By Mary Wagner
Everyone remembers last year’s online price testing debacle at Amazon.com. The poster child for Internet retailing sparked a major backlash from consumers when it gave different DVD buyers different discounts to test sales response. When shoppers found out — and some who’d been offered lesser discounts complained — the buzz got so negative that Amazon changed it policy to give everyone the lowest price at the end of a test.
It was a lesson for online merchants in how not to do price testing: seemingly at whim, on a heavily-visited site featuring products with high price recall, and without appropriate communication to consumers.
But there’s random price testing—and then there’s strategic price optimization. The concept of price optimization is about trying to get the right product in front of the right consumer at a price that drives sales and value all the way back into the supply chain, and it’s core to retailing. As part of that, merchants have long conducted market research so as to develop pricing strategies. The difference is that now the Internet can give retailers information on con-
Consumer response to pricing in real time, and they can
adjust pricing online instantly if they want to. But when experimenting with
online pricing under the umbrella of price optimization, retailers must take
care both in how they position tests with consumers and in what they do with
the data.
Amazon’s early misadventure in online price testing aside, there’s growing interest in the subject among retailers. One indicator is the number of software vendors now offering analytics to retailers that track Internet sales trends and inventory in Internet time. While it doesn’t automatically push new prices out to the web based on the data, analytics software presents the information to retailers in a structured format, runs proven algorithms against it, and then issues price strategy recommendations, making what’s been an art increasingly into a science. “The web offers opportunities retailers never had in the offline world,” says Craig Zawada, senior analyst with McKinsey & Co. “So the questions become: To what extent are they real opportunities? And can you use your online channel not only for pricing more effectively along the dimensions of forecasting, time and segmentation, but also to inform your offline strategy?”
Knocking the barriers
While they have focused for 20 years on wringing much of the excess out of the supply-chain side to lower costs, merchants today are looking to the customer-facing side of their operations for profits. And the online channel’s unique ability to measure response and make rapid adjustments at much less cost than changing store shelf prices or producing catalog supplements remains a tantalizing prospect. So, as Amazon did, merchants are beginning to use the web to experiment with “optimizing” revenues by tracking a variety of customer behaviors and preferences around price. They’re testing online markdown strategies that help them capture greater returns once merchandise hits the virtual sales floor. They’re seeking out safer ways to do it, with tools and techniques that aim to squeeze out profits or sales without impinging on brand promise. And they are so far, not inclined to share the details.
“The web removes a number of barriers to gaining information about consumers’ tolerance to price elasticity,” says Paul Bieganski, chief technology officer of Net Perceptions Inc., a provider of online marketing and merchandising solutions. “You have almost real-time feedback regarding consumer reaction to price. But I have yet to see anyone vary price online to obtain price elasticity information successfully—or at least, they’re not talking about it.”
Those dabbling in online price optimization are keeping it close to the vest. At Santa Clara University, home base for think tank Retail Management Institute, associate director Kirthi Kalyanam says he’s aware of at least four pilots and project implementations in which price testing has generated profit gains in the range of 15% to 40% for e-commerce companies. Zawada says he’s seen companies that have “pushed the edge” in testing price levels online raising prices by 5% to 10% without cutting into sales.
Neither would disclose names. Though silent on specific client cases, however, industry consultants are downright gabby when it comes to describing methods merchants could use to optimize pricing online; there are a bunch of them. The keys to success are how the price tests are designed, how the customer base is segmented, and how price change information is communicated to the online consumer.
New opportunities
“What’s being presented to retailers by the Internet is an opportunity to put intelligence and rigor behind a pricing process that for years has been rules-based ,” says Chris Verheuvel, director of strategic consulting services for Manugistics Group Inc., a provider of analytics software and services. Indeed, price optimization in the brick-and-mortar world has depended on the time-consuming manual review of sales and inventory trends and competitive data. A standard practice has been to decide at the season’s outset how long merchandise would remain at full price, and follow up with incremental markdowns at set times.
But the new analytics software is changing that. Look at what happened when store retailer FashionStyle compared its traditional markdown strategy, based on rate of sale versus weeks of supply, with the software-generated recommendations of ProfitLogics’ Pricing4Profit tool. The goal was to clear inventory of a knit shirt while maximizing the retailer’s return on markdown merchandise. To price and time markdowns as the shirt neared the end of the season, the conventional method forecast that at week 10, the weeks of inventory supply would be greater than the number of weeks until the merchandise would be outdated. The recommendation: a price cut at week 10.
Not so fast
But ProfitLogic’s software went beyond the traditional forecasting methods and the manual processing of data. It added the rigor of more data points and a speedier analysis of the relationship between them to suggest a different scenario. The software determined that demand for the shirt was not likely to fall off until week 11, and recommended delaying the markdown until week 12. It also recommended that the price be taken down to a different level. Continuing analysis by ProfitLogic suggested an additional cut in week 15. The traditional approach, meanwhile, suggested a markdown in week 17, and a third one even later when the price cuts still didn’t clear the inventory. The bottom line? Implementing the two strategic price cuts recommended by ProfitLogic achieved total unit sales of 1,711 versus 1,508 using the conventional approach, at gross margins that outstripped it $23,270 to $20,622.
ProfitLogic marketing director Julie Driscoll says that while FashionStyle is a brick-and-mortar retailer, “The mathematics, how merchandise behaves, and how analytics are used to inform pricing hold across channels.”
A similar story from McKinsey’s Zawada supports that. He cites a consumer electronics retailer that used analytics software to dig deeper than traditional forecasting methods into the online sales trend for a new product. The software program judged that the sales trend would support full-price sales for two weeks longer than anticipated earlier. While Zawada isn’t disclosing the retailer or the analytics vendor, the numbers tell a compelling story on their own. The retailer increased profit on the product by 17% with its decision to delay the scheduled markdown, he says.
“Price optimization technology offers companies the ability to establish the perfect price points on an item-by-item basis in order to maximize profits,” says Paul Ritter, an analyst with The Yankee Group, Boston. “The beauty of the Internet as a sales channel is the ability it offers to respond in real time to changing inventory levels, consumer demand, and other factors that influence pricing and buying decisions, and to do it on an almost infinitely scalable basis.”
New rules
In theory, that is, for the opportunity does bear challenges as well. Online shoppers have come to expect dynamic pricing at auction and accept rapid price changes on sale merchandise, but analysts believe they’re less tolerant of prices that change online outside of those situations. Then there’s the problem of channel conflict. Price consistency among channels remains a major issue for multi-channel retailers who now dominate online.
Early on, when Internet retailing was populated mostly by pure-plays, the web was known to consumers as a place to get great deals as merchants gave away the store in a bid to acquire traffic and customers. But with multi-channel merchants now the big players online, web prices are more often anchored in some way to store and catalog prices.
“We haven’t seen a desire on retailers’ part to vary Internet pricing. Instead, they’re treating the Internet almost if it’s another store location,” Verheuvel says. “The intelligence they gather online is being evaluated as a kind of early warning signal. They can use the Internet to understand price sensitivity and whether a new product will move before they put it into stores.”
But that’s web price versus prices in other sales channels. What about a pricing experiment that pits web prices not against those in the merchant’s other channels, but against different web prices—offers on the same merchandise, but at different times and under different conditions—on the merchant’s own site?
This gets close to the exercise that got Amazon into trouble, but retailers have learned a few things since then. Whether merchants that test online pricing in this way wind up with valuable information or just egg on their face is to a large extent a function of how they present the different prices to consumers, experts say.
Companies are uncomfortable for the most part in varying prices on the web, says Zawada, but many are finding ways to do price research while moving merchandise online in ways that minimize risk. “One example is the office supply industry,” he says. “For years, it’s been part of their offline strategy to have zoned pricing—it’s accepted. So some now have zoned pricing on the web. When shoppers sign on, they provide their zip codes. That gives them pricing consistent with store pricing for their region.”
The timing of online price changes also is critical to consumer response. Companies are bypassing random price testing among individual online shoppers to experiment with time as the basis for segmentation. That means shoppers who buy one week could get a different price from shoppers who buy a week later. It’s an approach that’s riskier for some product categories than others.
A rose by any other ...
If a product on a retail site has high price recall among consumers and gets a lot of repeat visits prior to purchase, a wide-screen TV, for example, it’s going to be more difficult to change the price. In such a case, “You want to extend the test for a longer period,” says Zawada. “You might also want to mitigate the risk of a customer having a problem with it by giving him some kind of reward. You don’t have to tell them why you’re doing it, but you’ve given them something in recognition.”
By contrast, specialty retailers of products such as giftware and gadgets face fewer constraints. Less heavily researched week to week, and less frequently purchased, they’re tied to a specific gift giving event anchored in time and little affected by price changes.
In experimenting with online pricing, it’s a rule of thumb that prices that are labeled “promotions” go over better with customers than outright price changes, though interestingly, the revenue impact on a retailer might be the same. “From the customer perspective, a promotion versus a different price tends to be more acceptable,” says Lisa Plaskow, director of revenue and pricing optimization at Manugistics. Merchants also are seeing that it’s less risky to vary pricing by customer segment than by customer. “It’s not very acceptable to change price based on the individual customer. But retailers are finding that it may be more acceptable to offer different promotions to different types of customers. More loyal customers, for example, might get a different promotion than those who don’t order as much,” says Plaskow.
Racing ahead
E-mail is helping retailers reach targeted segments online. “E-mail is still an untapped opportunity,” Bieganski says. “There are other means of discovering consumer price elasticity than outright modifying prices online, and I’d put e-mail in that category. Right now, e-mail offers are viewed as simply promotional events, but the opportunity is there to start varying price with the goal of obtaining price sensitivity information.” Bundling is yet another way to test the take rate on different products at different prices while minimizing the risk of negative reaction from shoppers, by packaging a printer with cables one week, for example, and with cartridges the next.
Though the web offers the potential to change prices almost instantly based on demand data, retailers are still struggling with how to use the data they can now easily collect and analyze. Technology moves faster than consumers’ acceptance of it. Consumers’ expectations are conditioned to a continuum of pricing that may change from minute to minute at online auctions, to prices they expect to remain stable for days or weeks at web stores, and weeks or months on the shelf or in catalogs. It may be that as channels grow closer together — and as technologies such as electronic shelf pricing develop—consumers will become more accepting of more frequent price changes across channels, increasing retailers’ ability to respond to trends and demand in shorter and shorter cycles.
In the meantime, companies that have catalogs or stores continue to move cautiously in implementing price changes in the interests of price consistency among channels. “They may use the data for pricing the next product, or the next season,” says Zawada. “Others less constrained may use the information to change prices immediately.”
But even though the web and software developers bring a new level of precision to pricing, science will never completely replace the art of merchandising, experts say. “We’ve never had this kind of science in pricing before. It’s gotten phenomenal at making decisions based on large amounts of data, but merchandisers still have to put their perspective on it — the mathematical model may say something is the right thing to do, but a merchandiser knows it just won’t fly in the marketplace,” says Verheuvel. “Pricing is starting to be a blend of about 70% art versus 30% science. We’ll probably see that balance out to 50/50 over time.”
mary@verticalwebmedia.com
