Your e-mail list has a large number of older addresses. How can you minimize list atrophy and restore your list to its former glory? Reconfirmation can help separate the wheat from the chaff on your list. The following presumes all the recipients on your list opted in at some point. Any addresses that didn’t should never have been on the list and should be removed straightaway.

Approaches to Reconfirmation
To reconfirm addresses, send a message to recipients informing them they’re on your list. They can either remain on the list or unsubscribe. Some mailing list software can be configured to do this automatically periodically, but most can’t. Reconfirmation is an excellent tool for those list segments that aren’t at either end of the spectrum. They’re not so old they should be entirely discarded, and they’re not so new they can be used without concern.

There are two approaches to reconfirmation: opt-in and opt-out. The opt-in approach requires recipients to take a specific action to remain on the list (usually clicking on a URL). Opt-out requires them to take action to be removed from the list (usually clicking on a link or sending a removal request).

The opt-out approach is generally more attractive to marketers, as it loses fewer addresses. However, it doesn’t clean the list as effectively. Recipients of messages that are black-holed or quarantined aren’t removed by opt-out. Further, users have long been advised not to unsubscribe from spam. Some recipients will complain or take no action rather than opt-out. A few believe since they hadn’t opted in before they shouldn’t need to opt out now.

Selecting the Recipients
Though many factors may be considered and precise determination depends on circumstances, there are two key dimensions to think about:

• Recency. How recently was the segment contacted by e-mail? Within the last six months? Within the last year? Never?
• Provenance. How good is the record and evidence of their opt-in? Full source, date, and IP address? A flag indicating the source of the subscription? No record?

Writing the Message
Craft a clear, concise reconfirmation message. You can provide an example of what the recipient will receive, but avoid making the message look like a marketing solicitation. This will only confuse recipients about the message’s purpose. The message should:

• Inform recipients why they’re receiving the message
• Remind them when, where, and how they originally signed up
• Tell them what they can expect to receive from you in the future
• Tell them what they should do to continue or stop receiving messages from you

Sending the Message
ISPs and recipients generally react favourably to these kinds of reconfirmation messages. However, many run automated blocking systems. A high bounce rate could block your message, at least temporarily. And some recipients may complain about an opt-in reconfirmation.

To minimize the effect of this, send the message in batches and monitor bounce rates and complaints. If a particular segment proves troublesome, switch it from opt-out to opt-in (if it’s not already) or discard it completely. If you’re experiencing deliverability issues with a particular ISP, talk with it about what you’re doing and why. ISPs are generally very receptive to people trying to do the right thing. Hold off on sending more messages until the issue is resolved. With a well-crafted reconfirmation message, you can effectively clean up an otherwise problematic list. Once it’s clean, keep it that way with regular communications and good list hygiene practices.

By Derek Harding

Have you experienced it yet? If not, it’s only a matter of time. One ClickZ columnist touched on it in not just one but two columns back in September. I thought it was a tempest in a teapot until I met an interactive agency professional at a conference who referenced it. And just when I thought it was over, a listserv I’m active in had its liveliest debate of the year about it.

Let me lay out the issues, the counterpoints, and my thoughts. Then, I’m curious to hear what you think. Give me your perspective; I’ll include your responses in a future column.

Seven Reasons to Get Rid of HTML E-Mail

• HTML takes more resources/bandwidth than text. Yes, it does. HTML tends to be more expensive to produce, and the end result will almost always be an e-mail with a file size larger than its text equivalent.

That said, HTML provides more flexibility in color, layout, fonts, and image inclusion. You can use HTML judiciously without going over the top. For instance, add just a graphic logo and a two-column format (which can be easier to read than text because of the shorter line lengths).

Case in point: Rick Gardinier, VP of Interactive at Blattner Brunner Inc., relayed a case study to me. He had a business-to-business (B2B) client who was sending plain text e-mail; performance was not that great. A simple HTML design, utilizing the company’s logo and colours, delivered a big lift in response rate.

• HTML is prone to deliverability problems. HTML messages “can get caught in spam filters” is how one critic described it. Yes, they can. So can text e-mail. The new technologies coming online, SPF (define), Sender ID, and Domain Keys, should help with this problem.

Earlier this year, ClickZ reported false positives were approaching 19 percent, up from the previous two years. This is a problem. But there are many different types of filters, and most don’t equate HTML with spam, so text messages are caught as well.

Text messages aren’t a silver bullet to get around the deliverability issue. Even if they were, you’d have to factor in response before you could declare a clear winner in the text versus HTML debate.

• HTML doesn’t always render properly. Very true. You need to code HTML specifically for a variety of e-mail clients (not just for Web browsers) to be sure it looks OK. Even then, the e-mail clients can throw you a curve; some of the newer versions won’t show graphics unless the sender address is in the recipient’s address book. Be sure the message will still be understood (e.g., by making headlines text instead of graphics) even if the images don’t appear.

Understanding all this, does it make sense to just go back to plain text? I don’t think so. I haven’t seen my clients’ open rates or response rates decline significantly since these new e-mail clients became available.

DoubleClick reports a decrease in open rates and click-throughs in the past year, but even that is small in context. Its open rates have gone from 38.8 percent to 36.0 percent, a loss of 2.8 opens per hundred e-mail messages sent. Click-throughs have declined from 8.3 percent to 7.7 percent, a loss of just 0.6 per hundred clicks.

These decreases don’t justify ditching HTML altogether right now. Further analysis shows e-mail had similar metrics in 2002, suggesting this may be a natural up-and-down cycle.

• People dislike HTML. This is an emotional topic for many. Comments such as, “I don’t read HTML e-mail, and neither does anyone I know” and “I assume all HTML e-mail is spam and delete it” are prevalent.

I actually like HTML. As a recipient, it’s easier to read and grabs my attention. As a marketer, there are some products that just lend themselves to HTML e-mail. One of my clients is a large toy, game, and collectibles manufacturer. It’s much easier to make a toy seem appealing by showing a photograph of it than to try to explain its cuteness in a few sentences. Sometimes a picture really is worth a thousand words.

• There are viable alternatives to HTML. Many suggest we should “send a text e-mail with a link” to a Web page. Others chime in that “e-mail should be enough text to arouse sufficient interest to click on the link” to the Web page.

Been there, done that. It was the mid-1990s, before HTML e-mail was viable. And those of us in that place at that time were so happy to suddenly have the option of columns, fonts, and other formatting (not to mention graphics). Even as a recipient, I was excited by HTML and hyperlinks. I just can’t imagine going back.

Unless, of course, there’s a good reason to go back. This is where testing comes in. No one I encountered quoted statistics showing that text e-mail performs better than HTML. Paul Broni, executive VP/client services director for Inbox Interactive summed up my experience when he said, “HTML is far and away the favourite [over text]; our metrics speak for themselves.”

• E-mail is about content, not design. The most concise explanation of this part of the anti-HTML platform is, “E-mail is about text, conveying information. It’s not about formatting or design. I’m not opposed to attractive design, but simple always beats pointlessly complex.”

I’ve always said content and copy are king in e-mail, and design, formatting, layout, and so forth should support these. Copy is what gets your message across; even the best picture will probably benefit from a strong copy-based call to action.

So why do people assume that when we talk HTML, we mean something that’s complex, something that will detract from, instead of add to, the message? The answer isn’t to get rid of HTML; it’s to teach designers and marketers to use it judiciously.

• HTML is “dangerous???. So is the bathroom. Isn’t that where most accidents occur? But I hope none of us ever decides to stop bathing. HTML opponents say, “Blocking HTML keeps viruses from getting through to your computer.” So does good virus software.

Unfortunately, viruses are a fact of life on the Internet. The danger isn’t limited to e-mail. Certain browsers can make your computer more susceptible to attacks. The best way to protect yourself is not to go online, at least not with your own computer. Is it a risk? Yes. Does the risk outweigh the benefit of HTML e-mail? Not for me. Not yet.

Where I Stand
I’m not ready to give up on HTML e-mail just yet.

As a marketer, I was taught to go with what generates response, as long as it’s ethical, and complies with all regulations and you won’t be mortified if an article about what you’re doing shows up on the front page of The New York Times (or ClickZ). My clients get good response rates with HTML e-mail, and I can’t see moving back to text-only. Not at this point.
?
John Verba of Adgression said it even more eloquently: “[Companies doing HTML e-mail must be] quite happy with the return… or they’d just be cranking out text-only, all the time.”

By Jeanne Jennings

Too often, marketers cling to the same campaign elements and tactics without thoroughly testing e-mail variables to determine what will drive performance higher. In fact a Jupiter Research study I recently concluded on the topic found that overall, about 60 percent of marketers don’t test. Challenged by resource constraints, most of these non-testers simply feel they don’t have the time to regularly test their campaigns.

Yet our research found the marketers who are testing on a regular basis (at least every other mailing) are almost twice as likely to have e-mail conversion rates that exceed industry averages by one to two percent, as compared with marketers who don’t test. Clearly, the effort of testing e-mail variables (e.g. subject lines, content) is offset by better results. Testing is really the only way for a marketer to determine which element drives higher results. It’s critical for optimizing e-mail campaigns. To ease into testing, let’s take a look at some of the rudimentary items to test.

Identify Marketing Goals. While this sounds very basic, I do find on occasion that e-mail becomes a primary promotional tool simply because it’s inexpensive. It is easy to get mired in the monotony of getting the e-mail out and lose sight of specific campaign or mailing goals. Before you can conduct any testing you must first know what the desired outcome is.

Identify Consistent Audience Segments. One of the first elements to test, and an early deliverable that comes out of testing, is audience segmentation. Ensure your e-mail solution can support multiple lists, and customer profiles can be easily divided into segments. Some attributes you’ll want to test related to audience segmentation include demographic data (e.g. income, age, sex); behaviour (e.g. open, click); purchase history (e.g. recency, frequency); acquisition source; attitudes and domain (e.g. AOL, Yahoo!). I’ve discussed using behaviour and attitudinal segmentation in previous columns; take a look at those for more ideas.

Experiment with Message Formats. While HTML is the preferred format for most marketers, anti-spam measures, including image and HTML blocking at many leading ISPs, underscore the importance of testing message format. Look for differences in delivery, open rate, click through rate and conversion. Testing formats on a domain level and obtaining insight into the domain level performance is critical to really understand the impact the message format has on performance.

Tinker with Content. One area likely to have the largest impact on campaign performance is testing different creative elements. Experiment with the subject line, personalized campaign elements, number of products and/or offers presented, and message tone. Research I’ve conducted has found personalization can have a dramatic impact on e-mail conversion rates. Determining if personalization is appropriate for your campaigns can only be achieved through testing.

Test Frequency. In order to determine your optimal mailing frequency or the impact of triggered lifecycle campaigns, test message frequency over time. My column on message frequency offers some ideas on alternate mailing intervals.

While testing is beneficial and all the above items are just some of the variables you can test, it’s very important to adhere to the following basic testing guidelines to measure which element is driving the results.

• Test only one variable at a time. Do this to determine which element is impacting performance test offers and subject lines, but don’t test both at the same time.
• Maintain a control group. To understand how the segments that are being tested behave over time, maintain a control group that doesn’t receive any of the testing treatments.
• Ensure tests occur on the same day. To minimize fluctuations in day-of-week response patterns, ensure all segments receive the tests on the same day of the week. That is, unless you’re testing day of week mailing patterns. If so, constrict the test to that one variable.
• Ensure tests are statically accurate. While the size of mailing test cells depends on your mailing list and testing method (A/B, Nth, etc.), it’s recommended that each test cell can return at least 100 qualified responses. Based on response rates, this may require cells that contain as many as 10,000 to 15,000 names.

Trial and error is a central part of marketing. Hopefully, these suggestions will make your trials a bit more scientific. Good luck — and let me know how it turns out.

By David Daniels