CHALLENGE
“Kids can be moving targets because their wants change regularly,” says Anand Shah, COO, Karmaloop. “Many times, they are looking at three or four separate pieces of content simultaneously when they open your email. All in all, we were looking to do a better job of offering products of importance in order to make a difference.”

With teen shoppers being Karmaloop’s primary demographic, Shah knew he must serve content that would push his audience’s relevancy buttons. By last fall, their house list had grown to 300,000, and Shah didn’t think their email efforts maximized their house list’s potential.

He and his team were already making offers based on past purchases, so they hoped to augment their email program with an automated system that gives customers control over what merchandise they wanted to hear about. Shah was concerned that teens would sign up for too many types of products and water down his goal of more-targeted email.


CAMPAIGN
First, Shah had his team survey a large portion of their customer base to determine how many brands to include in the new program. The results confirmed what he knew from their sales sheet -- that they had a handful of wildly popular brands, such as Puma, adidas, Nike, Paul Frank and Obey. But he was somewhat surprised to see the interest expressed in dozens of lesser-known brands.

From this, they developed a system that allowed customers to preselect by brand or by clothing category what types of merchandise they wanted to get emails about. Here are the five steps they followed:

-> Step #1. Pinpoint categories & brands

Once they pored through the survey data, they incorporated an alphabetical list of nearly 100 brands to choose from on the checklist for both boys and girls (see link to sample image below), deciding that as long as the list was relevant to the audience, such an extensive length was OK. If the list started to underperform, they could easily shorten it.

“We knew going in that a considerable portion of our customers were brand fanatics,” Shah says. “A different kind of consumer might go shopping for sneakers, but our customers come looking for certain styles of adidas or Nike.”

They also incorporated their standard 24 product categories into the checklist. The system was designed to recognize IP addresses, so that when a customer came back to modify his or her list, the categories and brands they had already selected would be checkmarked to remind them of what they were signed up for.

-> Step #2. Signup buttons on landing pages

To keep their regular email list from being cannibalized by the Alerts system, Shah had the team place prominent signup buttons on the landing pages for their brand keyword buys -- which were producing 80% of their regular search-driven traffic. While they knew existing customers would sign up for Alerts because of the buttons they added across the site, a key goal was to capture more SEM-generated leads. They also placed the signup buttons on brand and category pages across the site.

“We didn’t just want our past customers to be signing up. We wanted prospects to be a big part of the mix.”

-> Step #3. Behavioral targeting

Once this was finished, they programmed the system to take into account behavior information, such as abandoned shopping cart products and recent repeatedly viewed items, and combined them with both the pre-selects and past purchases. In terms of purchases, similar products or cross-sell items were part of the Alerts’ equation. Shah and his team held weekly meetings to discuss the data and whether or not to tweak the algorithms.

-> Step #4. Message timing

How many emails customers and prospects received was dependent on current sales and new products being driven with the brands they had selected. One week, they might have received two Alerts but none the next week. Shah didn’t want to bombard teens with constant emails, so he set up the program to limit it to no more than two messages a week about sales/new arrivals related to their selected brands.

-> Step #5. Target non-responders four months later with new offer

When recommendations didn’t convert after four months but customers hadn’t opted out of the Alerts email, the system rotated merchandise to the next group of items on the team’s set behavioral scale. (Items abandoned in a shopping cart get one of the top weights, while products viewed three times are weighted higher than ones viewed once.)

For example, if a female customer hadn’t made a purchase, the Luxirie by LRG jeans offers she was getting were replaced by Soundgirl jeans or another jeans line she had repeatedly clicked on.


RESULTS
Shah definitely achieved the relevancy he was looking for. In fact, he’s borderline ecstatic with the bottom-line numbers: email-generated revenue skyrocketed 318%, while conversions-to-sale increased from 3% to 4.6%. The Alerts feature also has helped overall onsite sales, which have been growing incrementally by 3%.

Nor does it look like the big checklist was too overwhelming for young people to wade through, so Shah has no plans to shorten (or lengthen) it. The average number of preselects chosen so far has been 6.4 per signup user. “We are seeing sales climb and more customer-retention because of Alerts. It just proves that there are two kinds of people out there: the ones who like to hear about everything you offer and those who just want to hear about Puma.”

Hooking up brand SEM keywords with the Alerts feature also proved to be a wise idea, attracting prospects the way Shah envisioned. More than 40% of people on the list are new to the brand, and the file size is increasing at least 20% monthly. “The regular list is only growing at the same pace it was last year, while the Alerts list is growing phenomenally.”

Meanwhile, 12% of their entire house list has signed up for the Alerts program. The behavioral targeting part of the mix was also a key driver, Shah says, especially when it comes to bringing lagging customers back into the fold months later. “While we wanted our customers to feel empowered by the preselects -- and that’s what we are seeing in the results -- the [behavioral] part allows us to make sure the offers are relevant in an up-to-date way.”

The timing of the messaging has been a boon to their performance as well. They’re seeing a clickthrough rate of 5.8%, compared to 4.5% that they would normally get from all other email programs not using this system. Using the same comparison, opens increased from their regular 17% average to 21.3%.


Useful links related to this article

Creative samples from Karmaloop:
http://www.marketingsherpa.com/cs/karmaloop/study.html

MyBuys - vendor who helped create the Alerts:
http://www.mybuys.com

Karmaloop:
http://www.karmaloop.com