Tuesday, December 21, 2010

An adjustment to how we calculate open rates

Tonight we've made a slight adjustment to how our system detects an "open" on a broadcast email campaign.

First, some background

Email marketing systems detect that an email has been opened by inserting a 1-pixel image at the bottom of the email message and waiting for that image to be downloaded from the server. Some email clients block images by default, and therefore they prevent email marketing systems like JangoMail from detecting the "open". If an email message can be understood by its recipient by not seeing the actual images, because perhaps there is enough text to explain the email message, then it's even more likely that the 1-pixel image won't be downloaded.

Therefore, we have always only counted an email as being "opened" if the 1-pixel image at the bottom of the email was downloaded, meaning that the recipient had viewed the email, and that images (even if there aren't any in the email body) are turned on.

New way to detect an open

We recently determined, however, that in many cases, a recipient may click a URL in an email (thus counting as a click), but never downloads the images in the email (thus NOT counting the open). And since clicking the email is indicative of also opening it, we've made the following change to Open Tracking:

If an email is clicked, but that recipient has not yet been counted as an "open" for that email campaign because images haven't been downloaded, the recipient now WILL get counted as an "open" anyway due to the click.
The net effect of this change is that you will now see higher Open Rates, and the Open Rates will be more accurate and reflective of what recipients are actually doing with the email message.

How much of a difference does this make?

Our analysis shows that this makes a big difference. For a recent email campaign we did to a segment of our own customers, these were the statistics in Reporting:

Recipients: 6,369
Unique Opens: 993 (15.6%)
Unique Clicks: 384 (6.0%)

An analysis of the data showed that there were 118 unique recipients that clicked that never registered as having opened the email. So if we add the 118 to 993, that's now a total of 1,111 opens, for an actual open rate of 17.4% instead of 15.6%.