GM.

Every now and then I mention something that happened on our blue little marble.

Ben Bader — 25 years old, dropping dead in a sauna, turned out to be heart disease. Tragic. Terrible news to see a guy in his 20s like I am losing his future.

Dan Kennedy— stepping back, Magnetic Marketing getting absorbed into Russell Brunson's machine, a real passing of the torch in the direct response world. Worth noting.

But both of these events probably don’t move the needle for your business.

I toss these into my emails every now and then not to be morbid or gossipy, but because I think our world runs better when we're paying attention to what's actually happening, not pretending everything is fine and the old playbook still works.

Most of these things, though? They don't hit your inbox. They don't hit your wallet. You read, you nod, you move on.

This one's different and should raise an eyebrow for you writers and email marketers out there. And it's not everyday that I get to nerd out over data.

So anyway, Gmail made two changes recently. Some newsletter operators noticed them and started posting about it quietly, LinkedIn, Twitter, the usual corners of the internet where the people who actually care about this stuff gather.

I want you to understand both of them clearly, because the implications are real.

The first thing messes with how you measure engagement. The second messes with who you think is on your list. Let me take them one at a time.

Gmail prefetches images inside your emails, including your tracking pixel, before your subscriber ever touches anything. Google does this for speed and security. Makes sense on their end. The problem: your ESP receives that pixel fire and logs it as an open. The subscriber may never have looked at the email at all.

SparkPost(an ESP) analyzed nearly 10 billion Gmail open events and found consistent inflation of around 2 percentage points. Bird's research puts the false open range at 1–6% of Gmail events. That sounds modest until you realize you may be running segmentation rules and automation triggers off those numbers, meaning Google's servers are qualifying leads, not your actual readers.

And Gmail isn't alone. Apple Mail Privacy Protection now accounts for nearly half of all email opens, which means half of every "open" your ESP reports isn't a human reading your email. It's Apple's proxy server preloading a tracking pixel. Campaigns that tracked 28% open rates before Apple's change now show 52%, with identical click and conversion numbers. The open rate went up. Nothing else did.

The second change is actually new, and most people are sleeping on it.

Gmail recently started rolling out a feature allowing users to change their primary @gmail.com address while keeping the old one active as an alias. Emails sent to the old address keep arriving in the same inbox, and the original address still works for signing in to Google services.

Here's why that's a problem for you as a sender.

Users can change their Gmail address up to three times, for a total of four addresses. From your ESP's perspective, those are four completely separate people. There's no flag, no signal, no connection between them. So if someone on your list switches their Gmail address and re-subscribes with the new one, you're now mailing the same person twice, paying for both contacts, possibly triggering duplicate sequences, and if they unsubscribe on one address, the other keeps receiving your newsletter indefinitely.

The email address, the thing our entire list infrastructure is built on, is becoming a worse unique identifier. This could quite literally be the beginning of the end for open rates.

That's two legs of the stool gone wobbly at once.

What To Do About It

Here's the adjusted playbook:

• Stop running segmentation or automation triggers off open data alone. If your re-engagement sequence fires based on "hasn't opened in 90 days," you're potentially suppressing people who are reading every word and rewarding Google's servers for being active.
  ​
  ​
• Make click-to-open rate (CTOR) your new primary metric. CTOR measures the percentage of people who opened and then actually clicked, the 2025 median sits at 6.81% across industries. It strips out most of the prefetch noise and tells you what's actually resonating.
  ​
  ​
• Track clicks, replies, and revenue at the campaign level. These are behaviors. Even a ~1% reply rate is meaningful, it represents real audience investment, not passive consumption. Mailbox providers also treat replies as a strong trust signal, which improves inbox placement over time.
  ​
  ​
• Build a "same person, two addresses" detection protocol. Flag soft duplicates using surname combined with a secondary identifier, phone number, postcode, purchase history. Whoever engaged most recently is your live address. Suppress the other.
  ​
  ​
• Add a direct "update your email address" option inside your emails. It's the cleanest fix available right now. Most readers won't use it. But the ones who do are your most engaged, and you solve the duplicate problem cleanly when it works.
  ​
  ​
• Run a list verification pass now, before this gets worse. Bounce rates above 3% trigger deliverability penalties from Gmail, Outlook, and Yahoo, yet 39% of senders rarely or never conduct list hygiene. With addresses multiplying via aliases, that number is only going up.
  ​
  ​
• Audit what your automations are actually measuring. If any sequence uses opens as a qualifying trigger, a lead score threshold, a sales alert, a re-engagement window, go through each one and ask: what am I actually measuring here? If the answer is "Google's servers," change the trigger.
  ​
  ​

The irony is that email is still the best channel we have. The email marketing industry is projected to grow from $7 billion to over $16 billion by 2030. Email still generates some of the highest ROI of any marketing channel, ~$20-$50 for every dollar spent! but success increasingly depends on technical excellence and real engagement signals, not vanity metrics.

The people who figure that out now will have cleaner lists, tighter segmentation, and better deliverability than the people who find out six months from now when their numbers go sideways and they don't know why.

Metrics can change and often do change all the time and it is important for us to know if metrics are no longer reliable.

Worth knowing.

Fathi

​Unsubscribe | Update your profile | 600 1st Ave, Ste 330 PMB 92768, Seattle, WA 98104-2246

‍