Notification Fatigue

For years notifications were treated as a lever with no cost attached. Send one more push notification and more people opened the app that day, and the increase showed up on the dashboard the next morning. The other side of the trade — the person who got annoyed — was recorded nowhere. When only one side of a decision has a number, that side wins every meeting.

Pagedeck went through exactly this. In a weekly review Ravi reported that doubling push volume the previous week had lifted daily active users (DAU) by 8 percent, and nobody had a number to argue with. Three months later the team shipped a feature that genuinely mattered and announced it the same way. A third of the usual audience opened it. The share of users who still had notifications turned on had quietly fallen below half.

Notification fatigue exists to put that late invoice on the table. The gain from sending more arrives this week and the cost arrives roughly a quarter later, so without a name for the effect it is hard to even claim that the two events share a cause.

Fatigue starts accumulating the moment a single notification is worth less than the effort of checking it. Taken one at a time no individual message looks bad. The problem is that people don't judge notifications one at a time — they judge the sender as a whole. Once "Pagedeck notifications are never anything" settles in someone's head, every later message gets the same treatment regardless of what it says.

It builds in three stages.

1. Swipe past. The notification still arrives, so reach holds steady while only the tap-through rate falls — from 6 percent to 1.5 percent, say. This stage is easy to miss because nothing in the delivery numbers moves.
2. Turn them off. Now nothing arrives at all. Unlike the first stage this doesn't decay gradually; for that user it simply stops.
3. Leave the channel. Unsubscribe from the email, or delete the app. By this point the decision is about the sender rather than any one message, which is why a well-written announcement can't undo it.

A smoke alarm that keeps going off for nothing ends up the same way. You check the kitchen the first couple of times, and after that you pull the battery out. Once the battery is out, a real fire makes no sound at all.

The hard part is that stage two almost never reverses. Users rarely walk into settings on their own to switch notifications back on, and the channel you would have used to ask them is the one that just closed. Notification fatigue is less a metric that moves up and down than a number that only ever moves one way. Reach lost this quarter does not come back next quarter by sending less.

The phrase comes up first when a team that uses messaging for growth sits down to interpret a result. There are four common places.

1. Reading an experiment. Run an A/B test for two weeks with higher push frequency on one side, and the higher-frequency side usually wins, because the bill has not arrived yet. Messaging experiments therefore run longer than usual, or carry a separate line that must not be crossed regardless of who wins. Opt-in rate and unsubscribe rate — the share of recipients who end their subscription after a message — are the two lines teams use most.
2. Cutting what goes out. Batch a day's worth into one morning digest, hold anything overnight, or cap how many messages one person can receive in a week. All three reduce volume, but they feel different on the receiving end.
3. Moving to a quieter channel. Anything that doesn't need to light up a device goes into an in-app message instead — a notice that appears only once someone is already inside the product, so it never interrupts the people who stayed away.
4. Doubting a retention number. Retention measures how reliably existing users come back, and it reads high when messaging is doing the returning. Stop the external trigger and the visits stop with it. Splitting out the users who have notifications off and reading their retention separately shows whether the product itself is holding anyone.

"If we add this push too, we're heading for notification fatigue. That's four this week already."

The objection is about the week's total volume, not about whether this particular message is useful.

"Opt-in rate is down 12 percent over two months. I'd read that as notification fatigue."

The audience for every future message is shrinking, which is a bigger problem than any single campaign.

"If fatigue is the worry, let's batch the day and send once in the morning."

Keep the content, cut the number of interruptions. It is the usual compromise when a team is unwilling to drop messages outright but agrees the pace is too high.

At Pagedeck the term surfaces mostly when a growth experiment is up for approval. When Ravi brings a proposal to push people twice a week about documents they've left untouched, Mia's first question isn't about lift — it's how many messages the same user is already getting that week. If three are scheduled, the new test waits until next week or takes the slot of an existing message. The week's messaging is handled like a fixed budget rather than an open tap.

Theo keeps the numbers for that argument on a dashboard. One panel shows the share of users with notifications enabled, split by signup month. When that share drops noticeably in a given week, the team works backwards to find what went out.

The most common misreading is that fatigue is purely a volume problem, solved by sending less. The word points at quantity, so the assumption is natural. What actually drives people to the settings screen, though, is relevance. One message that has nothing to do with you does more damage than ten comment alerts on your own documents. Halve the volume without changing who receives what and you end up with the same small audience, now hearing about half as much.

The second mistake is filing users who turned notifications off as users who already left. They disappear from the messaging numbers, so it looks that way. Many of them are still working in the product every day — you have simply lost the ability to reach them from outside. Miss that distinction and the next pricing change or feature removal reaches nobody, and the first you hear of it is a cancellation request. Counting that group separately, rather than folding it into churn, is what keeps the mistake visible.