance figures. Refits tune your site selection model to increase accuracy, but they don’t rework the model entirely.
For example, when we initially created the model, you may have had six competitors in the area and were about to open a second location. Now there are nine competitors, and your location is more established. The trade area has also gentrified, and you’d like to consider a few sites that weren’t options before. This is an excellent time to refit the existing model to the new information.
Updates for a model refit can include:
- The demographics of the trade area
- Where your store locations are now versus when the model was created
- Comparison points made for any new stores opened
- The sales performance of each location
- Market optimization seed points (potential sites for new locations)
- Competition in the trade area
- Any custom variables you’ve set
For each of these variables, we make sure the data is accurate and up to date, and we reweight them in the model based on the past twelve months.
Annual model refits help to keep your forecasts current without requiring the intensive research steps (and expense) of the initial model creation. But eventually, the time will come to do a more thorough update, and that’s where model recalibrations come into play.
When to do a model recalibration
Model recalibrations are much more thorough than refits. The work involved is more similar to the initial model creation. We start the analytics process over, updating customer data, testing trade area rules, re-profiling customers to identify shifts in behavior, and retesting all potential model variables in machine learning algorithms.
Model recalibrations represent significantly more work and come at a higher cost, so you’ll want to plan accordingly. When the time comes, a recalibration will leave you with a brand-new site selection model that’s fully optimized for current conditions.
We usually recommend a model recalibration every two to three years or as needed based on changes in market conditions. Examples of factors that could necessitate a recalibration include:
- Changes in your portfolio after acquisitions
- The maturity of your portfolio
- Newly closed or opened locations
- Major retail events (such as the changed landscape from COVID-19)
- Changes to channels
- New investments into ecommerce
- Drastic changes in consumer preference
- Plans to opening in new markets
If we see a need for a model recalibration, we may reach out to you proactively to discuss that option. And if you feel that a recalibration may be in order, connect with your account manager and we can talk through what that would look like.
Build (and maintain) the best site models with Tango
Tango’s predictive models are best in class because we refuse to take a cookie-cutter approach to site selection.
Rather than using the same algorithm across the board, we work with our customers to gain in-depth information about their business needs, pulling in more relevant inputs than anyone else. Then we use advanced AI and machine learning—paired with our own extensive experience—to tailor a site selection model for every client, resulting in the most accurate forecasts possible.
To keep those models up to date, we offer both refits and recalibrations as needed. Our utilization of machine learning makes both of these processes far easier (and faster) to implement. If your site selection model is due for an update, reach out to us to get that process started.
And if you’re ready to see what Tango can do for your business, request a demo today.
