• AI summaries in Gmail, Outlook, and Apple Mail are already live – and they’re changing how people “read” emails
• Open rates are going up (for the wrong reason) while click-through rates are dropping
• Your email structure now affects how you’re filtered, not just how you’re read
• A contents summary at the top of your newsletter is a smart move – for humans and AI
• Front-load your value, keep your paragraphs short, and make your CTAs impossible to miss

Something has changed in the way people receive your emails, and if you send a regular newsletter, it’s worth knowing about. You need to structure your emails differently because AI is now reading them.

It sounds a bit abstract, but it’s actually true – and it’s already happening in the inboxes of the people on your list.

Let me explain what’s going on, what it means practically, and what I’m doing differently as a result.

So what’s actually happening in inboxes right now?

The big three email clients – Gmail, Outlook, and Apple Mail – have all rolled out AI features in the last few months that change how emails are presented before a real life person even looks at them.

Gmail now has what it calls AI Overviews, which summarise email threads and surface key information without the reader having to dig through the message. There’s also a new AI Inbox that filters out emails it considers low-value, effectively making them invisible even when they’ve technically been delivered.

Outlook has Copilot-powered summarisation built in for most Microsoft 365 users now. And Apple Intelligence on newer devices gives Apple Mail users priority message summaries automatically.

So when someone on your newsletter list gets your email, there’s a reasonable chance they’re seeing an AI-generated summary of it before they read a single word you actually wrote.

Does this mean open rates are lying to me?

Kind of, yes. (But lots of spam detection link checkers have always made open rates lie.)

Because Gmail’s AI needs to read the email to generate a summary, it appears to auto-open messages in the process. That inflates your open rate. Meanwhile, click-through rates are dropping, because if someone reads a decent AI summary and feels like they’ve got the gist, they might not click through to the full content at all.

So your opens go up, your clicks go down, and neither figure means quite what it used to. It’s worth keeping that in mind if you’re looking at your newsletter stats and wondering why the pattern has shifted.

Is the AI also deciding whether my email is worth showing?

Yes, this is the bigger thing. It’s not just summarisation – AI is now acting as a filter.

Gmail’s AI Inbox deprioritises emails it considers generic or low-value. These don’t necessarily go to spam; they just get buried where people are much less likely to see them. And unlike a spam bounce or a complaint, you won’t get any signal that it’s happened. The email landed. It’s just not very visible to the user.

SUMMARY

The clarity and structure of your email content is now a deliverability signal, not just a readability one. Vague subject lines, heavy padding, dense paragraphs, buried value propositions – all of these are things an AI will interpret as “low relevance” before your subscriber ever makes that call.

Should I be putting a contents summary at the top of my newsletter?

I’ve been thinking about this for my own weekly email, and the answer I’ve landed on is: yes, and here’s why.

An anchor-link contents section at the top of a newsletter does three things at once.

For your human readers, it means they can scan what’s in this week’s issue and jump straight to what interests them – which is what a lot of people do anyway, even without anchors.

For AI summarisation, it’s essentially handing the algorithm a pre-written outline of your email. A clear, structured summary at the top means the AI is more likely to represent your content accurately when it generates its snippet, rather than pulling a random sentence from the middle that doesn’t reflect what you actually wanted to lead with.

And for filtering, it signals that the email is organised and substantive, not a wall of text or a vague pitch.

It takes a bit of extra effort to set up, but I’m gambling on it being worth it.

What does good email structure actually look like now?

None of this requires a total overhaul. Most of it is just disciplined writing. But here are the guidelines I’m going with:

The first 100–200 characters matter most. That’s the window an AI uses to generate its initial read on what your email is about, and it’s also what shows as a preview in most inboxes. Don’t waste it on a greeting or a “hope you’re well.”

I really keep my emails human, with an intro about my weekend – but I’m moving that to near the end now.

Short sections beat long paragraphs. A dense 400-word section covering three different points is hard for an AI to extract a clean summary from, and it’s hard for a human to skim. Split it up.

Put your most important content and CTA near the top. Not buried in section four. If your email has a main point, lead with it.

Make your subject line do actual work. Not clever for the sake of it. What is this email actually about? Say that.

Anchor links at the top if you have multiple sections. Especially for longer newsletters.

None of this is really new advice. Clear structure, front-loaded value and short paragraphs is largely good writing, but what’s changed is that the consequences of ignoring it are now more immediate. You’re not just at risk of people skim-reading and losing interest; you’re at risk of the AI misrepresenting your email, or deciding it’s not worth showing at all.

Does this affect spam filtering too?

It does, though the spam side of things is a slightly separate conversation. Gmail has been using machine learning for spam filtering for years, and it’s been getting more sophisticated. What’s newer is that it’s moved beyond the technical signals (authentication, sender reputation, bounce rates) and is increasingly weighing content quality and engagement as part of the picture.

If people consistently open your emails and don’t engage with them – or if the AI reads your content as thin or misleading – that can feed back into how your future emails are treated. It’s another reason to make sure what you’re sending is genuinely worth reading, not just technically compliant.

The authentication side of things (SPF, DKIM, DMARC – the things that prove the email is actually from you) still matters and is non-negotiable if you’re sending any volume.

In summary

Your subscribers’ inboxes now have AI in the middle, and that AI is reading your emails, deciding how to summarise them, and in some cases deciding how prominently to show them.

The good news is that adapting to it doesn’t require doing anything weird or technical. It mostly requires writing clearly, structuring your content well, and respecting your readers’ time – which are things you should have been doing anyway.

Add the contents section at the top of your newsletter. Front-load your value. Keep your sections short and focused. Make it easy for both the AI and the human to understand what you’re saying and why it matters. And that’s it, really!