Ex Machina: Jack Graham's Blog

I went to Cheetahmail’s “Relevance Tour” client summit the Tuesday before last. This post is a summary of what I saw and heard.

This was the first year they’d done the event in Boston, but it left me wishing I’d attended previous years’ events in NYC. The companies present covered a broad swath, including ad firms, retailers, and one other travel company. I think I was the only person there who was the sole representative of their company.

Most of the presentation revolved around best practices and was given by two people from Cheetahmail’s client services group. At the end, they held a panel with Q&A that included their creative director and a few people involved in strategic services. The presentation was broken down into several broad topic areas, including:

• Acquisition (including eAppends)
• Retention
• Testing
• Segmentation
• Reactivation
• Miscellaneous Topics
  

Acquisition
Here are some techniques that Cheetahmail’s clients are using:

• Acquisition pop-ups for new site visitors. J. Crew and a few other retailers are using an acquisition pop-up to solicit e-mails from all new site visitors. “New” in this case is defined as un-cookied. Once someone visits, they’re cookied and will no longer see the pop-up, whether they choose to fill out the form or not. In terms of the mechanics, one should do this with a CSS overlay rather than an actual pop-up to avoid pop-up blockers. This is potentially a very powerful lead gathering technique. However, I would be somewhat concerned from a user experience standpoint that it would confuse or annoy past site visitors whose cookies expire or who access the site from a different browser.
• As an alternative to pop-ups, some sites use a customized ad banner with lead acquisition call to action for new visitors. (Returning visitors instead see a promo or other information).
• Pop-unders for abandoning users. Some sites attempt to create a pop-up underneath the user’s current browser window if it looks like they’ve abandoned a form. The pop-under is then present when the user closes the front browser window, providing another chance to grab their eyes.
• Automatic opt-in. Automatically opting in anyone who buys from us is CAN-SPAM compliant, provided they haven’t previously opted out. Whether it’s a good idea to do this without getting permission is another policy; it depends on how aggressive your organization is.
• Required fields. Make e-mail fields required in all forms; customers now expect this.
  

eAppends
Although eAppends are CAN-SPAM compliant, data suggests that users don’t like being appended. Big surprise. For most of Cheetahmail’s clients, names added to their list via eAppends performed only 50% as well as organic leads. Cheetahmail recommends doing eAppends only once every six months, or quarterly if volume is high enough, but eAppending more frequently doesn’t appear to affect performance. Most of the discussion on best practices for eAppends had to do with keeping the new user from immediately opting out or hitting their spam button. They recommend doing a series of welcome messages rather than immediately hitting eAppends with promotional mailings. A typical campaign might consist of three or even four mailings: a welcome message (also serving as an opportunity for users to opt-out), an informational mailing about the company, a survey or qualification e-mail, and then maybe an attractive offer.

For what it’s worth, I’ve never thought eAppends were a good idea, and a lot of people who blog about e-mail marketing would agree with me. True opt-in is the way to go, in my opinion.

Retention

• There was a discussion of e-mail preference centers. We were shown a very creative example from Barneys, and while Barneys’ cheeky ad copy wouldn’t fly well with my company’s septuagenarian demographic, the structure of the page was something any business could duplicate. A good preference center should have the following elements, and in most cases they should appear in this order:
  
  1. Frequency controls. Promising users to mail them less is the first line of defense against them unsubscribing outright. This has to be implemented in such a way that users actually do get noticeably less mail, or they might just hit their spam button next time.
  2. Unsubscribe button.
  3. Change of e-mail address.
  4. Receive all/increase frequency. Barneys’ page calls this the “Please Stalk Me” button. Barneys has implemented this so that users who choose this option receive all of their promotional e-mails.
     
• Unsubscribe surveys have been used successfully by some companies to gather data on why people leave their lists. One jewelry retailer found that about 15% of people who unsubscribed responded.
  

Testing
The best practice here is to realize that different things will work in different situations. Therefore, test frequently, and test a variety of different properties of the mailing. Some things to try include:

• Subject line testing. This is a very common test. The most common metric to watch is open rate, but some clients found that subject lines affected clickthroughs, too. Test results can be surprising in some cases. Some examples:
  
  • General subject lines versus sweepstakes. For survey & qualification e-mails, some clients found that a more general subject line worked better than one that called out a sweepstakes or prizes.
  • Recognizable names versus general branding. Companies who had athletic endorsements or other recognizable names sometimes found that using those names in subject lines worked well, although this wasn’t the case 100% of the time.
    
  • Length (short versus long). One company found that subject lines of 25 characters or less performed best, subject lines of 50 characters or more were second best, and mid-length subject lines (25-50 characters) were worst. Cheetahmail speculated that this might be because the shortest subject lines leave a noticeable white space in the recipient’s inbox, while long subjects lines might generate curiosity because they get cut off at the end.
    
• Time of day testing. Strategy people on their panel suggested re-testing time of day and day of week every six months, as the optimal times can change, as well as retesting seasonally.
• Hosted link versus no hosted link. Some companies found that putting a link to a hosted version of the e-mail drew a lot of clicks.
• Mobile-friendly hosted links. In some cases links to mobile-friendly versions of e-mails performed very well.
  
• Creative Testing. Can be based around most-clicked links (i.e., where calls to action or other links are located in layout). Should be done across multiple mailings for a valid sample.
• Remail campaigns. Test lift gained from remailing a campaign within a few days of the initial mailing.
  

Segmentation

• Polling. This is a novel strategy some companies have tried in place of traditional qualification surveys. Rather than trying to get people to fill out an entire survey, set up a poll with results visible to users (e.g., “Where is your favorite destination to travel? … A) Holland & Belgium, B) Southern Africa, etc.). Then record responses to user records. This demands much less of the user than filling out a whole survey but can still yield useful data.
  
• Geographic Targeting. Some companies have tested segmenting by users’ geographical locations. This works particularly well for retailers selling things like sporting goods that might be useful in one part of the country but not another.
  
• Browser Data. With site-wide login and cookieing, it becomes possible to track what a user is looking at on the web site and then use this to segment. Cheetahmail clients who took this approach of tying segmentation to web site usage saw very good performance — as much as a 50% lift.
  

Lifecycle Triggers
Many of the practices here tied in to how users interact with company web sites.

• Remarketing to shopping cart or form abandoners. If someone gets part way through filling out a form or ordering online and then abandons, a campaign is triggered. Staples did an automated cart abandonment program that would send users who browsed a product and then abandoned before completing the order a promo for that product and three similar products.
• Web site reactivation. Campaigns can attempt to target web site visitors who looked around a lot but then don’t come back for 30-1000 days. For the really long-lapsed visitors, some companies tried unusual or creative subject lines to grab attention, figuring they had nothing to lose. (One company send mails with a subject line about marauding aliens attacking the Earth and apparently saw some good results!).
  

Miscellaneous Topics

• Some companies actively solicit reviews and/or testimonials from customers.
• Video. Messages can contain a static video frame that opens a video, or an anigif taken from frames of an actual video. (Holland America uses a product called LiveClicker to do this).
• Go green strategies can convert more offline users from paper mail to e-mail/web.
  

Thought Leaders

Based on what I saw, watching how the following companies handle certain aspects of e-mail marketing might reveal some good ideas:

• Bass Pro Shops & Giant Eagle — for eAppend welcomes (although this would be very difficult to monitor from outside).
• Drugstore.com — for early lifecycle and synchronization of e-mail content with the company web site.
• Barneys — for preference centers.
• Puma & Urban Outfitters — for lead acquisition techniques. I also love Urban Outfitters’ welcome campaign.