Picture the standard observation cycle. An administrator schedules a time, arrives at the classroom, spends thirty to sixty minutes scripting what they see, then spends another hour formatting those notes into a rubric-scored document. The teacher receives feedback days later. The process resets the following year.
Most people in K-12 education have participated in some version of this. Most have the same quiet concern: is this actually producing anything?
The honest answer: the formal observation produces a document, a score, and a completed compliance requirement. What it often fails to produce is an accurate picture of what a teacher actually does over time.
The reliability problem nobody talks about
A single observation visit — even a well-conducted one — captures a small, non-random slice of a teacher's practice. It reflects that day's lesson, that day's class, and the dynamic created by an observer in the room. It may or may not reflect what that teacher does on every other Tuesday.
The formal observation hasn't survived because it's the best tool for the job. It's survived because it was the only tool that could be done consistently at district scale, with the documentation systems that existed.
That constraint is no longer fixed. The question is whether the field is willing to examine what it's actually been measuring.
What evidence actually looks like
Great teaching leaves evidence everywhere, all the time — in the goals a teacher sets in August, in the coaching conversation where a principal and a third-year teacher work through a questioning technique for six weeks, in the video a teacher records of her own lesson and shares with her coach with a timestamp.
None of that is typically visible to the evaluator. Coaching happens in one system. Goals live in another. Video reflections exist in a personal workspace no administrator can see. The formal observation arrives as a fresh event, evaluating the teacher as if none of that prior work existed.
This is not a technology problem. It's a design problem — and it has a design answer.
The question worth asking
If evaluation's goal is to produce accurate, growth-oriented assessments — not merely satisfy a compliance requirement — then the question changes. It stops being: how do we make the observation visit more efficient? It becomes: how do we build a complete, honest picture of a teacher's practice from everything that's already happening?
When you reframe it that way, the formal observation stops being the thing. It becomes one moment in a continuous flow of evidence — contributed by the teacher, the observer, and the tools they both use.
The formal visit still matters. The rubric still matters. But both become more defensible when connected to a longitudinal record of what's actually been happening in that classroom.
What this requires in practice
The evidence-first model requires three things traditional systems have never provided together. First, both teacher and observer must be able to contribute evidence — evaluation shouldn't happen only to a teacher. Second, the privacy architecture must be one teachers genuinely trust: the moment a teacher suspects coaching conversations will appear in formal evaluation, candid coaching stops. Teachers must control what evidence moves between spaces, and that control must be real and visible. Third, there must be a way to find patterns across evidence over time — not to score teachers automatically, but to surface the connection between a lesson in October and a goal set in August when a rubric indicator is evaluated in March.
A different place to land
The documentation burden of traditional observation is being automated away. What replaces it matters. A faster version of the same model — observer visits, notes processed by AI, scores generated more quickly — doesn't address the reliability problem. The snapshot is still a snapshot.
A genuine evidence portfolio, built collaboratively across a cycle, connected to coaching and goal-setting work, and interpreted by a human evaluator with full context — that's something different. Not faster forms. A more honest picture.
Sibme's Observations module launches July 2026, bringing formal and informal observation into the same platform as coaching, goals, and video — within a trust architecture designed for teacher agency. Learn more at sibme.com/observations.
.png)