Testing P-O-P Synergy

The research techniques that helped secure an Allied victory in World War II are now being used to help marketers conquer their own battles in-store. Multi-variable testing (MVT), now being pioneered in the field of marketing and merchandising by a research firm named QualPro, enables marketers to test various sales tools such as P-O-P, promotions and price points -- simultaneously. The synergistic interactions of these multiple testing factors not only helps to validate test findings, but they also provide more accurate, real-world results."The significant thing about this form of testing is that it's a departure from what was thought to be the scientific method, testing one thing at a time and holding conditions constant," says Kieron Dey, account director and principle at the Knoxville, Tenn. based consultancy. Dey is also a trained statistician.QualPro's statistical techniques date back to the turn of the century. "It was pointed out that if you tested many things at once, you really had a real world test because the world doesn't work one thing at a time," says Dey.MVT practices were right on target in 1943 when they were used by the Royal Air Force (RAF). Mathematicians reporting directly to Winston Churchill were asked to improve the effectiveness of antiaircraft shells that protected London from German bombers. Their calculations supplied the RAF with more accurate shells that took down German planes.Through the use of sophisticated statistical formulas, MVT techniques allow researchers to measure the impact of each element in a testing program, while simultaneously measuring each factor's success in tandem with other elements. Also, accurate testing results can be achieved by testing in relatively few locations."In the 1920s statisticians typically tested half a dozen things at once," says Dey. "In the 1930s they realized you could get up to 20 or 30 things at once. By the 1950s, the testing was mainly done in the manufacturing sector."Though the MVT process has mainly been used to streamline manufacturing processes, Dey says he is now applying MVT techniques to other areas including merchandising. The results, he says, can drastically affect the way successful merchandising programs are launched.The company's experiments have found that on average only a quarter of ideas generated through traditional focus groups and consumer interviews actually improve sales. Another 23 percent of those ideas hurt sales, and 53 percent make no difference. Translated, that means that fully three out of four marketing initiatives don't help.MVT Put To The TestWhen Dey starts a new project, he begins by interviewing marketing managers, salespeople, designers, store clerks -- anyone who may have some contact with the merchandising or marketing of a product. This helps determine which factors should be tested and which have some likelihood of increasing sales. "It doesn't matter how good the statistical design is if the testing factors aren't there in the first place," says Dey. "That's a key part of what we do. We know how to shake loose the right ideas for testing."Dey recalls one project for a leading brand of sneakers where he initially identified 70 possible factors for testing, which included P-O-P and other sales-inspiring ingredients. He quickly eliminated factors that couldn't be tested easily and inexpensively. "We only test those things that are practical, feasible and cost-efficient so it can be manageable," he says.In the case of the sneaker study, Dey scaled the original 70 possible testing factors down to eight, which included testinga large, expensive permanent display;a small, portable, inexpensive display;segregating the colors of shoes on the display;and increasing the price of the shoes.A special testing matrix was designed so that all these ideas could be tested simultaneously, with accurate results in a limited number of stores. Each store ran a specific combination of testing elements for 90 days. Sales data was compared to same store sales from one year earlier.The testing factors were actually tested in 32 stores, though only 16 stores were required, according to Dey's calculations. "There are some fairly simple calculations that give you an adequate sample of stores so that you can see with enough precision the results," he says. "There are also some practical issues of what's acceptable to management and what's reasonable."Sometimes, someone will argue for more stores than I need to have," he says. "For example, with the sneakers, I could have done it with 16 stores. The reason they [the client] went with 32 was just a general sense that they usually ran tests in 100 stores. We still saved them a lot of cash."The final analysis of the data yielded some surprising results:The expensive, large display, which was based on customer focus group input, actually caused sales to drop.The cheaper display was able to dramatically increase sales, however, only when shoes were separated by color on the display.A price increase of 5 percent had no effect on sales."It's not uncommon for us to reverse conventional thinking through this process," says Dey. "We hit some tense moments at this stage. People here really question the credibility and it's a struggle for them to accept new ideas, but we always prevail."The sneaker company implemented the temporary display, segmented the shoes by color and increased the price by 5 percent. Almost immediately afterward, annual sales jumped 30.3 percent.Though MVT is well documented as a tool to improve manufacturing processes and performance, it's new to the in-store environment. Dey and his staff are attempting to expand the scope of this testing technique and prove that a statistical approach to understanding merchandising can provide accurate, real-world results."Statistics is often thought of as data analysis, and that's certainly an important piece of it, but I see it as a way to accelerate learning," says Dey.

Testing P-O-P Synergy

P2PI MEMBERS

JOIN P2PI

Related Images

Business Topic