If you work in email marketing, you’ve probably heard this warning many times:
Don’t trust your open rate. It’s inflated.
Since Apple introduced Mail Privacy Protection (or MPP), Apple Mail can download images (and tracking pixels) in an email automatically. That means your email platform can register an “open” even when the recipient hasn’t actually read your email.
So yes, open rates can absolutely be inflated – and it’s something we’ve explored in the past.
But while analyzing data for our upcoming Email Detective annual report, we found something that goes completely against this:
Open rates can also be undercounted.
And as our data revealed, it happens surprisingly often.
How can someone open an email without registering an open?
To understand this, it helps to remember how open tracking actually works.
When you send a newsletter through an email marketing service, a tiny invisible image, usually a 1×1 pixel, is added to the email. When the recipient opens the email (and their email client loads that image), the sender gets a signal.
Pixel loaded = open recorded.
The problem is that the reverse isn’t necessarily true; that is, pixel not loaded ≠ email not read.
Someone can open your email, read the subject line, introduction, offer and call to action, but if external images don’t load, the tracking pixel doesn’t load either.
As far as conventional open tracking is concerned, that person never opened the email. Which means you might actually be getting more opens than you realize.
Luckily, this is something we can observe with Email Detective, our browser extension for Gmail and Outlook.
Because Email Detective runs inside the recipient’s inbox, we can see that the user has genuinely opened an email. We can then check whether the remote images in that email were actually loaded.
And the results surprised me.
Nearly 6% of Gmail opens have images blocked (and the number is even higher in Outlook)
In our testing period, Email Detective observed 143,921 genuine email opens. Of the 142,383 Gmail opens in our dataset, images were blocked in 5.8% of opens.
Put another way:
About 1 in every 17 Gmail opens we observed would not have triggered a conventional tracking pixel.
Now, that doesn’t necessarily mean Gmail is blocking the images itself in every case. Gmail normally displays external images automatically, although users can change this setting. Google can also withhold images when it considers a sender or message suspicious. Browser privacy tools and other factors can interfere with remote images too.
From the email marketer’s point of view, however, the outcome is the same:
A real person is reading the email, but the tracking pixel can’t see them.
The effect was even larger in our Outlook on the web data. Out of 1,538 opens, 11% had images blocked. That’s roughly 1 in every 9 Outlook web opens.
Now, it’s true that our Outlook sample is currently much smaller than our Gmail sample, making it harder to conclude anything definitively. But my theory is that, because Outlook is notorious for putting so many emails in junk, the images simply never get loaded.
So is a 40% open rate actually 43%?
Let’s turn these numbers into a simple example.
Imagine you send your newsletter to 1,000 subscribers:
- 700 use Gmail
- 300 use Outlook
- Your email platform reports a 40% open rate
That’s 400 recorded opens.
For simplicity, let’s assume that’s:
- 280 Gmail opens
- 120 Outlook opens
But our data suggests some genuine opens may never have fired the tracking pixel. If we correct those numbers using the image-blocking rates we observed, we’d end up with roughly:
- 298 Gmail opens
- 134 Outlook opens
That’s about 432 genuine opens instead of 400.
So the reported 40% open rate could be closer to 43.2%.
At the same time, other technologies can be pushing your reported open rate in the opposite direction.
Apple Mail Privacy Protection is the obvious example. Apple can fetch remote email content automatically, causing a tracking pixel to load regardless of whether the recipient actually reads the message.
Security software and automated systems can create additional activity that marketers may mistake for human engagement.
And the way individual email marketing platforms identify and filter this activity varies – some exclude Apple MPP opens and bot activity, while others don’t.
So what actually ends up in your email marketing dashboard? Very roughly, the calculation looks like this:
Real human opens
+ Automated or privacy-generated opens
– Real opens where the tracking pixel didn’t load
= The open rate your ESP reports
A few more complications
There’s another implication of this data that I think is particularly interesting. Suppose two companies send equally good emails to equally engaged audiences:
- Company A has an audience made up mostly of Gmail and Google Workspace users.
- Company B has a much more Microsoft-heavy audience.
If different inboxes and email clients have different rates of image blocking, their reported open rates could differ even if their subscribers are equally engaged.
There is one more complication worth mentioning. Many email platforms don’t rely exclusively on the open-tracking pixel.
If somebody has images blocked but then clicks a tracked link in the email, some platforms will infer that the email must also have been opened and count it accordingly. Which makes sense – you can’t click a link inside an email without interacting with the email first.
And while that helps to deliver an open rate that’s closer to reality, the problem is not all ESPs do this.
All of this makes it harder to arrive at a hard-and-fast rule on how to calculate open rates accurately. A tracking pixel can’t reliably tell you that a human paid attention, which is why the distinction between “opened” and “read” is becoming increasingly important.
Does this mean you should stop looking at open rates?
Not at all. Despite all of these problems, I still look at our newsletter open rates. I just don’t interpret them as literally as I once did.
If one newsletter gets a higher open rate than another, I dig into potential reasons why. And open rates can be helpful for comparing your own campaigns over time, particularly when the composition of your audience and your sending setup remain relatively stable.
Here’s what I would focus on:
- Compare your campaigns with your own previous campaigns.
- Look for significant changes rather than obsessing over tiny differences.
- Combine opens with clicks, conversions, replies and other stronger engagement signals.
- Be cautious about comparing your open rate directly with another company’s benchmark.
And above all, remember that the number in your email marketing dashboard is an estimate produced by a tracking system. It’s not the same thing as the number of people who actually read your email.
That’s just one of the findings we’re exploring for our upcoming Email Detective annual report, where we’ll dig further into what actually happens to marketing emails once they reach real inboxes. To receive this report when it’s released in early 2027, make sure you’re subscribed to our newsletter.