We Turned Off Work Notifications After 6pm for a Quarter. Nothing Broke.

Last January my manager asked our team a question I did not expect: what would break if nobody got a chat notification from work after 6pm? Our team is eight backend engineers. We have a real on-call rotation with a pager tool. Chat was never supposed to be how emergencies reached us. And yet most of us checked it every evening.

So we ran an experiment. For one quarter, everyone set their work chat to silence notifications from 6pm to 8am, and on weekends. The on-call engineer kept pager alerts on, as always. We agreed to keep a shared log of anything that went wrong because of the change.

The setup was boring on purpose

We did not install anything new. Everyone used the built-in schedule setting in our chat app. We posted a short note in our team channel and in two adjacent teams' channels: "We are not seeing messages after 6pm. If something is urgent, page the on-call."

We also agreed on one rule for ourselves. If you did open chat after hours, you were not allowed to reply to anything that was not an incident. You could read, but you had to wait until morning to answer.

What the log said

After 13 weeks, the log had six entries. Six. Here they are, roughly:

• A product manager asked a question at 9pm that was answered at 8:30am. She said it was fine.
• Someone in another time zone needed a code review approval. They waited until morning, lost about half a day.
• A deploy failed at 7pm. The on-call got paged by the deploy system, not by chat. No change.
• Two people on our own team realized they had been answering each other's evening messages out of habit, not need.
• One genuinely urgent message went to chat instead of the pager. It was a customer data question from support. It sat for 40 minutes before the support lead paged us. That was the only real miss.

We fixed the last one by adding a "page backend" button to the support team's tools.

What we noticed in ourselves

This part is less measurable. In our end-of-quarter retro, five of eight people said they slept better. Three said they had not realized how often they looked at their phone in the evening until it stopped buzzing. One person said she actually felt more anxious for the first two weeks, worried she was missing something. By week four that faded.

Our deploy frequency did not change. Our incident count did not change. Our sprint completion rate stayed within a couple of points of the previous quarter.

Why it worked for us

I do not think this works everywhere. It worked for us because we already had a real paging system for real emergencies. Chat had quietly turned into a second, worse paging system, one with no rotation and no escalation.

Once we named that, the fix was obvious. We kept the schedule after the experiment ended. Two other teams in our department have since copied it. The only thing anyone has asked for is clearer docs on how to page us, which is a much better problem to have.