The report shows an open rate of 48%. Does that mean almost half the people read the newsletter? No. It means that a tracking image loaded for some delivered messages — and sometimes not even that.
I wrote the original version of this article in January 2014, when Gmail began loading images through its own proxy. At the time, it looked like a major disruption to measurement. Today, there are even more reasons for caution. I would therefore use open rate as a rough operational signal, not as proof of interest.
An open is estimated using a small image
The sending system inserts a unique transparent image into a message. When the email client downloads the image, the server records a request and assigns it to the message. That creates an "open."
The method has two fundamental weaknesses:
- a person can read a message without loading images, and the open will not be recorded;
- an application or proxy can load an image automatically, and the open will be recorded without actual reading.
Measurement was therefore never a mind reader. The report merely sometimes acted like one.
Gmail caches and serves images through a proxy
In December 2013, Google announced that Gmail would show images automatically and serve them through secure proxy servers. The official Images Now Showing announcement explains that the sender does not deliver the image directly to the user from the original external server.
That does not mean Gmail erased all opens. It means that technical details such as repeated loading, IP address, or device do not carry the same informational value a sender might expect. The historical assumption that opens could no longer be measured at all after the change was too black and white.
Apple can load remote content in the background
For Mail Privacy Protection, Apple says that it hides IP addresses and downloads remote content in the background regardless of whether a person interacts with the message. The description is in its official documentation on privacy protection in Mail.
The practical impact is simple: some automatically recorded opens cannot be reliably separated from genuine ones. The "opened email" segment is therefore not a safe foundation for important automation, sales evaluation, or database cleaning.
What open rate can still tell you
I would not discard it completely. Within the same tool, similar audience, and the same period, a sharp change can draw attention to a delivery, subject-line, or measurement problem. You must, however, allow for the change also being caused by the mix of devices and privacy features.
I do not use open rate to claim that people read the content, or to compare different companies. A benchmark without the same methodology and audience looks precise, but does little to help a decision.
Measure steps closer to actual intent
- Delivery and rejection. Whether the message technically reached the mailbox provider.
- Complaints and unsubscribes. Whether the communication fails to fulfil its promise or reaches the wrong people.
- Relevant clicks. Not every click on a logo; follow the path to the content or offer promised by the message.
- Replies. In B2B communication, one substantive reply can mean more than hundreds of automatic opens.
- Follow-on action. An enquiry, registration, use of a tool, purchase, or another event matching the goal.
- Long-term quality. Activity of relevant contacts, business value, and the state of sending reputation.
I tag newsletter links with consistent UTM parameters so visits can be distinguished in analytics. Even a click is not automatically business. For every mailing, I decide beforehand what it should enable for a person and what decision I will make based on the result.
Beware of automation built on opens
The "did not open, send again" scenario can misclassify people because of privacy protections. Likewise, lead scoring that awards points for every open can reward an email application instead of a person's interest.
If automation needs a stronger signal, I use a conscious action: a click on a specific topic, a reply, a preference selection, or a completed step on the website. Even then, I watch the context. Mindlessly adding points is merely a more sophisticated way to create a pretty number.
Less certainty in a metric can lead to better marketing
The limitation of open rate is not a disaster. It forces marketing to stop mistaking an easily available metric for a business result. In the end, that is quite a useful service.
A newsletter is not read because a pixel was downloaded. It is read when it gave a person something. That is harder to measure — but at least we are measuring the right thing.
Practical communication setup continues in the article How to connect a blog and email marketing. You can find broader experience with deliverability in the email deliverability test and a technical evaluation of website steps in the Google Analytics 4 guide.