The search giant today rolled out new ways for marketers to understand the in-store impact of their ads.
(Page 2 of 3)
“This price optimization tool allows a merchandiser to manage products at a level they never could before because there just aren’t enough hours in a day,” Smith says. “The tool does the number crunching for us. It compares the demand curves for particular products across classes and sub-classes with other demand curves for other products or for other years and selling seasons.”
Merchandise managers are free to either accept or reject the optimization system’s recommendations, and to implement recommended price changes either immediately or gradually until the recommended target date, though not without the company holding them accountable for the final performance of a pricing strategy. “We’ll report the history on what was recommended, what the buyer did, and call attention to how we’re performing,” Smith says.
Getting the system to produce good recommendations, however, takes extensive and careful preparation. While preparing its system, the Goody’s I.T. team transferred three years’ worth of product sales data-2004 through 2006-to Oracle, a process that took nearly five weeks.
Then the retailer’s merchandise and store operations managers spent about three weeks in daily 3- to 4-hour meetings to establish the business rules to regulate the markdown recommendation process, such as setting minimum improvements to margins in order for the system to recommend a markdown. Other rules require the system to factor in the costs of labor and materials used in executing markdown recommendations.
“A lot of work went into these meetings to set business rules with a great deal of involvement from people in merchandising and store operations,” Smith says, adding that the planning process helped to win buy-in from merchandise buyers and store personnel.
Involving store operations staff as well as merchandisers also helps to determine just how granular and flexible to get in applying business rules across product lines and store locations, Smith says. One of the biggest challenges in deploying a price optimization system, experts say, is deciding how far to drill down into product data across classification and SKUs as well as into regions and subgroups of stores, or even individual stores. The more detailed the data on products and stores analyzed by the system, the more precise the recommendations on how, when and where to change prices.
“If we limit the system to chain-wide data, it can only recommend optimized prices across the entire chain,” Smith says. “But the more flexible we are with the business rules in analyzing data throughout product lines and individual stores, the better Oracle can make recommendations.”
Value of recommendations
The value of recommendations applied to more products and stores, however, must be weighed against the additional time and costs involved in executing the -recommendations, he adds. “The offset is the question of whether more granular recommendations can be executed in the stores, and that’s a discussion that’s been going on for some time-how granular to go with the recommendations,” Smith says. “If we say the Oracle system can mark down products on any SKU on any day in any store, we’d be doing nothing but changing prices every day.”
And after factoring in the cost of labor and materials involved in changing the price tags on marked-down SKUs in individual stores, the net effect on margins might not be as favorable as initially expected, Smith says.
One step at a time
Goody’s will take it a step at a time, he adds, first getting comfortable using the system’s price recommendations for broad levels of product classifications across the entire chain, then experimenting with recommendations for sub-classes of products and across regions of stores.
“We’ll look for the lowest level in the product hierarchy where our team feels the demand curve is accurately portrayed, though this can differ from one apparel department to another,” he says, adding that it will compare the value of such information against the cost of execution.
In extending the recommendations into small groups of stores, Goody’s expects to mostly focus on groups with similar weather patterns for analyzing common demand trends as applied to pricing strategies. “We probably won’t go lower than climatic zones,” Smith says.
Reaching a high level of performance in the use of optimization systems, however, also takes careful planning in other areas, experts say. Although many retailers have reported good results, “we also know of large retailers where it hasn’t worked as well,” LeHong says, declining to name the under-performers.
To make these systems work, he and others say, retailers need to have accurate product data throughout their enterprise software applications and trust in the ability of the optimization technology to make worthwhile recommendations on when and how much to change prices.
“Any price optimization system can’t work without a good data repository and central data management,” says Sahir Anand, retail industry analyst with research and advisory firm Aberdeen Group, a unit of Harte-Hanks Inc.
With accurate data synchronized across the enterprise, including consistent product definitions and sales data, an optimization system is capable of producing usable price recommendations. Without accurate data, merchants will never come to believe those recommendations, experts say.
“It boils down to trust,” LeHong says. “For merchants who have been planning pricing strategies with spreadsheets and a gut feeling about sales trends, it can take time to build trust in a software application’s recommendations. But these applications are complicated, so you have to use them, then do your homework and check the recommendations and see if they held true.”
The more Goody’s uses its price optimization system, the more it expects to find new ways to apply its information, Smith says.
For example, retailers may decide to run markdowns more or less frequently based on customer response to markdowns in month-to-month or year-to-year comparisons, which can also be analyzed in the system.
“If a retailer marks down men’s pants once a month, but finds customers over extended periods very responsive to price changes, it might make more sense to do markdowns more frequently,” says Chris Morrison, vice president of planning and optimization solutions for Oracle.