A write-up holds up when a teacher who disagrees with it can still not argue with it — because every sentence points at something that happened. The craft is a pipeline: script what you see in low-inference notes, tag the evidence to components, then let the evidence carry the judgment. Every teacher and class below is invented.
Three rules before the examples
- Quote what you saw. "Students were engaged" is a conclusion. "26 of 28 students wrote for the full seven minutes; two conferred with the teacher" is evidence. Write the second; the reader reaches the first on their own.
- No adjective the evidence didn't earn. If the word could survive being challenged in a post-observation conversation, keep it. If it needs your authority to stand, replace it with what happened.
- Rubric language belongs in the write-up, not the notes. In the room, script times, quotes, and counts. At the desk, connect them to the component's language. Mixing the two in the room slows your scripting and colors what you notice.
Example 1 — a strength (3b, Questioning and Discussion Techniques)
Scripted in the room:
9:12 — T: "Why does the author repeat that line? Turn and talk — be ready to share your partner's idea." All pairs talking within 10 seconds. 9:14 — T pulls a name stick: "M., what did your partner say?" 9:15 — T: "Who can push back on that?" Three hands; T waits, two more. 9:17 — Student to student, unprompted: "But that only works if the narrator is lying."
Tags to: 3b — questioning distributed across the class, students responding to each other.
The write-up paragraph:
Questioning distributed the thinking across the room rather than collecting it at the front. After the central question at 9:12, students rehearsed ideas in pairs before anyone answered publicly, the random call at 9:14 held every pair accountable for that rehearsal, and the invitation to "push back" produced student-to-student argument — including an unprompted challenge at 9:17 ("that only works if the narrator is lying"). Discussion technique, not teacher charisma, produced the engagement.
Example 2 — a growth area (2d, Managing Student Behavior)
The evidence carries the judgment; the writer never editorializes.
Scripted in the room:
10:31 — Two students at the back table trade a phone; T mid-sentence, no response. 10:34 — Same table, louder; T: "Guys." Talk resumes within a minute. 10:39 — T stops the lesson: "I'll wait." 40 seconds of silence, then the same table resumes. No named expectation stated during the period; no consequence referenced.
Tags to: 2d — responses to behavior, clarity of expectations.
The write-up paragraph:
Responses to off-task behavior at the back table followed a pattern of escalating pauses without a named expectation: a non-response at 10:31, "Guys" at 10:34, and a 40-second whole-class wait at 10:39, after which the same conversation resumed. The behavior cost instructional time three times in eight minutes, and at no point in the period was an expectation stated or a routine invoked. In our conversation we agreed on one next step: name the expectation once, to the table, with the class working — not to the room, with the class waiting.
Notice what the paragraph does not say: "poor classroom management," "struggled," "chaotic." The timeline is harder to dispute than any of those words, and kinder.
Example 3 — outside the classroom (4e, Growing and Developing Professionally)
The out-of-classroom components — preparation and professionalism (P&P) — are cited on a write-up from evidence noted in the 15 school days before the observation, not from memory of the whole year.
Source: NYCDOE Advance Policy Summary — evidence "observed within 15 school days"; see Every Advance deadline for 2026–27.
Noted, dated twelve school days before the visit:
5/6 — R. brought student work samples to the grade team unprompted and asked the team to score them against the rubric; revised the task that afternoon and shared the revision with the team.
Tags to: 4e — Growing and Developing Professionally.
The write-up line:
On 5/6, R. brought student work to the grade team for calibration unprompted, then revised and re-shared the task the same afternoon — feedback sought, received, and acted on within one day.
The pipeline, named
Script low-inference in the room. Tag each piece of evidence to its component at the desk. Then write each paragraph as: what happened, with times and quotes; what component language it speaks to; what was agreed next. However you run that pipeline — paper, a document, or WellNoted, which keeps the evidence tagged and builds the write-up from exactly what you captured — the standard is the same: nothing in the write-up that wasn't in the room.