How Email Works, Part Four: Measurement and Tracking

I had intended to have this just be a three part series that covered how emails are deployed and delivered. Recent conversations made me realize that not looking at the content of the messages leaves out key pieces of the puzzle. Specifically measurement and tracking – what we know and what we can infer about our campaigns.

In part three
 we looked in detail at the headers of an email message. This time we’ll complete the picture with the message body to see how opens and clicks are actually measured.

Let’s start where we left off with the Content-Type header, as shown below.

This header indicates the message contains two separate versions of the content. One is plain text and the other HTML. I’ve simplified the copy to improve legibility and colored the multipurpose internet mail extensions (MIME) headers yellow, the text copy green, and the HTML blue.

There is no absolute requirement to use multipart/alternative. It is also not mandatory to have just two parts, that the parts be HTML and text, or contain the same content. For example, the Apple Watch has introduced a new type: text/watch-html. Some organizations send a boilerplate text version which links to the Web browser view, whereas writers even believe that the text version should be discarded entirely.

That’s the structure – now to the tracking. The essential response metrics for email are click-throughs and opens. In both the text and HTML versions of the message there is a link that reads:


This is not the link that was placed into the message in the ESP’s user interface. That link was replaced with this one during the submission process we covered in part one. Encoded into the URL is information about the recipient, which link this is, and which email it is in.

When a subscriber clicks on the link their browser visits the ESP’s site which records the details then tells the browser to go to the actual destination page. All of this happens almost instantaneously and enables the click-through response tracking.

The other key element is a tracking pixel. At the bottom of the message there is a line starting with ” < img.” This was added by the ESP during the submission process. If a subscriber has images turned on, their email program will request that image in a way similar to a click-through. All that will be returned is a 1×1 pixel transparent or invisible image, but it will enable the ESP to know that they viewed the message. This forms the basis of the open-rate calculation.

There are a number of implications of these two technologies.

  1. Open tracking is only possible in HTML, since text doesn’t support images.
  2. Open tracking only works when the recipient has images turned on. This can cause wide variations in open numbers, depending on whether a given ISP or email client loads images by default. It can also under report actual reads.
  3. During the click-through or open tracking event, your email client sends a number of other pieces of information. This includes which email program you use, your operating system and browser, and your IP address – which can be tied to your ESP and location. When Gmail added image caching it was partly to hide this information.

You may have noticed that I have not mentioned another key metric – the inbox delivery rate – and this is because that information is generally not available. Non-delivery is notified by the sending message transfer agent (MTA), but inbox placement is far harder to determine.

Marketers use seed addresses to determine whether an email made it to the inbox. Seeds are added to the send. Next, a provider retrieves the seed addresses’ email to see if it is in the inbox or the spam folder. The problem with this approach is that the seeds are becoming less representative of overall filtering results, as ISP filters increase in sophistication.

To address this, some providers have started to use panels that consist of millions of email users that all permit providers to monitor their mailboxes. These panels can provide more accurate information on email delivery for large senders.

A solid understanding of how email is measured, including firm grasp of the strengths and weaknesses of each metric, places you in a better position to make good use of the data available.  

Did I leave anything out? Anything that isn’t clear? If there are any questions you’d like answered write a comment and let me know.

Related reading