Email Has Been 'Dying' Since 1995: What Obituaries Get Wrong

Email died again last month.

This time the killer was a Gmail change to how tracking pixels are handled, which made open rates read lower across every email platform at once. Substack told its creators the obvious thing after investigating: subscriber behavior hadn't changed, delivery hadn't changed, clicks hadn't changed — a measurement broke, not the medium. But the headline wrote itself anyway, because "email is dead" is the most reliable headline in technology. It has run, in some form, almost every year since 1995.

The funny part is the chart underneath the obituaries. Every time someone announces email's death, more people are using email than the last time someone announced it. The medium that has been dying for thirty years now has around 4.7 billion users and moves more than 370 billion messages a day. It is the single most-used application on the internet, and it has been pronounced terminal at every stage of becoming that.

This is worth understanding for a reason that has nothing to do with email specifically. The same obituary template is being printed right now about a different channel — the open web, the website, the thing you actually own — and the people printing it are making the exact mistake their predecessors made about email. If you know why the email predictions were wrong, you know how to read the website predictions.

A brief history of email's death

Here is the timeline, with the cause of death and the autopsy result.

Read the right-hand column top to bottom and a pattern falls out. Not one of these predictions was stupid at the time. Spam really was an existential nuisance in 1999. Slack really did eat a chunk of internal company email in 2014. The AI assistants of 2023 really can read and answer your mail. Each prediction correctly identified a real new thing. Each one then made the same leap: because this new thing is arriving, the old thing is leaving.

That leap is the error, and it is the same error every time.

The mistake under every obituary

Channels are almost always additive, not substitutive. A new channel shows up, takes a slice of the jobs the old channel was doing, and the old channel keeps the rest — usually while growing in absolute terms, because the whole pie is growing too.

Slack didn't replace email; it replaced the worst use of email, the twelve-person internal thread that should have been a chat. Email kept everything Slack is bad at: reaching someone who isn't in your workspace, sending a receipt, contacting a customer, leaving a durable record. Instant messaging didn't replace email in 2002 for the identical reason. Social networks didn't, in 2007 — they ran on email, using it as the verification layer and the re-engagement channel. The new channel almost never kills the old one. It carves off a use case and the headline rounds that up to a funeral.

The people writing the obituaries confuse a shift in the mix with a death of the channel. The mix shifts constantly. Less of your communication is email than it was in 2009; that's true and unremarkable. It does not mean email is dying any more than the existence of restaurants means home cooking is dying.

But there is a deeper question the additive-vs-substitutive frame doesn't answer, and it's the one that actually matters: some channels do die. Vine died. Google Reader died. Organic reach on a Facebook Page died. So the real question isn't "why does email survive every prediction" — it's "what separates the channels that genuinely die from the ones that don't." Because that distinction is the whole game.

The variable that actually predicts survival

Line up the channels that survived against the channels that died, and one variable sorts them cleanly: does anyone own it.

Email has no owner. SMTP is an open protocol, published in an RFC, implemented by everyone, controlled by no one. There is no company that can deprecate it, no boardroom that can shut it down, no algorithm that decides whose messages reach whose inbox, no quarterly priority that can throttle your reach to juice engagement. Your list of email addresses is yours — a flat file you can export and carry anywhere. The relationship between you and the person at that address has no intermediary who can revoke it. That is why thirty years of new channels carved off use cases without ever landing the kill: there was nothing to switch off.

Now look at the channels that actually died.

Every channel on that list had an owner with a switch. The death was never the public losing interest first — it was a company deciding, for its own reasons, to change the rules or pull the plug, and the public having no recourse because they never owned anything. You didn't own your Google Reader subscriptions. You don't own your Facebook Page's reach; you rent it, and the landlord raised the rent to roughly one to two percent of your own followers so you'd buy ads to reach the rest. You don't own your follower count on any platform. You're a tenant, and tenancy can be revoked.

This is the line that splits living channels from dead ones. Owned, open channels don't die, because death requires an owner with a motive and a switch, and they have neither. Rented channels die routinely, because the owner's incentives and yours are different and theirs win.

The half of the obituary that's actually right

So the "email is dead" chart is wrong about email — and it accidentally describes something completely true.

The thing the doomsayers keep sensing, and keep mislabeling, is real channel fragility. They feel a channel slipping and reach for the nearest one, email, because it's the famous one. But the channel that's actually slipping under their feet is never email. It's whatever rented platform they built on. Reach you rent does disappear overnight. Ask anyone who had a million-follow Facebook Page in 2013 and a thousand-view post in 2016, having done nothing different except exist on the wrong side of an algorithm change. Ask the developers whose entire business was a Twitter app the morning the API closed.

"Email is dead, so move your audience to our app" — the actual subtext of most email obituaries, including the Substack-vs-Gmail round — gets the risk exactly backwards. Email is the channel that can't be taken from you. The app is the channel that can. The pitch to abandon the thing you own for the thing they own is the single worst trade in distribution, and it's sold to you as forward-thinking every time.

Email isn't dying. But your reach on someone else's app can vanish in a product update you'll read about after it's already happened.

The channel getting the email treatment right now

Here's why this thirty-year-old pattern matters in 2026: the obituary has moved to a new target. The headline now is some version of "AI will kill the website." AI search answers the question without the click. Agents browse on the user's behalf. Why would anyone visit your site when the model just tells them the answer?

Structurally, this is "Slack will replace email" with the nouns swapped. It correctly identifies a real new thing — AI-mediated discovery is genuinely taking a slice of how people reach information — and then makes the same leap to a funeral. And it will be wrong in the same way, because the website has the same property email has: it's an open, owned channel with no switch for anyone else to flip. The mix is shifting hard — search referrals are down sharply and AI assistants send almost no traffic back, which is a real and painful change to absorb. But "less traffic from this one source" is a shift in the mix, not the death of the destination, exactly like "less of my communication is email now" was in 2009.

What actually changes is the same thing that always changes: the traffic gets smaller and higher-intent, and the channel you own becomes more valuable per visit, not less. The website isn't dying. Its role is concentrating — from a place lots of low-intent people browse into the place where the people who chose to come, or whom an AI agent routed to you deliberately, actually convert. That's a channel getting stronger per visitor while a rented one (search referral volume) gets weaker. The obituary writers will call it death. It's the email story again.

What's actually worth owning

If the lesson of email is "build on channels nobody else can switch off," the practical version for any business in 2026 is short. There are two channels you genuinely own, and the entire point is to route value through them and treat everything else as rented traffic that feeds them.

The first is the one this whole post is about: an email list. A flat file of people who agreed to hear from you, portable across any provider, with no algorithm rationing your reach. The recent Gmail and AI-inbox changes don't undermine that — though they do change how you have to write the email so the machine reading it first surfaces your point, and they make authentication a hard pass/fail rather than a nicety. The channel is fine. The craft moved.

The second is your website and the conversation that happens on it. When a visitor lands on a page you own and starts a chat, there is no intermediary deciding how much of your answer they're allowed to see — no feed algorithm, no priority filter, no API you don't control sitting between your reply and their screen. It's the same property that kept email alive: a direct line with no owner in the middle. This is the cleanest reason an on-site chatbot has held up while so many "engagement" channels came and went — a chatbot embedded on your own domain is an owned channel by construction, the real-time complement to the email list, and it answers the higher-intent visitor that AI-mediated discovery is now sending you instead of the casual browser it used to. Where live chat ties you to staffed hours, the automated version keeps the conversation on a surface you control around the clock. You don't have to oversell that to see it: it's simply on the right side of the owned-versus-rented line, which is the only side that survives the next obituary.

Everything else — your reach on a social platform, your ranking in a search engine, your placement in an app store, your visibility in an AI assistant's answers — is rented. Useful, often essential, worth optimizing hard. But rented. The discipline is to use rented reach to drive people onto the channels you own, and to never, ever do the trade the obituaries keep selling: abandoning the channel you control for the platform that controls you, because someone declared the owned one dead.

The next obituary is already written

There will be another "email is dead" headline within the year, and then a "the website is dead" one, and eventually "the AI chatbot is dead, agents replaced it." Each will correctly spot a real new channel. Each will make the leap to a funeral. Each will be wrong in the specific, predictable way all of its predecessors were wrong — mistaking a shift in the mix for the death of the medium, and mistaking the fragility of rented reach for the fragility of channels that can't actually be taken away.

The move is not to chase whichever channel is being declared alive this quarter. It's to notice which channels have no owner and no switch — email, your domain, the conversation you host on it — and to build there, while treating every platform that can change its rules on you as exactly what it is: traffic you're renting until the rules change. Email has been dying since 1995 because email is one of the few things online nobody can kill. The channels worth owning are the ones the obituaries can't touch.

Build your chatbot for free →

No credit card required.