T-Mobile is one of first advertisers to run a 1-minute video ad.
E-retailers that know what they're looking forget the most out of the torrent of analytics data.
When one of Brandon Proctor's roughly 65 marketing and analytics staffers says she'd like to pull a custom analytics report or gather a specific set of customer information, the first question Proctor asks is "Why?"
"Many companies collect data, but very few actually put it to use," says Proctor, vice president of marketing at home improvement e-retailer Build.com. "Today you can pull information on just about anything. But you have to make sure the data is usable."
Making the distinction between profitable and pointless data is especially important during the crucial holiday season, where time is short and stakes are high. Analytics providers such as IBM Corp.'s Coremetrics and the Omniture business unit of Adobe Systems Inc. can be sophisticated to a fault, offering so much data that companies suffer from information overload. But a measured approach to analytics can help retailers prepare, boost sales and avoid disaster during the holidays. Experts offer a few tips to help retailers make the most of their analytics data during the holidays. Among them: Test, create goals, examine holiday shopping trends from last year and look at consumers' buying patterns during this year.
Deciding on discounts
Proctor conducts tests to make sure his company is not giving away bigger discounts than it needs to capture a sale. Build.com tests 1%, 2% and all the way up to 10% discounts to find what Proctor calls a discount tipping point. It has found on some items a 6% discount generated just as many sales as 10% off. And for one product, he found that a small 2% discount did the trick as well as larger price cuts. "So many companies offer a 10% discount by default," Proctor says. "In some categories, we don't have much margin to play with. A few percentage points can make a big difference."
Retailers may be surprised with test results. While Build.com discovered in some cases that shoppers were happy with 2% off, consumers might be more stubborn when it comes to another type of offer, like free shipping. E-mail services provider Experian CheetahMail found during the holidays last year that free shipping offers with no minimum spending requirement had 70% higher conversion rates, triple the revenue per e-mail and more than double the average order value compared to free shipping offers with minimum purchase requirements.
Danielle Savin, vice president of multichannel marketing and retail at e-commerce consultancy FitForCommerce, recommends keeping tests simple. She says it's easy for retailers to over-complicate them. For example, if Build.com is testing a discount percentage in an e-mail, it should keep every other element, such as text and graphics the same, Savin says. If a retailer tests a 10% discount e-mail with a Christmas tree graphic and catchy subject line against a 5% discount e-mail with no graphics and a brief headline, that's too many variables. A retailer won't know what triggered the shopper to buy, Savin says.
"Be clear in your A/B testing and then look at the statistics and go with it," she says. "Don't try and test a promotion and text and creative at the same time."
Start with a goal
David Zucker, vice president of marketing analytics at members-only flash-sale site Gilt.com, looks at analytics with a specific goal in mind. Rather than trying various e-mail offers or discounts to see what garners the most sales, Gilt, which uses analytics vendor Mu Sigma, decided it wanted to get women to buy more men's products and then used analytics to help it reach that goal.
After looking at analytics Gilt found many women who bought men's items were buying gift items, such as cufflinks, rather than items that required a size, such as shoes or jeans. Gilt.com was able to identify female members most likely to shop in the men's category, and then market to them with specific messages and promotions. As a result, the top 30% of women deemed most likely to buy in the men's category bought 70% more men's products.
"It can even work on a higher level," Zucker says. "If we have a goal of getting more members, we then might look for the products members saw right before they activated their membership and try to find a trend. Companies get weighed down in the sexiness of all the detail they can get, without thinking first about what they want."
Looking at last year
So when should retailers start testing and setting goals to for the holidays? Early, Savin says. "In February or March, after returns come in, every retailer should be pulling together marketing and creative staff to complete a post mortem," she says. Retailers should look at what worked the best and the trends for the season just past, she adds.
For example, Savin suggests merchants look at order flow from last year to help determine warehouse staffing needs. Look at peak sales days last year and increase staff during those days this year, she says. She also recommends analyzing web site traffic to find the peak traffic days from the previous year, then conducting load tests to make sure the site can handle at least that many visitors at once.
But retailers can't expect holiday shoppers to be carbon copies of the year before. Merchants should also account for this year's trends, Proctor says.
More bang for the buck
For example, Proctor, who uses an Adobe Omniture analytics package, is noticing that average units per order this year are higher than last, while order value is about the same. That means a shopper who in 2009 may have spent $2,000 for a kitchen faucet might still spend $2,000, but will buy a kitchen faucet, a ceiling fan and a light fixture. More items per order could make offering free shipping over the holidays much more expensive for Build.com, especially because his company ships from several distribution centers. Shipping three items from three centers is much more costly than shipping one product from a single hub, he says. As retailers uncover trends, they can shift around promotions to save them moneyÑfor example, by offering a discount or a mail-in rebate instead of free shipping.