Where email marketing actually stands in 2026 – and what 2027 is already doing about it

Let's start with an uncomfortable question: when's the last time you actually trusted your open rate?

I ask because I don't think most of us do anymore, and the data backs that up. Email is still the highest-ROI channel most of us have access to – Litmus has been putting the average return around $36 for every $1 spent for a few years running, and nothing in 2026 suggests that's slipping. But the way we measure whether email is working has become steadily less trustworthy every year since Apple rolled out Mail Privacy Protection, and 2026 added a new wrinkle on top: AI agents are starting to read your emails before your subscribers do.

So I pulled real benchmark data from Klaviyo, Validity, Braze, Mailchimp-adjacent sources, and Data Axle's own research, plus a long scroll through Really Good Emails to see what's actually winning creatively right now. Here's what's true, what's changed, and what I'd actually do about it if I were running your program.

The benchmarks, stripped of the marketing spin

First question: do you actually know what "good" looks like anymore, or are you still comparing yourself to a number from 2021?

Klaviyo's 2026 benchmark report – built from over 183,000 brands – puts the average email campaign open rate at 31% across all industries, with the top 10% of senders reaching 45%. Click rate averages 1.69% for campaigns. That's the headline number everyone quotes. But it buries the more useful finding underneath it: automated flows (welcome series, cart abandonment, post-purchase) are outperforming one-off campaigns by a wide enough margin that treating them the same is actively misleading you. Flows are pulling a 5.58% click rate against 1.69% for campaigns, and generating something like 41% of total email revenue from just 5.3% of total sends.

Here's a table worth screenshotting, because it's the one I keep getting asked about:

Deliverability moved the goalposts in 2026, and a lot of teams haven't noticed

Validity's 2026 Email Deliverability Benchmark Report – built from trillions of inbox data points – puts the global average inbox placement rate at 87.2%, up 3.7 percentage points year over year. That's genuinely good news. It also means roughly one in seven to one in eight marketing emails still never reaches a visible inbox at all, even when your ESP dashboard proudly reports a 99% delivery rate.

Here's the question that trips people up: do you know your inbox placement rate by provider, or just as one blended number?

You should, because the spread is bigger than most people expect. Gmail – the largest mailbox provider globally by a wide margin – is sitting around 89.8% inbox placement in Validity's most recent data. Microsoft's Outlook ecosystem is meaningfully behind that. If your list skews toward Outlook or Hotmail addresses and you're only looking at a blended deliverability number, you're probably underestimating a real problem hiding in one specific provider.

Validity's report also introduced something I think is genuinely useful: a metric they're calling the Disaffection Index – essentially, how fast your program is losing engaged subscribers before they ever hit unsubscribe. It's a leading indicator instead of a lagging one, which is the direction deliverability measurement needed to go. Complaint rate and bounce rate tell you about damage that's already done. Something like a disaffection index tells you it's coming.

What's actually working creatively right now

I spent a chunk of this month scrolling Really Good Emails looking for patterns, and a few things showed up over and over that are worth stealing.

Hyper-minimalism, not more elements

Litmus has been citing roughly nine seconds as the average time someone spends looking at an email. Nine seconds. The winning designs in 2026 are responding to that by cutting, not adding – one message, one image, one clear call to action. The era of five competing links in a footer is fading, and honestly, good riddance.

Dark mode is table stakes, not a nice-to-have

Every major client – Gmail, Outlook, Apple Mail, Android – now handles dark mode, and every one of them handles it slightly differently. The brands doing this well aren't just slapping a dark background on their existing template; they're designing with enough contrast and foreground separation that the email holds up whether the client leaves their colors alone, partially inverts them, or rewrites them entirely.

Brand-forward hero sections, not just a logo

Mike Nelson at Really Good Emails flagged this one directly: more brands are building their opening section around bold typography and big, confident headlines instead of leaning on a small logo in the corner. The idea is simple – if someone only sees the top three inches of your email before they decide whether to keep reading, that space should be doing more work than a logo can do alone.

Interactive elements, especially in B2B

AMP-powered emails that let someone RSVP, fill out a short survey, or browse a product carousel without leaving their inbox are gaining real traction – and not just in retail. B2B teams are using the same interactivity to simplify event registration, collect quick feedback, or remove a step from onboarding. Every one of those is friction removed between "saw the email" and "did the thing."

Plain-text is having a real moment, especially in B2B

This one surprised me less than it probably should have. A growing share of high-performing B2B campaigns are skipping HTML entirely in favor of something that reads like an actual person wrote it – because, increasingly, it feels like spam filters and subscribers both respond better to something that doesn't look like a template.

Quick gut check: if you stripped all the images and buttons out of your last campaign, would the writing alone be good enough to keep someone reading?

The part that actually matters most: AI isn't just personalizing your emails anymore, it's reading them first

This is the shift I think most email marketers are underrating right now, and it's the one I'd genuinely lose sleep over if I were running a large program.

Gemini is already sitting inside Gmail. Microsoft Copilot is already inside Outlook. Both are being positioned to summarize threads, prioritize what a subscriber actually sees, and in limited cases, take action – booking things, replying, filing things away – without a human ever opening your email at all. That's not a hypothetical for 2027. It's already happening in 2026, just unevenly rolled out.

Braze's 2026 Global Customer Engagement Review put a number on the trust gap this creates: 93% of marketing leaders believe AI is helping them understand what customers want, but only 53% of actual consumers agree that brands are getting it right. That's not a small gap. That's most of the industry grading its own homework and getting it wrong by half.

Here's why this actually matters for how you measure success: an AI agent can trigger an "open" event while scanning your email without a human ever reading a word of it. It can also read, understand, and act on your content – book the appointment, forward the offer, add the item to a list – without ever generating an open at all. The metric that's supposed to tell you a human saw your email is becoming less reliable in both directions at once.

So what do you actually measure instead?

Right now, honestly, nobody has a clean answer – the concept of an "agent action rate" (did the agent surface your content or take the action you wanted) doesn't formally exist as a tracked metric yet at most ESPs. But the direction is clear enough to act on: lean harder on downstream signals you already have – replies, conversions, revenue attributed to email – because those require an actual outcome, not just a pixel loading. That's exactly the same shift Wellsent has been telling people to make because of Apple Mail Privacy Protection. Turns out it was good advice for a second, unrelated reason too.

What 2027 probably looks like

I'd hold these a little more loosely than the 2026 data above – they're direction, not certainty – but here's where the signals are pointing.

• Downstream metrics become the default, not the nice-to-have. Replies, conversions, and revenue attribution move from "the metrics we check quarterly" to "the metrics on the dashboard," because open rate will keep getting noisier as more of the inbox is triaged by something other than a human.
• Accessibility becomes a deliverability signal, not just an ethics one. Mailbox providers are already rewarding well-structured, accessible HTML. Expect that to get more explicit, not less, as inbox providers keep leaning on structural signals to separate real senders from junk.
• Agentic AI in marketing will have a real failure wave. Gartner's prediction that more than 40% of agentic AI projects get shelved by 2027 due to unclear ROI or weak guardrails is worth remembering next time a vendor pitches you a fully autonomous campaign engine. The winners will be the boring, well-scoped use cases – not the flashy end-to-end ones.
• Predictive send-time and content personalization becomes normal, not enterprise-only. The infrastructure for this has existed for a while; what's changing is cost and accessibility for mid-market and small teams, not the underlying capability.

What I'd actually do about all of this right now

Not a 40-point strategy deck. Just the handful of things that would move the needle most based on everything above.

1. Stop reporting open rate as your headline metric in any deck, dashboard, or report you send someone else. Lead with click-to-open rate and revenue per recipient instead – they survive both Apple MPP and AI agents scanning your inbox.
2. If you're not running welcome, cart abandonment, or post-purchase flows yet, build those before you build another one-off campaign. The revenue gap between flows and campaigns isn't marginal – it's the difference between a program and a habit.
3. Check your inbox placement rate by provider, not blended. If you don't know where to start, run it through the Inbox Placement Troubleshooter – it'll narrow down whether it's a Gmail-specific, Outlook-specific, or list-quality issue in a few minutes.
4. Rebuild at least one campaign this quarter around one message, one image, one CTA. Nine seconds isn't a lot of runway for anything else.
5. Test your templates in actual dark mode on an actual phone, not just a light-mode preview in your ESP. If your logo disappears or your brand colors invert into something unrecognizable, fix it before your next send, not after someone screenshots it.
6. Start tracking replies and conversions as seriously as you track opens, even if it feels early. The inbox is being triaged by something other than a human more of the time every month – the sooner your reporting reflects that, the less painful the transition will be.

Email isn't going anywhere. It's still the channel with the best ROI, the most durable audience relationship, and the least platform risk of anything in the marketing stack. But the way we measure whether it's working is genuinely changing under our feet, and the teams that adjust their metrics now are going to look a lot smarter in 18 months than the teams still bragging about a 45% open rate that an AI agent generated without a human in the room.

Stephen Taunt