B2B Email Metrics Part One – The Big Open Rate Lie 

Why email open rates can’t be measured

It’s time for the inevitable technical bit. I’ll keep this brief and skip over some details.

The systems and standards that underpin internet email date back to academic communication in the 1970s, long before there was any thought of using email for marketing purposes. Emails were plain text. Systems were designed to ensure delivery of email – that is, to make sure that the message you send gets to the recipient’s inbox. There was never any provision for finding out whether the delivered message was ever opened. Presumably it was sufficiently unusual and exciting to receive an email that one opened everything!

Skip forward to the 1990s. HTML emerged as a standard for graphical content on the web, and many email readers started to support the display of HTML emails, which could include embedded images.

Email marketers realised that these embedded images offered a sneaky way to find out whether an email had been opened. Because the image isn’t sent with the email itself, but rather it is downloaded dynamically when the email is displayed, we could detect and record these dynamic image requests. So, the logic was:

Embedded image requested => email has been rendered in an email client => email has been opened by the recipient

Count the number of image requests, and you’ll know how many opens.

(An aside: if you also add some personalised information to the embedded images, you can tell which individual recipient opens any given email. This is used in a lot of marketing automation systems. And because this approach is open to abuse, a lot of email clients will attempt to block or restrict it.)

Now the truth is that this approach was never all that reliable. People read email in lots of different ways and different email clients treat images differently. Microsoft Outlook, for example, doesn’t download images unless a user specifically asks for them. So, a person could open and read an email without generating an image request. And there are also ambiguous situations such as when the same person opens the same email on two devices – do we count that as one open, or two? So, at best “open rate” was a rough proxy for true user behaviour.

Recently the situation has become much worse with a new privacy feature from Apple “Mail Privacy Protection”. This means that emails sent to Apple devices will appear to ALWAYS be opened. The “open rate” becomes pretty much meaningless for any campaigns that are delivered to any significant number of Apple devices. It’s no longer even much use as a proxy measure.

To see why this is the case, consider this extreme example:

Campaign 1 is sent to 100 people who use Apple devices with MPP. None of them open the email. It will report an open rate of 100%. But the true open rate is. 0%.

Campaign 2 is sent to 100 people who use Outlook on a Windows desktop. All of them open the email, but with image leading turned off. It will report an open rate of 0%. But the true open rate is 100%.

That’s just about as misleading as it is possible to be! And admittedly it’s an extreme example. But it shows the underlying problem – open rate today is determined by the mix of devices and email systems that your recipients use, and not by whether they open or read your content.