Cross-sell in Confirmation Emails: 111% Higher Conversion Rate

by Adam T. Sutton, Senior Reporter

CHALLENGE

Stephen Fuller-Rowell, Director, eCommerce, ChinaBerry, and his team operate the health and wellness retail company Isabella. The traditionally catalog-based company has been transitioning to an ecommerce strategy as consumer behavior has shifted.

Following this trend, the team invested last year in adding personalized product recommendations on the site, including product and shopping cart pages. It also added the recommendations to weekly alert emails.

By the fourth quarter of last year, the team realized that recommendations lifted the site's conversion rate to 6.2% from 5.6% for the same quarter of 2008, when it did not have recommendations. Furthermore, the team found that 7.8% of people who clicked through to visit the site from a weekly alert email purchased a product.

"We found that a good indication that things were moving in the right direction," Rowell says.

Rowell and his team were happy with the results. However, they wanted to extend the value of the product recommendations further, and improve the performance of the program as a whole.

CAMPAIGN

Rowell and his team saw their order confirmation emails as the perfect opportunity. Customers already expected to receive the emails. Adding product recommendations could help the team encourage repeat sales, as well as capture more data on which recommendations customers responded to.

Here are the steps the team took to extended its product recommendations:

Step #1. Establish recommendations system

As mentioned, the team had tested and established a product recommendation system on its website and in its weekly alerts email program. The team tailored recommendations individually to users, based on:
o Browsing history
o Purchase history
o Products added to shopping carts
o Products similar to those in which they had shown interest

The system also analyzed which recommendations generated additional sales, and used that information to improve suggestions in the future.

- Ability to manually tweak

The recommendation system was not entirely automatic. Team members could identify products that should never be suggested in combination with a non-complementary product, and they can also insist on certain products being paired with others.

"It's not entirely machine-driven," Rowell says, "so it is not losing the expertise of our merchandisers"

Step #2. Grab low-hanging fruit first

After establishing the system, the team added recommendations to key areas of its website, such as on product and shopping cart pages (see creative samples below). These areas, along with the team's email alerts, are well-known high-impact areas to add product recommendations on retail websites.

If your team is just getting started with product recommendations, you should pursue a similar strategy. Start with the locations mentioned above. Then monitor performance and tweak the display to improve conversions further.

After you've exhausted these well-known areas, you can move on and test new areas, as Rowell's team has.

Step #3. Redesign confirmation email

Once the team decided to push ahead with adding recommendations to its order confirmation emails, the next step was to upgrade the text-based messages to an HTML format.

For years the team had used text-only emails to help ensure customers could easily read messages, regardless of email provider. However, the team could not add images of the products to text-based emails.

The team designed a new confirmation email with the following features (see creative samples):
o Company logo and tagline in header
o Order confirmation and date in header
o Thank you message
o Contact information
o Text-based list of products purchased
o Order total and shipping information
o Three product recommendations

Product recommendations were included in the right third of the email, in a separate box. Three product images and their titles, which are hyperlinked to their product pages on Isabella's site, are included.

The suggestions were based on the same data and algorithms as those on the Isabella website. To serve suggestions, the team's system identified who opened the email and immediately sent the three suggestions from its server.

"It acts pretty much like a webpage in that way," Rowell says.
RESULTS

"The impact of the personalized recommendations on sales is comparable to the impact that Google AdWords has for us, which for us is the 800-pound gorilla in our marketing," Rowell says. "[The recommendations have] probably had the largest impact on our Web sales in the last 12 months."

In July, the total revenue generated from the team's recommendation engine equaled 90% of the total revenue the team generated through Google Adwords.

The team's recommendations in confirmation emails, its latest addition to the program, are boosting sales. An analysis from late July to late August revealed:

o 400-500 customers per month click the recommendations in confirmation emails

o 19% of those who do click a recommendation complete another purchase

o Recommendations in transaction emails generate a 111% higher conversion rate among those who click a recommendation than the team's weekly alert emails

- Improving the program's performance

Adding the recommendations to confirmation emails also gives the team additional data on which recommendations work best, Rowell says, and which do not spark interest. That data is used to further refine their system.

Useful links related to this article

CREATIVE SAMPLES
1. Confirmation email with recommendations
2. Product page with recommendations
3. Shopping cart with recommendations

Members Library: Revamped Recommendations Lift Order Value 15%: 5 Steps to More Relevant Suggestions

MyBuys: Supplies the team's product recommendation service

Isabella

Chinaberry, Inc.