Is my open rate completely useless now?

No, but it answers one question instead of several. A sudden collapse across the whole list is a real alarm worth investigating the same hour. A high or rising rate mostly reflects how much of your list reads mail in Apple Mail, and comparing your rate against another company's tells you nothing.

Can I tell which opens came from Apple's background download?

Not reliably from a platform report. Some platforms filter known pre-fetchers and label the rest, and the filtering is a best effort, not a clean split. The practical move is to stop setting targets on opens rather than to try to clean them.

Should I remove the tracking pixel from my emails?

You can keep it. It still powers the collapse-to-zero alarm and it costs nothing in deliverability terms. What has to change is the decisions you hang on it, above all deleting subscribers who look inactive when the signal cannot see them.

Apple Mail Privacy Protection: what it does to open rates

Say a 20-person online school moves its newsletter to a new platform. The first send comes back with an open rate near seventy percent, roughly double what the old tool reported, and nobody changed the subject lines. The marketing lead is delighted for about a week. Then the enrolment numbers for the month arrive and they look exactly like last month.

Nothing improved. The measurement changed. Open tracking stopped meaning what it used to mean once Apple changed the way its Mail app loads remote images, and most small businesses are still steering by it. That means tuning subject lines against noise and, worse, deleting subscribers who look inactive and are not.

This is one of the first things we look at when a client tells us their email stopped working. Often the email is fine and the dashboard is the thing that is broken.

What an open actually is, and why it was always fragile

There is no open event in email. There never was. Every open rate you have ever seen is inferred from a one-pixel image hosted by your sending platform at a unique address per recipient. When a mail client fetches that image, the platform logs a request and calls it an open.

The whole method rests on one assumption: that a mail client downloads remote images at the moment a human looks at the message. That assumption is the part that broke, and it broke in two different ways at the two mailbox providers most of your list uses.

What Apple Mail does with your pixel

Mail Privacy Protection is a setting in Apple's Mail app. Apple's own wording is the clearest statement of what it does to a sender: Protect Mail Activity "downloads remote content in the background by default, regardless of whether you engage with the email". The image is fetched when the message arrives, not when anyone reads it.

Apple also routes those downloads through two relays run by different parties. In Apple's description, the first relay sees the reader's IP address but not the content, and the second sees the content but not the address, so no single party can identify both the reader and what they received. Checked 14 October 2026.

For your dashboard the consequences are specific:

A message delivered to an Apple Mail account with the setting on can register as an open whether or not a person ever looked at it.
The open timestamp is close to the delivery time, not a reading time, so any best-time-to-send analysis built on it is measuring your own sending schedule.
The location and the device in your report describe a relay, not a reader.

Two limits are worth knowing, because they are why the number is a mixture and not a flat ceiling. Apple says the protection does not extend to links or to attachments, so click tracking survives it. And it depends on the network: Apple's guidance notes that if the option to limit IP address tracking is turned off for a Wi-Fi or Ethernet network, the address is not hidden on that network, and that with a VPN active Mail may show "Unable to load remote content privately" and offer a manual Load Content instead.

What Gmail does, which is not the same thing

Gmail does not pre-fetch on arrival. It proxies. Google's admin documentation says Gmail serves the images that appear in messages through Google's secure proxy servers, which protects users and the domain against image-based vulnerabilities. Gmail's help page for readers is explicit about the sender's side of that: senders cannot use image loading to get information about your computer or location and cannot set or read cookies in your browser, but senders "may know whether you've opened an email that has an image". Checked 14 October 2026.

So a Gmail open is closer to a real signal than an Apple Mail open, and it still arrives stripped of everything around it. You get a yes, not a who, a where or an on what. And a reader who keeps images off produces no open at all while reading every word you wrote.

What is left of the number

An open rate today is a sum of at least four different things: messages a person read with images on, messages a machine fetched on arrival, messages read with images off that counted as nothing, and messages a security scanner pulled before any human saw them. Your platform reports one figure. It cannot separate those four, and neither can you, because the inputs that would let you are exactly the ones Apple and Google removed on purpose.

That does not make the number worthless. It makes it good for one job. A sudden fall to near zero across a whole list is a delivery or rendering problem and it is worth looking at within the hour. A slow rise usually means your audience mix moved towards Apple Mail. And forty percent on your list against sixty percent on somebody else's tells you nothing about either one.

What we check first on a client account

Whether authentication passes, before anyone talks about content. If SPF, DKIM and DMARC are not aligned for the sending domain, every engagement question is premature.
Whether the platform reports unique opens or total opens, and whether it filters known pre-fetchers. Platforms differ, and a switch between those two definitions explains a surprising share of overnight improvements.
The share of the list on Apple addresses and on Gmail, which sets roughly how much of the open rate is mechanical.
Whether anyone has already deleted subscribers for looking inactive, and when. This is the expensive one, because deletion is not reversible and a re-import starts the engagement history over.
What clicks and conversions did over the same period, since those are the series whose definition did not change.

The numbers that still mean something

Clicks. A click is a deliberate action on a link, and Apple states that the protection does not cover links. Clicks carry their own contamination from security scanners and link checkers, which is why we read click-to-open as a trend on one list over months and never as a benchmark against another company's figure.

Replies and conversions. If the newsletter exists to get people to book a call or buy a course, the booking count and the order count are the numbers. They are smaller, they move slower, and they are the only ones that are not a proxy for something else.

Postmaster Tools. For mail sent to personal Gmail accounts, Google reports user-reported spam rate, message authentication, delivery errors and a compliance status view. Google's sender guidelines say to keep the spam rate reported there below 0.10% and never let it reach 0.30% or higher. Two caveats from Google's own pages, checked 14 October 2026: the spam rate covers DKIM-authenticated messages sent to Gmail accounts, and the data is not real time, typically updated within 24 hours and sometimes longer. Google's notice about the old Postmaster Tools interface also says the Domain and IP Reputation dashboards will be retired, and it does not name a date, so do not build a report that depends on them.

Bounces and complaints from your own platform. Hard bounce rate per send and complaint rate per send are counted at the receiving end rather than inferred from an image, which makes them the sturdiest things on the dashboard.

Pruning a list when you cannot see opens

Here is the collision nobody warns you about. Google's guidance for senders suggests you consider unsubscribing recipients who do not open or read your messages, and that advice is sound in spirit: a list of people who ignore you is how a spam rate climbs. But the open signal the advice depends on is the one Apple has made unreadable for a large part of your list. Follow it literally and you delete paying customers for owning an iPhone.

What we do on a client list instead:

Define inactive on clicks over a long window, twelve months rather than ninety days, and never on opens alone.
Keep anyone with a purchase, a booking, a reply or a support ticket in that window, whatever the tracking says about them.
Treat repeated bounces as the hard signal. Google's sender guidelines list automatically unsubscribing recipients with multiple bounced messages as a requirement, and a bounce is measured at the far end, not inferred.
Before deleting anyone, send one plain re-permission message with a visible link to stay on the list, and let the people who want the mail say so.
Make leaving trivial. One-click unsubscribe headers in place, a visible unsubscribe link in the body, and requests honoured inside 48 hours, which is the window Google's subscription guidance names. A reader who cannot find the unsubscribe link reaches for the spam button, and the spam rate is the number that actually costs you delivery.

What we watch afterwards

We rebuild the reporting around what survived, then watch it for a quarter: clicks and conversions per send on one consistent definition, spam rate and authentication in Postmaster Tools, hard bounce rate per send, and unsubscribes as a share of delivered mail. Opens stay on the dashboard as a smoke alarm, labelled as one, so nobody sets a target against them again.

The hard part is internal rather than technical. If somebody has reported open rate to a board for three years, changing the metric looks like hiding a bad number. We write the one-page explanation that goes with the change, in plain language, so the first question in the next meeting already has an answer.

How Guanacos Tech helps

We do this as part of deliverability work for small and mid-sized teams across North America and Latin America: verify the authentication floor first, work out how much of the open rate is mechanical, rebuild the short list of metrics that still mean something, and rewrite the list hygiene rules so nobody prunes a customer for using Apple Mail. How we work fits on one page. If you would rather have it done than researched, we offer email deliverability consulting for small business, and half an hour is enough to go through your last three sends and say which numbers you can trust.

Your open rate went up and nothing got better: what Apple Mail Privacy Protection and the Gmail image proxy did to open tracking