What a Bricks-and-Mortar Business Can Learn From Website Analytics
By Maura Ewing
If you don't yet have a well-designed website for your bricks-and-mortar business, now there's even more reason to get to it: Your site's traffic could help you predict the pace of offline sales in stores.
A recent study from the Kellogg School of Management at Northwestern University, analyzed clickstream data for a $10 million U.S.-based manufacturer that sells customized industrial products. The findings: The more often a customer visited the site and the longer she lingered on the page correlated with the likelihood that she would place an order offline.
Time is money
The time that elapses between a typical click on a website and when a person places an order is called "lead time." "The greater the lead time, the more operational value a company gets from the click information," says Jan Van Mieghem, a professor of managerial economics and co-author of the study.
E-commerce sites such as eBay or Amazon generally have a very short lead times. "This type of company can do very little updating of production and inventory planning with clickstream information because it has no ability to react," Van Mieghem says.
That's not the case for retailers who display their products online, but take orders offline. Customers generally go their websites while researching products, and often wait days, weeks, or even months before placing an order. "The longer the lead time, the greater the opportunity for a business to improve production and inventory plans using the method we created," Van Mieghem says. If the lead time is too long, however, it becomes too noisy to forecast.
Not all clicks are created equal
The researchers produced a regression equation that quantifies the likelihood that a particular click will lead to a purchase based on an in-depth analysis of the company website's clickstream traffic.
The longer a visitor stays on the site, the more likely he or she is to purchase -- but only up to a certain point. "Initially as the length of time goes up the probability of purchase goes up, until you hit a certain trigger point," Van Mieghem says. After that trigger point, the click apparently becomes less informative. "I think these are cases where the page is up, but nobody is looking at it," he says. "Unfortunately it's not possible from our data set to track the amount of eye contact a person is devoting to the page."
A reliable crystal ball?
With the data they collected and analyzed, the researchers found that clickstream data predicted, with statistical significance, the propensity, quantity, and timing of offline orders. Some visitors were easier to accurately analyze than others, however. For example, repeat customers were much more forward-thinking with their purchases, and thus easier to accurately analyze. But with a sales forecast based on clickstream data, the manufacturer was able to strategically plan its inventory, and reduce holding and backordering costs by about 5%.
The study is the first of its kind for offline sales channels, and there were limitations to the research. First, because the company sells to other businesses, it was able to identify those visitors by IP address -- something a consumer-facing small business wouldn't be able to do unless it required visitors to log in.
Also, this approach is most useful for businesses selling products that require testing, customizing, or price negotiations because of the amount of time clients spend researching before purchasing.
Such accurate forecasting using this method probably isn't possible for most retailers, but that's not to say you can't glean useful data from clickstream traffic. "Any retailer should be tracking their website data. Just looking at your top visitors, and what pages they look at can be revealing," Van Mieghem says. "Patterns in click length and frequency are still good sales predictors."