Close
Join 237,000 weekly readers and receive practical marketing advice for FREE.
MarketingSherpa's Case Studies, New Research Data, How-tos, Interviews and Articles

Enter your email below to join thousands of marketers and get FREE weekly newsletters with practical Case Studies, research and training, as well as MarketingSherpa updates and promotions.

 

Please refer to our Privacy Policy and About Us page for contact details.

No thanks, take me to MarketingSherpa

First Name:
Last Name:
Email:
Text HTML
Join Our Research Team at DMA 2014
Apr 02, 2009
Case Study

How to Get 70% Conversions: American Greetings’ 6-Step Campaign

SUMMARY: One size doesn’t fit all when creating email offers for subscribers. But segmenting lists can prove challenging for subscription-based sites with lots of content and no marketing database.

Follow one marketer’s 6 steps to segment an email list without a marketing database. The American Greetings team got a 70% lift in conversions from a simple email test that matched subscribers’ preferences to content in the offer.
CHALLENGE

American Greetings had two kinds of customers – members and nonmembers – when Kathy Hecht joined the company a year ago. She knew that two segments weren't enough to help her team start talking to email subscribers in more relevant ways.

“There were some communications that were different,” says Hecht, SVP and General Manager, AG.com. “If you were a member, we didn’t try to get you to join. We reserved that for the non-members. But outside of that … it was a one-size-fits-all approach.”

Hecht and her team weren’t deterred by the lack of a marketing database. Find out how they succeeded in segmenting email subscribers to boost conversions.

CAMPAIGN

Step #1. Master the basics

Hecht made sure all measurement and uniformity aspects of the site and email system were in order. Here’s her short checklist of questions to ask before implementing a segmentation strategy:

-Can you track email open rates, CTRs, click-to-conversion rates?
-Can you track how people click through the website? Where they leave?
-Do the website’s “join” or “subscribe” buttons take people to different offers?

It’s important to track clicks and email metrics, as well as make sure all “join” and “subscribe” buttons take visitors to the same offer. This ensures the testing environment is solid. It also helps get everyone on the same page in terms of how and what to measure.

Step #2. Start with known data

The team didn’t have a marketing database. They had to look at order histories, housed in a data warehouse, to get a sense of how to segment subscribers.

It was almost like looking at fulfillment data, Hecht says. The team looked through records of every e-card sent by subscribers. They asked themselves: How can we talk to these people differently?

Step #3. Build attributes

Hecht’s team tagged attributes of the e-cards sent based on tones, such as funny, traditional and heartfelt, and recipients, such as sister, mom, and brother.

“We didn’t have a marketing database, so a lot of our analysis, a lot of our segmentation has had to be done in a very long and laborious way,” says Hecht, noting that the team didn’t let that stop them. “With sheer brute force we got through it in a year.”

Step #4. Define segments

Hecht listened to everyone’s opinions about how to segment subscribers. She also looked at the data for clues.

-Get ideas from customer service

Hecht found that customer-service representatives knew customers’ wants and needs based on questions customers asked. The reps gave great insight on how to segment customers.

-Look for patterns in the data

The team looked at patterns in the data, such as the number of people sending humorous e-cards. If there were a lot of them, that was an easy way to segment that group and start talking to them in a relevant way.

They also looked at anomalies in the data, such as people who only sent birthday e-cards. “I told my team, we’re just looking for pitchforks – [not needles in the haystack] – because there are some very easy ways for us to think about the customers,” Hecht says. “If you only send cards to your girlfriends, those are the kinds of things we should put in front of you.”

-Don’t count on averages

Hecht stayed away from calculations, such as the average number of e-cards sent, because they didn’t help her determine how to talk to subscribers differently. "Even knowing that you sent 20 cards doesn’t help me out,” she says. “It becomes more meaningful if I know … you sent 20 cards to the same person or to 20 different people.”

Step #5. Develop a segmentation strategy

Hecht’s team segmented subscribers based on:

Example #1. Lifecycle of subscribers

They figured out whether subscribers were new acquisitions, re-subscribing members from the past, or those whose subscriptions were up for renewal.

Example #2. Tone of the e-card

They determined the tone of e-cards subscribers sent to friends and family.

Example #3. Kind of sender

They classified subscribers into: “happy senders” (i.e., those who sent many e-cards to numerous people) or “obligatory senders” (i.e., those who sent a few e-cards to a small number of people on major occasions, such as birthdays or Mother’s Day).

Step #6. Test emails to segments

Hecht’s team got great results from tailoring email offers to the tone of e-cards subscribers had sent.

-Match tone to preference

They matched the tone of e-card offer in the email to the tone of the e-cards subscribers frequently sent to friends and relatives.

Sherpa Tip: Traditional online publishers could implement this kind of testing strategy as well.
Segment subscribers based on the kinds of articles they frequently read, for example. Then, match the tone of articles offered in an email test to the tone of content that subscribers most frequently viewed to see if you get higher open rates.

-Send two email versions

AG sent out two versions of the email – one promoting a traditional St. Patrick’s Day e-card, the other promoting a humorous St. Patrick’s Day e-card.

Naturally, traditional e-cards went to subscribers who’d sent lots of traditional e-cards. Humorous e-cards went to subscribers who’d sent lots of humorous e-cards.

-Conduct multiple tests

The team conducted several tests to several segments to determine which segments got the best results (i.e., lift in conversions and number of e-cards sent).

“It’s not impossible, which is important for people to understand,” Hecht says. “We didn’t have all the technology. We didn’t have all of the wiz-bang tools in place.”

It took weeks at the beginning to pull data, evaluate and send test emails because some of the data choices were very complicated, she says. But now it takes about 10 days.


RESULTS


AG saw a 70% lift in conversions from the St. Patrick’s Day e-card email test.

Conversions, in this case, were people who went to the site after seeing the email and purchased the e-card or a $15.99 annual subscription.

Open rates were relatively flat, says Hecht, because it was challenging to tailor subject lines to the offer. “But after people open it, we’re getting more of them to come to the site and then take us up on that offer because they’re seeing an offer that’s incredibly relevant to them,” she says.

Another metric AG used to measure success was the number of e-cards subscribers were sending. Were they sending more e-cards than they had in the past? “Within any of our test segments, the answer to that is yes,” Hecht says. The metric mattered because the more e-cards subscribers sent, the easier it was to prove the value of the subscription in renewal offers.

Useful links related to this article

Creative Samples from American Greetings:
http://www.marketingsherpa.com/cs/americangreet/study.h
ml


American Greetings:
http://www.americangreetings.com/


Post a Comment

Note: Comments are lightly moderated. We post all comments without editing as long as they
(a) relate to the topic at hand,
(b) do not contain offensive content, and
(c) are not overt sales pitches for your company's own products/services.










To help us prevent spam, please type the numbers
(including dashes) you see in the image below.*

Invalid entry - please re-enter




*Please Note: Your comment will not appear immediately --
article comments are approved by a moderator.