- Grow 📧 Monetize
- Posts
- 📩 Open rates miss the AI reader
You spent the morning on your newsletter.
Picked the stories. Worked the opening. Gave the sponsor a good spot. Rewrote the subject line three times.
Then a machine read it and wrote its own introduction.
That might be the version your subscriber uses to decide whether you are worth their time.
I have been chewing on this middleman for a couple of months now, and I wanted a clearer answer to what it actually means for publishers.
Does AI change whether we reach the inbox?
Does it give readers enough to skip the click?
And what happens to the sponsor paying to be seen?
So I went through the provider documentation and some of the testing behind the claims.
In partnership with
Blu Dot surpasses 2,000% ROAS with self-serve CTV ads
Home furniture brand Blu Dot blew up on CTV with help from Roku Ads Manager. Here’s how:
After a test campaign reached 211,000 households and achieved 1,010% ROAS, the brand went all in to promote its annual sales event. It removed age and income constraints to expand reach and shifted budget to custom audiences and retargeting, where intent was strongest.
The results speak for themselves. As Blu Dot increased their investment by 10x, ROAS jumped to 2,308% and more page-view conversions surpassed 50,000.
“For CTV campaigns, Roku has been a top performer,” said Claire Folkestad, Paid Media Strategist, Blu Dot. “Comping to our other platforms, we have seen really strong ROAS… and highly efficient CPMs, lower than any other CTV partner we've worked with.”
Using Roku Ads Manager, the campaign moved from a pilot to a permanent performance engine for the brand.
*Ad
Three things to take away:
A machine can shape your first impression. The summary may land before your preview does.
A summary, a priority label, and a spam decision are different things. Lumping them together leads to bad fixes.
Your reporting will not tell you everything. Some reading looks like nothing, and some nothing looks like reading.
The second reader is here.
Understand what it does before you rebuild anything around it.
The preview you wrote might not survive
Most publishers treat the subject line and preview text as a pair.
The subject creates interest. The preview gives one more reason to open.
That second half can change now.
Apple Mail can show an AI summary instead of your preview. Apple documents this directly, and users can switch back to the message's opening lines. Availability depends on device, language, region, and settings.
Outlook can replace the preview too. With Copilot's Prioritize feature on, eligible incoming messages get a priority assessment and a short summary. Microsoft says messages that require action tend to get higher priority.
Yahoo makes the decision especially fast. Its Catch Up feature shows summaries of unread messages in the mobile app, with options to delete or keep. It even celebrates how many emails the reader cleared and how quickly.
Your newsletter could end up as a few lines in a cleanup session.
Which makes those few lines worth checking.
Gmail is doing several things at once
People talk about Gmail's changes like they are one feature hitting every reader the same way.
They are not.
Conversation summaries help people catch up on long threads after they open them. That documentation does not show Gmail rewriting the inbox preview of every standalone newsletter.
The newer AI Inbox surfaces suggested tasks and topics to catch up on. It is still in beta, limited to eligible plans, US users, and English. Google says it pulls from Primary and leaves out Promotions and Social.
Separately, the "most relevant" sorting in Promotions weighs the senders and brands a recipient engages with. Readers can switch it to most recent.
For a publisher, those are three different questions.
Is the newsletter being summarized?
Is it showing up in the view the reader actually checks?
Is the reader choosing to interact with it?
An open rate will not answer all three.
And the address on your list does not tell you which app someone reads in. A Gmail subscriber could be reading through Apple Mail.
A bad summary is not a spam problem
This is where I would separate what we know from what sounds convincing.
The documentation shows these features summarize messages, assess priority, and change presentation.
It does not show that making your newsletter easier to summarize improves where it lands.
AI already plays a role in spam filtering. Google says so itself.
But reader-facing summaries are not a published shortcut to better deliverability.
Same caution with falling clicks.
Maybe a summary gave readers what they needed.
Maybe the stories were weaker.
Maybe your audience shifted, your frequency went up, or your bot filtering started stripping out activity you used to count.
Each of those leads to a different decision.
"AI did it" is not enough to change how you send.
Before you blame the middleman, look at what it actually showed your reader.
Stop typing what you could say in 10 seconds.
Wispr Flow turns your voice into clean, professional text inside any app. Emails, Slack, client updates: speak once, send without editing. 4x faster than typing. Works on Mac, Windows, iPhone, and Android.
*Ad
Make your priorities impossible to miss
Twilio and Inbox Monster tested different versions of one email to see how Apple summarized it.
In that test, the subject line and live text near the top shaped the summary. Text inside images and image alt text did not carry the same weight.
That is directional evidence, not a law for every newsletter and every inbox.
My takeaway: make the editorial priorities of each issue easy to read.
Give the subject line a complete thought. It should still make sense if your carefully written preview disappears.
Use the opening to say why this issue matters. Lead with something more useful than housekeeping.
Write descriptive story headings. A roundup can cover a lot of ground. Each story should make sense without leaning on an image or a vague teaser.
Keep essential information in live text. That includes the main story and the sponsor's message.
Then test your own template. See what survives before deciding what needs to change.
What does the reader get that a recap leaves out?
A newsletter that summarizes the news is going to have a different relationship with AI summaries than one built on original reporting or analysis.
That does not make a good digest worthless.
Saving someone time is a perfectly good reason to exist.
But it does raise a question.
What does your reader get from you that a shorter recap leaves out?
Maybe it is your judgment about which stories matter.
Maybe it is an original interview, a sharp comparison, a strong argument, or the way you explain what happens next.
That is where I would keep investing.
I would not respond by making headlines more mysterious or burying the point further down. Readers still need a reason to trust you and come back.
If you depend on website traffic, give people a clear reason to keep going past the summary.
If the full experience lives inside the email, success might include someone getting value without ever clicking out.
Your business model decides what you should measure.
Your sponsor is in this too
There is one more person affected by the machine's version of your newsletter.
The advertiser.
A sponsor can have a prominent spot in your layout and still get almost no attention in a summary.
Putting the sponsor's name and message in live text gives it a chance to show up. It does not guarantee it.
So when you review a sponsored issue, check three things.
Does the summary mention the sponsor?
Does it describe the offer accurately?
Does it confuse the ad with your editorial coverage?
Keep the sponsorship clearly labeled, and look at what the reader actually sees before you promise anyone visibility.
Be careful with reporting too.
A delivered email, a recorded open, a sponsor click, and a qualified lead all tell you different things. Be clear with advertisers about which one you are reporting.
See what's really happening in every deal
Aligned shows you what buyers are doing between meetings inside your most important deals then surfaces the risks, openings, and next steps. No more guessing where a deal stands or finding out a champion went quiet too late. Just real visibility, so you can act before it's too late.
*Ad
The reader might never open it
Google now has an official Gmail MCP server in Developer Preview.
It lets connected AI applications search threads, retrieve messages, manage labels, and create drafts.
I already use MCP on the sending side to help build segments.
Now picture an agent working for the person receiving your newsletter.
Here is my inference: an agent could pull a message's text and use it to answer a question without ever loading the tracking pixel.
That retrieval alone would not register as an open.
Your reporting could miss a real interaction.
Meanwhile, automated systems that load images or follow links can create activity with no human reading or clicking. Omeda documents this problem with non-human interactions and click filtering.
So we have activity that looks like readership but is not.
And possible readership that does not look like activity.
That is why I want several signals:
Clicks with bot filtering applied.
Replies.
Meaningful website activity.
Paid subscriptions and renewals.
Advertiser outcomes, where you can get them.
Use whatever fits your publication, and keep looking for proof that people value what you send.
One warning.
Do not turn the idea of invisible readership into an excuse to keep mailing everyone forever.
An unmeasured interaction is possible.
It is not proof that a particular subscriber is engaged.
Your move this week
Take your next issue and test it in the inboxes your readers actually use.
Start with Apple Mail on a supported device with Apple Intelligence and summary previews turned on.
If Microsoft 365 matters to your audience, add an Outlook account with Copilot Prioritize available and enabled.
Screenshot the preview and the summary. Then read them as someone who has never heard of your publication.
Do they explain why this issue is worth reading?
Is the main story represented accurately?
If there is a sponsor, how does that message come through?
If you would not approve what you see, revise the subject line, the opening, or the story headings, and test again.
Make it part of reviewing every issue, right next to checking links and layout.
You already did the work of deciding what your readers should know.
It is worth seeing how much of it survives the introduction someone else is writing for you.
Hit reply with a screenshot of what Apple or Outlook wrote for your last issue. I want to see how far off the machine is getting.
GM.