4 ms·
Disclaimer: I work for Webex Teams, another collab tool. I think a lot of this depends a lot more on the expectations of the whole group. This needs to be defi
by athenot 8y ago
Disclaimer: I work for Webex Teams, another collab tool.
I think a lot of this depends a lot more on the expectations of the whole group. This needs to be defined at an org level and well understood by everyone. It's not so much the tool as how it's used.
There is now a continuum of interactions:
- FYIs that are put out in a topical chat. You get to those when you have a moment, when you go through your unread rooms to catch up. Same for topics that you are only peripherally following. Time is not of the essence there. Zero interruption here and no expectation for you to be always listening.
- Topical chats where you are @-mentionned. This is where a conversation that you're not actively following suddenly took a turn where your input is sollicited. But since you see the topic of the room, you are in control about whether it's worth an interruption. The group's expectation is that the person that was mentionned will join the conversation IF they are not in the middle of something (assuming the topic is not alarmist).
- New rooms that got created around a topic or you getting added to an existing room. Basically it's similar to the above, where you're getting dragged into a conversation.
- 1:1 chats with specific requests. Again, it depends who it's from, you triage based on origin. But the expectation is already managed at our group level.
In all these modes, you can chose to accept the interruption or defer it. The deeper your work, the more likely you can defer it. We can operate like this because we have the nuclear option we can always use when it's actually needed: PagerDuty. This brings a few benefits: the person paging a team or another individual is aware that they are putting a burden of urgency and there's a recorded audit log of that page. So it's not used lightly.
Once you've accepted the interruption though, you still get to control its modality.
- Very often, a "hey, quick question" can turn into something WAY larger than the requester could imagine. That's when we point them to creating a Jira ticket, effectively punting a synchronous interruption into an asynchronous unit of work. Or you can tell the requester to schedule a meeting to work more deeply on that. (Related topic: ALWAYS have at least one day where it is forbidden to schedule meetings, accros the whole org.)
- If the request is about an urgent matter, we just do an instant video meeting within Webex Teams. That converts long drawn out chats into quick visual meetings where the nuances are better expressed. Because in a long chat, the mind is already interrupted, so having to wait for typing and confirming nuance ends up taking a lot of time.