Plan For Reputation Impact in E-mail Segmentation

Aug 27, 2007 9:52 PM  By

If you use e-mail in your business, you know inbox delivery is all about reputation these days. A close inspection of reputation factors could reveal ways to improve inbox delivery, thus improving ROI.

There are two parts to assessing an individual message’s deliverability. Every major ISP maintains a reputation profile for every sending IP address. That reputation is based on two key factors: complaint rates and spam trap hits. There are other factors, such as mail volume, mail frequency, and unknown users, but marketers usually pass these tests easily through regular contacts and good data collection practices.

In addition, each individual message has 30 or more analyses applied to it to determine whether the message is spam. These include many of the same factors as the IP reputation monitor, plus content filters, real-time complaint rates, throughput rate (i.e. messages per hour – spammers are usually very high), bounce rates (high hard-bounce rates are a clear sign of a spammer), and much more.

With this background, e-mail marketers can make some changes that will lead to increased inbox delivery among the segments most likely to respond. Because each individual message is rated on its own merits, you can separate messages that have different likelihoods of triggering a spam filter or block.

  1. Separate first time sends – if this is the first time you are mailing to an address, send that message separately. The likelihood of a complaint is far higher to a first time recipient. If you don’t confirm or double opt-in, you will confine hard bounces to this group as well.
  2. Separate 90-day inactives – if a customer has not opened or clicked on a message in 90 days, sent that message separately. This is your most important segmentation, since some ISPs (unfairly) use abandoned accounts as spam traps or donate them to service providers like Symantec to include in its Brightmail Probe Network, which monitors millions of addresses for spam. (On paper, if Brightmail sees a message show up in several of its Probe Network accounts, the message is spam, since the accounts were never opted in to receive the message. In practice, the account may have been active at some point and opted-in, so you get penalized no matter how good your collection practices are.) If your team has the time, you could further separate your list into 90-180 day inactive and 181+ day inactive, and see how the two segments perform.
  3. Separate purchasers from non-purchasers – if you have been able to connect a transaction to the e-mail address, the likelihood of it representing a legitimate customer is substantially higher than an address without a purchase.

The end result is five basic segments of recipients – first time, inactive purchasers, inactive non-purchasers, active purchasers, and active non-purchasers. Even if you don’t version your content, separate your customers into these groups when you send your message. That way, the most important group, the active purchasers, won’t have their message ID contaminated by higher complaint rates or spam traps.

Of course, if you don’t use a deliverability reporting service like Pivotal Veracity, Habeas, or Return Path, you won’t learn much about the differences between these segments. But at a minimum you should see substantial e-mail performance differences (opens, clicks, conversion, etc.) between the groups.

Michael Greenberg is vice president of marketing for Loyalty Lab, a San Francisco-based developer of customer loyalty programs for the retail industry.