Most Agile writing assumes a team big enough to need a dedicated Scrum Master, a product owner distinct from the people doing the work, and a room full of people to run a standup for, none of which describes us at all. We tried adopting a stripped-down version anyway, mostly because the alternative, no process at all beyond a shared to-do list and vague email threads with clients, had started causing real problems, work getting dropped, client expectations drifting out of sync with what we'd actually committed to.
What stuck: two-week sprints with a short list of what we're committing to, written down somewhere both of us can see it rather than living in whichever person's head happened to take the client call. A shared board, we're using a simple kanban-style tool rather than anything Scrum-specific, so client work in progress is visible rather than living only in someone's head, which turned out to matter more than we expected the first time one of us was out sick for a few days and the other could just look at the board and know exactly what was mid-flight. A fifteen-minute check-in most mornings even when we're in the same room, which felt slightly silly at first, we're sitting six feet apart, why are we doing a formal standup, but having a fixed daily moment to say out loud what we're each working on catches misalignment days earlier than it would surface otherwise, before a client conversation makes the gap embarrassingly obvious.
What didn't stick: formal retrospectives with all the ceremony, sticky notes on a wall, a facilitator, timeboxed rounds of feedback, which for two people talking daily anyway just felt like a meeting for the sake of having a meeting. We tried it properly, twice, following an online guide almost to the letter the first time, and both times the actual useful content could have been, and eventually was, said in about five minutes over coffee instead of the forty-five the format called for.
We also skipped story points and velocity tracking entirely. Estimating in points relative to each other only pays off once you have enough historical data and enough team members to make the averaging meaningful, and with two people we found we were just translating points back into hours in our heads anyway, adding a layer of abstraction with no actual benefit over estimating in hours or days directly.
Sprint planning itself got simpler than the textbook version too, no formal backlog grooming session, no planning poker, just the two of us looking at what's committed to clients and what's realistically achievable in the next two weeks, and adjusting the list together in ten or fifteen minutes rather than a scheduled hour-long ceremony.
Take the parts that solve a real problem you have, skip the rest. The version of Scrum we're running now barely resembles the framework as written in any of the books we skimmed, but it's solving the actual problems we had, work invisible to one of us, no fixed cadence, without importing a pile of ceremony designed for a team five times our size that would have cost us more time than it saved.