Commentary

Bounceback Handling Debate

After reading Dylan Boyd's Q and A with Jeanniey Mullen last week, I realized I agreed with the posting and disagreed with it at the same time.

I agree that tracking bouncebacks should be a major initiative for e-mail marketers. It is a solid indication of your list quality, and it shows if there is something wrong with your sending infrastructure. It is also a critical component to your e-mail reputation. The era of laissez-faire list management is coming to an end. It's been clearly communicated by the receiving community that e-mail reputation will be the determining factor for delivery--not the brand name or logo behind the e-mail.

When it comes to bouncebacks, the unspoken worry has been shrinking your list by being proactive about cleaning it. List size was--and will continue to be--a measure of success for e-mail marketers. But why frantically fill a bucket that has a big hole at the bottom? Patch the leaks before you invest any more money in filling it. How do you patch it? It's all about data sources.

advertisement

advertisement

We tend to look at our e-mail database as one monolithic entity, when in fact it is a compilation of hundreds--if not thousands--of unique pieces. Think of all the different ways data was put in there, and then ask yourself these questions:

  • Are all the places you collect data on your own Web site consistent?

  • If you use partner or affiliate programs, how many different Web pages did the data come from (I estimate this in the thousands per program)? Are their data collection methodologies in line with yours?

  • If you have gone through any mergers or acquisitions, have you communicated the e-mail implications to your customers?

  • Have you gotten permission before moving customers to new programs?

  • Have you updated data collection forms to account for strategic shifts over time?

    Now that you've asked yourself these questions, apply your bounce rate problem to it. Are you able to isolate your bounces to specific segments of your list? Can you isolate problem sources in your data collection? Are one or two apples spoiling the barrel, or is it systemic across all data sources? Is the problem a recent development, an old issue that corrected itself, or is it ongoing? In the traditional world, unproductive sources of data are cut off. In the online world, it's not that we don't follow the same mentality. We just don't know where the problems are to fix.

    What did I disagree with? Jeanniey's case study example indicates that 80 percent of the bounces were associated to aggressive content filtering. While this may have been the case for the one campaign in question, it is very unlikely. (This may simply be a definitional difference for what is meant by a "bounce," commonly defined as responses from the receiving mail server saying that the mail has not been accepted for delivery.) I feel it important to make sure readers know that content has very little to do with bounces.

    There are two primary reasons why your e-mail doesn't get delivered:

    1) The mail server you sent the message to responds with a bounce. There are a variety of different kinds of bounce responses (the following assumes you have a good address to mail to). Commonly, the most serious bounces are "policy" blocks that indicate your mail isn't getting delivered because the receiver doesn't like the reputation of your mail server and isn't accepting mail from it. This may be driven by high complaint rates, high unknown user rates, consistently hitting spam trap addresses or a variety of other reputational issues. In addition, there are a wide variety of other bounce responses that indicate other problems, including "temporary failures" that indicate you should try back later. In our experience, the content of the message is very rarely the reason for one of these bounce codes. It happens, but is extremely rare.

    2) Your message gets filtered/discarded after the server accepts your message. Nearly all e-mail platforms have some filtering after the server accepts messages, and none that I know of register a bounce that gets sent back to the sender. Depending on the filtering criteria, changing aspects of your message may help here. While that may increase your inbox delivery rate, it has nothing to do with your bounce rate.

    Our data shows that playing the content filtering game is a bit of a goose chase. It's very rare that one word in your copy or a color in the HTML will trigger filtering. Moreover, our data and experience would indicate the majority of delivery problems are based on the reputation of the sending mail server--not the content of the message.

    In other words, if you sent the same message from another server, it would be delivered. Most times filtering occurs because your e-mail shares similar characteristics with known spam. So guess who's sending the previously known samples of spam? Probably you! Why? Likely because the e-mail you sent that triggered these characteristics had high abuse complaints, bounce rates or spam traps included. The most important way to ensure your e-mail gets delivered is to make sure all of your e-mail programs have a great reputation. It's time we move away from the symptoms and start to focus on the source of our problem: complaints, unknown user rates, data sourcing/hitting spam traps and having poorly configured sending infrastructure.

  • Next story loading loading..