Katherine Klinger, director of customer relationship management for artsy online marketplace The Grommet, says 50% of traffic to the retailer’s site stems from email and 2% to 4% of those emails convert. “What that says to me is that more than 95% are gathering information for their next purchase,” Klinger said in a presentation at the Shop.org Digital Summit in Dallas.

To give interested shoppers more data to help them make a purchase decision, The Grommet, No. 447 in the Internet Retailer 2016 Top 500 Guide, creates and sends a second email to shoppers who engage with the first email but did not purchase, Klinger said in the presentation, “The Odd Couple: Rainbow Shops and The Grommet.”

Because The Grommet sells unique items from artists (think a couch coaster that resides on the armrest of your couch so you don’t have to reach three feet to your coffee table for your beverage) the e-retailer often tries to include a maker or artist’s voice within its emails. For example, it might include a Q&A with the person who designs and makes a product. It also often includes product reviews from shoppers in its email.

“There are no words more powerful than those written by your community,” Klinger said. “We will say ‘Hey, the reviews are in and we think you might be interested in what people are saying about this product.’”

The Grommet attracts a loyal customer base, so it also send emails noting new products that will be available to purchase the next day. “We want you to be excited,” Klinger said.

A big part of The Grommet’s email strategy is A/B testing, which it conducts with up to 90% of its email campaigns. Such tests help a wide array of The Grommet’s employees do their job better, Klinger explained. For example, photographers can learn if lifestyle or product shots perform better or if subjects should be looking away from or directly into the camera. Email designers can gain insights on the type and number of buttons they should use for the best results. And copywriters can determine if vague or witty copy resonates with shoppers, Klinger said.

“You get insights from a sample and the winner goes to the rest of the list,” she said.

During her presentation, Klinger displayed A/B tests The Grommet had conducted and asked the audience which emails they thought performed better. In one example, she asked attendees if they thought a daily 10 a.m. or noon email performed better. 10 a.m. won, she said. And she said she is glad the Grommet eventually tested the timing as it had been sending noon emails for over six years. Sending at 10 a.m. instead of noon resulted in a 14% lift in revenue per email, she said. “We had always thought that sending over lunch breaks would be best and people would tell co-workers around that time,” she said.

She also displayed results for an email that tested lifestyle pictures versus product pictures where lifestyle images won, a standard email template design versus a customized template where the standard version won and an email with copy versus and email with just a headline where the headline-only message was the winner. However, with the last example, she noted the importance of looking at all the factors that can contribute to the bottom line. For instance, while the email with just the headline had a higher open rate, the email with additional copy had a far higher conversion rate, making the version with additional text more valuable for The Grommet.