OneStopPlus Group, formerly part of Redcats USA and which operates e-commerce sites for apparel and home furnishings brands including Woman Within, Brylane Home, KingSize, and fullbeauty, ensures that when it changes anything on one of its web sites—be it a hero image, product description or the shopping cart—the new version will lead to more clicks and conversions than under the old version, according to Mathieu Clavie, director of e-commerce. The retailer uses technology from vendor SiteSpect Inc. to run A/B and multivariate tests before every web site change it makes, he says.

While SiteSpect’s technology has helped OSP identify new online shopping features and functionality that have led to increased sales, it also has saved the retailer from investing too much time and resources in developing things unlikely to produce good results, Clavie says. Before devoting lots of work to a large task such as revamping the checkout flow from multiple pages of steps to one page, for example, OSP’s developers first validate that the change will provide the intended results.

“It’s really from a technology standpoint that we save a lot of money,” Clavie says. The savings usually come from the time its in-house developers spend on a project, he adds.

So far this year, OSP has completed eight individual tests with the web-based tool, Clavie says. For one, on plus-size women’s apparel site JessicaLondon.com, it used SiteSpect to determine that displaying “outfit shopping” on the home page—a banner image that shows shoppers what to wear for the current season—led to a 5% increase in the number of items sold on the site and a 2% increase in revenue over two weeks compared with displaying “previously viewed items” in the same area, he says. The retailer also is conducting another test to see whether shortening the checkout process from five or six pages to a single page boosts conversions on all its web sites. With the single-page checkout, conversions over several weeks have increased by three percentage points on average across the sites compared to conversions with the multi-page checkouts in that time, he says.

One of SiteSpect’s main advantages, Clavie says, is that it does not use code embedded in the web sites to run its tests, which means using it doesn’t slow down site performance. Instead, the technology works at the server level, much like a content delivery network such as Akamai Technologies Inc., he says. When a consumer goes to the web site, her browser calls the content delivery server to load the display and, at the server, SiteSpect runs a program that tells her browser which version of the site to load, based on the test OSP is running.

OSP also can create different static test pages and measure how shoppers react to them, Clavie says. This helps the retailer understand how a change may affect shopping activity before it’s fully developed. For example, it recently began an A/B test where one test page showed all the available sizes and colors of a T-shirt. The other page showed a single shirt as usual. OSP is running the test to see if displaying all available options spurs consumers to buy more. If the test shows that consumers will buy more, OSP will add further options to the page, such as whether the shirt also comes in long sleeves.

“We’re testing whether we keep going with that project,” Clavie says. He says that if test results show consumers don’t buy more, using the static pages rather than a more developed test will have saved OSP about 200 hours of development time.

In this way, SiteSpect pays for itself, he says. He declines to say how much OSP pays for its monthly subscription to the Internet-hosted tool. SiteSpect says its pricing starts at $4,500 per month and goes up according to the size and needs of its clients. The technology is also available as a licensed, on-premise tool, the company says.

Favorite