All the Right Moves
If there’s one truism in life and in business, it’s that there’s no “sure thing.” And in a hyper-competitive retail environment, making the wrong decision in areas like pricing, marketing, merchandising and remodeling — or failing to make the right ones — can put merchants in a pretty deep hole.
Mistakes are inevitable, especially when it comes to predicting consumer demand in an uncertain economy. But having the tools to accurately leverage data and adopt fact-based decision-making strategies can go a long way toward minimizing errors and maximizing capital investments.
This is the scenario that led PetSmart, the Phoenix-based retailer with nearly 1,200 locations across the United States, to adopt predictive analytics — an outgrowth of scenario planning, but far more detailed in order to facilitate decision-making in every area of the company and to gauge the potential results of a project.
Predictive analytics “is a tool we use to prioritize and structure our investment strategy,” says Jason Lenderman, director of enterprise analytics for PetSmart. “This has tremendous implications for everything from capital expenditures to merchandise allocation. It’s enabling us to be more proactive in how we manage our business.”
Overall, predictive analytics has given the chain a “higher comfort level” with various projects and is improving PetSmart’s ability to make decisions now and for the future, according to Lenderman. “I think it’s far more accurate than scenario planning and in the end provides us with better and more relevant answers.”
The ability to base business decisions on fact-driven analytics has become particularly important in the current economic environment. “The retail business is increasingly competitive,” Lenderman says. “This means placing even more importance on making the right decisions about things that will appeal to our customers.”
PetSmart’s use of predictive analytics began more than three years ago, when senior vice president and CFO Chip Molloy joined the company. At the time, PetSmart had no formal tools in place and was using ad hoc analytics throughout the organization. “We were in a pretty good place from the standpoint of having all the raw data,” Lenderman says, “but the issue was using it to accurately measure things that were happening in the store.
“You are never going to be 100 percent sure about anything,” he admits, but predictive analytics is allowing PetSmart “to scrub the data and eliminate a lot of indecision and variability.”
A key element in PetSmart’s strategy has been its relationship with Applied Predictive Technologies (APT), whose Test & Learn software solution is quickly becoming essential for testing new retail initiatives with a minimum of financial risk and operational disruption. “We’ve been able to measure a multitude of metrics ranging from sales and margins to operating expenses and customer satisfaction,” Lenderman says.
Implementation of APT’s solution took three to four months and was “very successful,” according to Lenderman. “By then we were up and running and measuring a number of different projects. We did add in some data sets that we originally didn’t think we’d use, and we customized metrics and control groups so we could add to the efficiencies gained through using APT. Overall, any changes or tweaks were about expanding the functionality of the tool and customizing it to PetSmart’s needs.”
Predictive analytics is now being used throughout virtually every area of PetSmart’s enterprise “to meet consumer demands more efficiently, including more effective merchandising and marketing,” Lenderman says. “Over the past 12 months, changes to the stores have been made using predictive analytics, and there’s more to come.”
The important pilot phase
Basically, the Test & Learn system enables PetSmart to test new ideas and concepts before rolling them out to the entire chain. “This way we have a relatively good idea in advance of what we think the impact will be on revenues, profit and return on investment,” Lenderman says. “So far we’ve been pretty close in our conclusions. Even in a difficult economic environment, I think predictive analytics can get you to a place where you’re making sound, fact-based decisions.”
A project is first rolled out as a test, the initial step in ascertaining the project’s potential in-store impact. The test “is really more about our comfort level and whether we can execute something from an operational, merchandising or marketing standpoint,” Lenderman says. “It usually includes a handful of stores. Then, we move on to a pilot phase with more stores and measure the impact of what we’re doing.”
The pilot is a critical component of the measurement process. “As long as we have a representative sample within the pilot we have a fairly good idea of what the impact will be before we roll out a full initiative,” he says.
PetSmart is quite bullish on the value proposition of predictive analytics. “Without a doubt we are getting better answers with the APT system and giving better direction to the senior team,” Lenderman says. “It’s put us on the path to where we’re making wiser investment decisions in the stores.”
The chain is not only able to make better decisions; it’s able to render them more quickly. “Timing varies,” Lenderman says, “but we can get results from a test or pilot the day after we roll it out. However, we’re a bit more conservative. We want to make sure that what we see is sustainable.”
PetSmart typically measures a test or pilot for one quarter, during which the project team provides weekly updates on all initiatives and how the tests are trending. “By the end of 13 weeks, we’re comfortable with the results and whether they are sustainable,” Lenderman says.
- The delight of a dinosaur: inspiration for improving the customer experience
- August is the Time to 'Keep That Drumbeat Going' on Internet Sales Tax
- Veteran Massachusetts Retailers President Honored for Service
- Three ways Macy’s has reduced friction for customers
- Expectations of stronger job growth should light a fire under retail sales for the rest of the year