Three months into the shift to remote work, engineering teams had split clearly. Some adapted smoothly. Others struggled hard. What separated them was interesting and practical.
Documentation as culture
The teams that adapted best to remote work had strong documentation cultures before the transition. Architecture decision records, runbooks, onboarding guides, and API documentation that existed before March 2020 reduced the knowledge transfer burden of going distributed. Teams that relied on implicit knowledge transfer through proximity faced the documentation debt immediately. The lesson: documentation is not a documentation-phase activity; it is a continuous practice.
Meeting reduction as engineering respect
Effective remote engineering organisations reduced meeting load, not increased it. The anti-pattern: replacing office collaboration with back-to-back Zoom meetings that leave no time for engineering work. The effective pattern: move status updates to written async formats (standup in Slack, weekly written updates), reserve synchronous time for decisions that genuinely require real-time discussion, and protect deep work blocks in calendars explicitly.
Video-first for complex technical discussion
Written-async communication handles status, decisions, and documentation well. Video-first is better for complex technical discussion where whiteboarding, real-time reaction, and collaborative problem-solving are needed. Architecture reviews, complex debugging sessions, and 1:1s for sensitive conversations benefit from synchronous video even in an async-first culture.
Team rituals that survived the transition
The team rituals that translated well to remote: weekly team retrospectives (structured, with asynchronous pre-work), team 1:1s on a consistent schedule, virtual coffee chats for informal connection, and asynchronous recognition (shoutouts in team channels). The rituals that did not translate: unstructured office hangouts, impromptu whiteboard sessions, and hallway conversations that surface blockers early. Replacing the latter requires proactive manager check-ins and low-friction async blockers channels.