Senders around the world are reporting the same thing: open rates for Gmail recipients have fallen sharply, with nothing changed on their side. Same lists, same content, delivery is fine, and yet the number in the report keeps sliding.
The explanation is simpler than it looks, and it points somewhere very different from “my campaign failed”.
How an open is actually measured
There is no technical way to know that a person opened an email. What every sending platform does instead is embed a tiny transparent image in the message body, a tracking pixel. When the mail client loads that image from the server, the platform records an open.
An open is an image load. Nothing more than that.
What Gmail did for years
Gmail never lets a recipient load images straight from the sender. Everything is routed through Google proxy servers. Inside that mechanism, Gmail spent years pre-fetching the images in incoming mail, before the recipient had looked at anything.
The result: Google servers loaded the tracking pixel, and the sending platform logged an open automatically. For years that was a fixed component of every sender’s Gmail open rate.
What changed in 2026
During 2026 Google changed that behaviour. According to the major deliverability vendors, Gmail has sharply reduced image pre-fetching and now applies it mainly where it expects the recipient to actually open the message.
Mailgun described the pattern in April across a wide range of senders: Gmail open rates falling, with almost no movement in clicks, conversions, complaints or bounces. That combination points to a change in measurement, not in recipient behaviour.
Validity measured roughly a one third drop in Gmail image-loading activity, and some senders reported Gmail open rates down 30 percent or more across a quarter.
In parallel, Google rolled out AI summaries inside Gmail and Gemini integration, letting people ask questions about their inbox and get answers drawn from their mail without opening it. A recipient who reads the summary at the top of the inbox got the information. The pixel never loaded, so no open was recorded.
Attachments: the blind spot in every open-rate report
There is a whole category of email where open tracking was never accurate, and is now less accurate still: mail whose actual content sits in the attachment. A price list, a bulletin, a brochure, a form.
The reason is simple. The tracking pixel lives in the message body. An attachment is a separate object. When the recipient opens the PDF, on a phone, on a desktop, or in Gmail’s own preview, nothing travels back to the sender. There is no mechanism that reports whether a file was opened, read, saved or printed.
Gmail makes this sharper. It previews attachments straight from the message list and lets people download them without entering the message at all. A recipient can read an entire bulletin while the pixel never loads once.
So a publisher whose product is the attached file sees an open rate that reflects only the people who also scrolled the body. The readers of the file itself were never counted, and never have been.
For this kind of sending, open rate was never a readership metric, and today it is even less of one. The metrics that matter are clean delivery, list stability, and the absence of complaints and bounces.
What it means in practice
The most important point: the mail is still arriving. Delivery has not changed, inbox placement has not changed, and recipients keep receiving exactly what they received before. What changed is the counting. People who used to be counted as openers automatically are no longer counted that way.
If you see a sharp drop concentrated in Gmail recipients, with no matching drop in clicks or replies, you are almost certainly looking at this change and not at a broken campaign.
How to confirm it
The deliverability vendors suggest a simple check. If these hold steady while opens fall, it is a measurement change:
- Clicks have not dropped proportionally
- Spam complaints have not risen
- Bounces have not risen
- Google Postmaster Tools reputation is still green
One more check: break your open rate down by mailbox provider. If the drop is concentrated in Gmail while Outlook, Yahoo and the rest hold steady, that is the signature of this change.
What this means for list hygiene
List cleaning is where this change does the most quiet damage. Engagement-based pruning built on open data is now working from a signal that shrank for reasons that have nothing to do with your subscribers. Run the old rule unchanged and you will remove people who are still receiving, still reading, and still buying.
The healthier approach is to separate two things that open rate used to blur together. Validity is whether an address still exists and accepts mail, and that is measured by bounces, by mail server responses, and by verification, none of which depend on a tracking pixel. Engagement is whether a person still wants your mail, and that is now best read through clicks, replies and complaints. Clean on validity with confidence. Judge engagement more slowly, and over a longer window than you used to.
What to do, and what not to do
Do not send more. The instinctive response to falling opens is to raise frequency or re-send to non-openers. That is a mistake. Repeatedly mailing people who never engaged is one of the strongest negative signals Gmail reads, and it can damage a reputation that is currently fine.
Be careful with open-based list cleaning. Rules like “remove anyone who has not opened in six months” were written for a world where Gmail counted differently. Today a subscriber who receives every send and even reads the Gmail summary each morning can look like a dead address. Automated pruning on opens alone will remove active people.
Move to clicks. A click is a deliberate action that cannot be generated by accident, which makes it the most reliable engagement signal available right now. If your mail has no clickable links, add some.
Track trends, not absolute numbers. This week against last week, same audience, still tells you something. The number itself no longer represents how many people read.
In short
Gmail did not stop delivering mail. It changed how it counts opens. Recipients still get your campaigns, some of them simply are not counted as openers any more. Understanding that lets you adjust the metrics and the rules you built on top of them, instead of reacting to a moved number with steps that damage the list.
Frequently asked questions
My open rate dropped. Is my email still being delivered?
Yes. Delivery and inbox placement have not changed. What changed is how Gmail counts opens. Recipients who used to be counted automatically are no longer counted, even though they still receive the message.
How do I know it is not my campaign?
Check four things: clicks, spam complaints, bounces, and your Google Postmaster Tools reputation. If all four are steady while opens fall, it is a measurement change. Then break the open rate down by provider. A drop concentrated in Gmail while Outlook and Yahoo hold steady is the signature.
Can I tell whether someone opened an attachment?
No. The tracking pixel sits in the message body only, and an attachment is a separate object. When a recipient opens a PDF, nothing is reported back. For mail whose content lives in the attachment, open rate never reflected readership.
Should I remove subscribers who have not opened in six months?
Not automatically, and not on opens alone. Pruning rules written before this change will now remove active subscribers. Combine clicks, bounces and complaints in the decision.
Which metric should I move to?
Clicks. A click is deliberate and cannot happen by accident, which makes it the most reliable engagement measure today. Alongside it, follow trends over time within the same audience rather than the absolute number.
Sources: Mailgun, “Gmail open rates suddenly look lower” (April 2026); Validity, “What is Really Behind Gmail’s Open Rate Drop” (May 2026); beehiiv, “Declining open rates: A diagnostic guide” (2026); MessageFlow, “Low Email Open Rate: Causes and How to Fix It” (May 2026).