How data and analytics can help retailers SmartTest across the chain

When retailers want to try something new with their stores — anything from price points, to in-store displays, to shaping offers — they don't want to blindly roll it out to all their locations and hope for the best.

First, they may try to predict consumer response by turning to market research to ask questions and track response on a five-point scale. Unfortunately, the reality in the marketplace, where so many brick and mortar retailers are fighting to survive, may deliver a very different experience and response. That's a risky gap that has cost many retailers dearly over the years. They may also decide to try it out in a few stores and see how it goes. That's a start. But a better idea is to try an in-market, experience-based process — a SmartTest — that measures and captures successful tactics at chosen locations to drive high ROI investment.

Test, measure and apply

Sounds simple: instead of guessing, implement your plan in select locations and then determine next steps based on your results. But unlocking that opportunity also means using data and analytics to understand and predict customer response across all kinds of variables.

There are a lot of questions to be answered. What do all the brand locations look like across the chosen parameters? What are the differences? How do they compare to the ones you are testing? Do you want to only test in certain pockets or more widely across the brand? It's worth the effort to find out, because establishing that baseline can give you important clues about what moves the needle for traffic and sales, guiding capital expenditure.

Sometimes it's black and white: after testing new decor and displays in certain stores, one fast food chain saw positive results and implemented as chain wide rollout Having tested and learned prior, they were able to move forward with a much higher degree of confidence.

Test-location findings may also lead to a mixed approach, for a lot of reasons. Not all stores and retail locations are the same, and the differences can clearly guide marketing choices. Are they franchise or corporate locations? Do budget constraints demand that you focus on where you'll get the most bang for your buck? Often, a modulated rollout makes more sense, and even then, different stores may get different treatments for the foreseeable future.

Maybe it's about testing an offer. Let's say an automotive services retailer is featuring oil changes. The entrance point for the oil change increases as the quality goes up, so the goal is to get the customer to trade up. The same may go for upselling tires or other products and services.

Often there is a physical entity such as an in-store display reinforcing this push. How's it working at select locations? What do the customers who use that location look like vs. those who do not? Are there socioeconomic differences? Do some locations have more bays or inherently skew higher for tire sales? You can track all these elements to see what impacts oil change sales and change the trajectory of customer behavior, first at those locations, then across the chain.

"Look-alike" locations

Let's say you have a thousand automotive services retail outlets and you set up a structured test with a subset of one hundred. You build out the prototype, put it in those locations and measure the impact. Do the people trade up more, is it more revenue, it is more profit? What does that look like? Did it bring in extra sales you weren't expecting? What is the ROI of the prototype and how long will it take to break even? You can compare these with a test vs. control to see how well it worked and performed against your key performance indicators. You can also learn why it didn't work, matching those variables against certain types of locations to keep guiding you smarter.

Then — and this is where the magic really happens — through look-alike processing and algorithms, you can find the stores not included in that initial test that look like those stores. This control group of stores is built to provide an apples-to-apples comparison. You factor in everything you've learned, then apply what worked to mirror the successful strategy and desired customer behavior, rolling out across the chain with a much higher degree of confidence and profit.

Just like brands can target look-alike customers, marketers can recognize retail locations' unique attributes that accurately reflect the synergy — and selling opportunities — between the many moving parts of a retail location and its customers and prospects. Having identified those attributes, the brand can match them up with other locations.

Banks may also approach it operationally, such as where to offer services after regular banking hours. They build out a control group of locations so it's an appropriate comparison of locations: this is what will happen vs. not doing it, controlling for all other relevant factors, from wealth measures, commute times and customer density. For them, like any of these other retail operations, it's a smart way to hedge their bet instead of gambling on what seems like a good idea across all locations.

SmartTests like these will not only be a lot closer to optimal than a "forced march" across the chain, they will let the retailer get smarter now and in the future about the relationship between their locations and their customers.

Rick Muldowney is chief analytics officer at marketing agency digm.