Ryan Allison, President, AWinestore.com, watched his 11-year-old company grow from being a drop-shipping startup to running their own warehouse with thousands of bottles of wine in their inventory. But he decided 15 months ago that they were just scratching the surface with the potential of their Web site.
“We needed to improve merchandising to give our customers more options and increase add-on sales,” he says. “And we wanted to enhance our regular traffic from prospects through higher search engine rankings.”
Allison surveyed the national emarketing services landscape before picking a vendor who happened to be in his Seattle locale (see hotlink below). He and the project manager started discussing merchandising systems and search programs and agreed upon a system that might kill two birds with one stone.
“We decided that recommendations in our product pages and shopping cart process could lift individual customer tickets and enhance search,” Allison says. “I was [anxious] about going into new territory, so we decided that we were going to run a few tests.” CAMPAIGN
Before the testing began, Allison and his team had to program the recommendations page. Here are the four steps they took:
-> Step #1. Merchandising set-up
Allison wasn't interested in simply pitching upsell products based on what others had collectively bought in similar instances. Instead, they went with a real-time behavioral modeling technology.
Onsite keyword searches and recent clickthroughs helped the program distinguish if the customer was male/female, single/not single, at work or shopping at home. They added a reverse IP lookup to determine geographic location, breaking down demographics according to the customer’s region or neighborhood.
“Portland, OR, wine buyers are different than Portland, ME, while Manhattan, NY, customers have different taste and spend different dollar amounts than our patrons in the borough of Queens,” he says.
Two other parameters included time spent on specific product pages during singular visits and whether they came from Google or Yahoo search.
-> Step #2. Test the recommendation system
Once the recommendation system was ready, they wanted to test it against a seriously competitive opponent, so they ran an A/B test in which the recommendations went against their top sellers. They tested for conversion rates and average order size.
“I wanted to make sure that the test was valid and the recommendations page wasn’t going to just beat the heck out of a page that didn’t offer other choices,” Allison says.
-> Step #3. Google vs Yahoo!
With the recommendations page nearing completion, Allison and his team were curious about the sales-conversion potential for natural search via Yahoo! and Google (their biggest SEO traffic drivers). They tested which search engine pushed the most qualified traffic to the recommendations page -- measuring the performance against sales conversions originating on the home page.
“We had always seen variations in purchase behavior from people that arrive from those two search engines,” Allison says. “It’s nothing as simple as Google people like cabernets and Yahoo! searchers buy white wines, but we’ve learned some things about their differences. We wanted to see how the two search sites worked with the new system.”
-> Step #4. Send to a friend & customer ratings
Next up was a shopping cart upgrade that allowed them to add a viral “Send to a friend” feature, along with the ability for customers to review and rate products and sign up for newsletter offers. They also included more images and copy on the checkout page, as well as on the preceding product information pages.
“We had already seen that the labels on the wine bottles sell more than any other factor -- almost like they do in a brick-and-mortar store,” Allison says. “So, we wanted to increase the number of wine labels on the site and needed to have a system that displayed them better.”
First off, Allison became a believer when he saw the results from the A/B test. The recommendations page converted sales at a 23% higher rate than their top sellers. “I was a little surprised at the large margin of difference between the two. At that point, I was more than ready to see it perform full time live.”
Real-life scenarios haven’t let him down either -- their average order sizes have increased by 22% since implementing the system. For the 2006 holiday season, average order sizes were 29.5% larger than the year before, and page views increased 79%. The system was a significant reason why their 2006 sales jumped 80% from 2005.
“Because of the way we have the system set up, we plan on seeing that trend continue without adding more paid promotions,” he says.
And here’s some intriguing data about how Google and Yahoo! organic search converted to sales on the new recommendations page versus the home page:
o The conversion rate for Google users landing on the recommendations page was 54% -- compared to 34% for the home page. Average orders increased 17% when Google users landed on the recommendations page compared to the home page.
o Yahoo!-driven visitors charted similarly, converting to sale 51% of the time at the recommendations page and 31% at the home page. Average order sizes were 16% higher for the recommendations page in the test.
As a result, Allison and his team have been tweaking their search strategies to take advantage of the findings.
In addition, the shopping cart “has allowed us to better utilize product descriptions and the images of the wine labels, and that’s helped our search results as well.”Useful links related to this article
Creative samples from AWinestore.com:
CleverSet - the company that helped Awinestore.com with the product recommendations page:
X-Cart - supplied the shopping cart upgrade: