What Should I Write Down? Why That's the Wrong Question During a Classroom Walkthrough

If you've ever stood in a classroom during a walkthrough and felt your attention split between watching the lesson and trying to capture it, you already know the problem this post is about. Admins are quietly being asked to do two cognitively demanding jobs at the same time — observing instruction and triaging what's worth writing down — and one of those jobs is winning attention at the expense of the other. The case I want to make in this post is that "what should I write down?" is exactly the wrong question to be asking during a walkthrough, that the triage happening in your head is costing you the very observations you most want to make, and that there's a meaningfully better way to think about capture now than there was even five years ago.

This is a follow-up to a post I wrote on what administrators should be looking for in the first five minutes of a walkthrough. That post made the case that the highest-signal observations are about the students, the room, the teacher's decisions, and the lesson's history — not just the teacher's surface performance. This post takes the next question seriously: once you know what to look at, what do you do with what you see?

The hidden second job

Every walkthrough has two jobs running in parallel. The visible job is observation — watching a classroom, taking in the learning, forming a professional read on what's happening. The hidden second job is capture — deciding what's worth writing down, what category it fits in, how to phrase it so it'll be useful later, and where it'll eventually go in whatever evaluation framework your district uses.

For most administrators, the second job is operating constantly, in the background, every minute of every walkthrough. And it's not free. Every cognitive cycle spent deciding whether to capture something is a cycle not spent actually observing.

The cost shows up in predictable ways. The administrator who's mentally cross-walking what they're seeing against a rubric is, by definition, not fully present to what they're seeing. The walkthrough that produced great notes was sometimes the one where the observation suffered. The walkthrough that produced great observation often left the admin scrambling to reconstruct what they saw later. Both jobs got done badly because they were competing for the same attention.

This isn't an admin problem. It's a structural problem with how walkthroughs have traditionally been conducted. The capture job has always been there — it just wasn't always being asked of admins in real time.

What triage actually costs you

The triage question — is this worth writing down? — sounds practical. In practice, it's expensive in three specific ways.

First, you lose the moments that matter most. The most consequential things to observe in a classroom are often the subtle ones. The quiet redirect that re-engaged a student without interrupting the lesson. The wait time after a question that gave the slower thinker a chance to find an answer. The way the teacher acknowledged a wrong answer without dismissing the student. These are exactly the moments admins are most likely to triage away in real time, because they don't look like rubric language — they look like teaching. By the time the admin recognizes the moment was significant, it's gone.

Second, you bias toward the obvious. When the admin is asking "what should I write down?" the mind reaches for the things that fit cleanly into a rubric box. Direct instruction. A visible behavior management move. A clear example of differentiation. These are the moments that read as "evaluable" in real time, which means they're the moments most likely to get captured. The result is an evaluation built from the most visible parts of the teaching — which is often not the most important parts.

Third, the triage produces evaluations that feel thin to teachers. A teacher who's been observed knows what they did during the walkthrough. They know the small decisions they made, the moments they were proud of, the choices that took years to internalize. When the resulting evaluation only captures the surface — the lesson objective, the question type, the visible engagement strategy — it doesn't feel like being seen. It feels like being scored on the easy parts.

The triage isn't lazy capture. It's overworked capture. The admin is doing the job the workflow has asked them to do — they're just being asked to do too many jobs at once.

A different way to think about capture

The reframe is simpler than it sounds, and I want to state it plainly before defending it.

Stop triaging. Write down everything.

Everything you notice about what the students are doing. Everything the room is telling you. Every micro-decision the teacher makes. Every transition, every question, every wait time, every redirect, every piece of student work you saw on your way to or from a desk. Whether it "fits" in a rubric box. Whether it's "important." Whether you know where it'll go.

Stop asking should I write this down? and start asking what am I actually seeing?

I know what the objection sounds like. "That's not practical. I can't write down everything. And even if I could, I'd be drowning in notes I have to sort through later." That objection has been correct for the entire history of classroom walkthroughs — which is exactly why admins learned to triage in the first place. The triage was the workaround for a real constraint.

But the constraint has changed. More on that in a moment.

First, the case for why richer capture is just better observation, even before the categorization question gets solved.

When you stop triaging, you stop pre-filtering what you let yourself notice. The peripheral observations — the corner-of-the-eye details that don't fit a rubric box but are part of what makes a classroom what it is — start landing in your notes. The micro-moves the teacher makes are no longer competing with "where does this go in TKES Standard 3" for your attention. You see more, because you're not editing in real time.

You also produce a different kind of record. A walkthrough log full of rubric-coded fragments is one kind of artifact. A walkthrough log full of the actual texture of what happened in a room is another — and the second kind is far more useful for the real job of an instructional leader, which is helping teachers grow. A teacher who reads "demonstrated effective questioning strategies (Standard 4)" learns nothing. A teacher who reads "you waited eleven seconds after the question about the Treaty of Versailles, and the student who finally answered hadn't said anything all class period" learns something they can use.

The cost of the new approach is that someone has to categorize the rich capture into the rubric-aligned evaluation eventually. That used to be the admin, doing it manually, in the evenings, after every walkthrough. The constraint that made triage necessary was that the translation from rich observation to formal evaluation had to happen in the admin's head — which meant some of it had to happen during the walkthrough.

It doesn't have to anymore.

What changes when translation isn't your job

EvalScribe was built around exactly this premise — that the translation step from raw observation to formal, framework-aligned documentation can be handled by a purpose-built tool, freeing the administrator to do the observation job without the capture job competing for attention.

In practice, that means admins can do what I've been describing in this post: write down everything they see, in whatever order they see it, without pre-categorizing or rubric-cross-walking in real time. The students were quiet during the first five minutes but eyes-on. The teacher waited eleven seconds after that one question. There's an anchor chart from a week ago that students kept referring back to. The kid in the back row hasn't moved his pencil. The teacher chose to coach through the wrong answer instead of moving on. Whatever you saw, in whatever language comes naturally — that's the capture.

EvalScribe then takes that raw observational text and translates it into formal, framework-aligned evaluation language. It connects the observations to the relevant standards. It applies consistent rubric-aware scoring. It produces an evaluation document that maps cleanly onto whatever framework your district uses — TKES, AQTS, FEAPs, Marzano, Danielson, others — without the administrator having to do that translation work in their head, during or after the observation.

The reframe matters more than the tool, honestly. The principle is "stop triaging, capture richly, let the categorization happen later." That principle is true whether you use EvalScribe, some other purpose-built tool, or any approach that decouples observation from real-time categorization. EvalScribe is the most thoroughly purpose-built version of that decoupling I know of — and I built it precisely because I kept watching administrators try to do both jobs at once and watching the observation suffer for it. But the practice insight stands on its own.

The administrator is still the evaluator. The judgment about what the observation means — what the scores should be, what feedback the teacher needs, what coaching conversation to have next — remains where it has to remain. What changes is the tax on the administrator's attention during the moments they should be most attentive.

The more an administrator captures, the richer the snapshot the translation has to work with — and the richer the resulting evaluation. Write more, observe better. Let the categorization wait.

What this looks like in practice

For administrators who want to try this approach, a few practical notes.

The capture can be messy. You're not writing for an audience. You're writing for yourself, in whatever shorthand makes sense in the moment. Fragments, names, times, half-sentences. The translation step doesn't need polish — it needs texture.

Capture as much non-teacher content as possible. Most admins instinctively focus their notes on the teacher. Force yourself to write about the students, the room, the work, the questions, the silences. The richer the non-teacher capture, the more genuinely useful the resulting evaluation.

Include what surprised you. The moments that made you pause, the small decisions you didn't expect, the things that didn't fit a category — those are often the most valuable parts of an observation. They're also the first things triage kills. Capture them deliberately.

Don't pre-edit for charity or harshness. Note what you saw. The evaluation is downstream of the observation, and tone-balancing is a job for that step, not for the observation itself.

Read the room on the way out. The last thirty seconds of a walkthrough — what students are doing as you leave, how the teacher closes a transition, what the room looks like settling — is often as informative as the first five minutes. Capture it.

A walkthrough conducted this way takes about the same amount of time as one conducted the old way. But the quality of the resulting observation is different, and the teacher on the receiving end of the eventual evaluation feels the difference.

The point

The administrators who give the best feedback to teachers are the ones who actually watched the classroom. The triage question — what should I write down? — has been quietly preventing that, by making admins observe and categorize in the same moments. The reframe is to let those two jobs happen at different times. Observe richly in real time. Let the translation work happen afterward, with a tool built for it.

For the kind of work EvalScribe was built to support, this is the whole point. A purpose-built translation tool isn't valuable because it produces evaluations faster — that's the visible benefit. The deeper benefit is that it changes what an administrator can do during the walkthrough itself. The capture job stops competing with the observation job. The admin can finally be where they are.

If you want to see how EvalScribe handles the translation step in practice — and how that affordance changes the kind of evaluation you can produce — an individual administrator license is $100 per year, deliberately set below most districts' procurement thresholds. You can decide for yourself, without a committee. Questions, or interest in a school or district license, reach me at [email protected].

And if you haven't read the companion post yet, it covers the what to look at side of the same question: The First Five Minutes of a Walkthrough: What Admins Should Be Looking For.

FAQ

What should administrators write down during a classroom walkthrough? The richer the better. Observations about students, the room, teacher decisions, student work, transitions, questions, silences — anything you notice, in whatever order you notice it. The traditional advice has been to prioritize ruthlessly because the admin has to translate everything into rubric language later. A purpose-built translation tool removes that constraint.

Is it possible to write down too much during a walkthrough? For traditional workflows, yes — admins were limited by what they could realistically translate into formal evaluation language by hand. For workflows that include a purpose-built translation tool, richer capture produces richer evaluations.

How long should walkthrough notes be? There's no universal answer. The shift worth considering is from "how short can I keep these so they're easy to write up later?" to "how richly can I capture what I'm actually seeing?" The notes are an artifact of attention, not a deliverable in themselves.

What's the difference between observation and capture? Observation is the act of attending to what's happening in a classroom. Capture is the act of recording it. The case this post makes is that these two jobs have traditionally been forced to happen simultaneously, at a cost to the quality of both — and that they can usefully be separated.

Does EvalScribe replace the administrator's judgment in an evaluation? No. The administrator remains the evaluator. EvalScribe translates raw observational notes into formal, framework-aligned evaluation language. The judgment about what the observation means, what the scoring should be, and what feedback the teacher needs remains with the human evaluator. The administrator stays in the loop throughout.

Which evaluation frameworks does EvalScribe support? TKES, AQTS, Marzano, Danielson, and others. The full list of currently supported frameworks is available at evalscribe.com/frameworks.

How does this approach affect feedback conversations with teachers? Generally, positively. Evaluations produced from richer observations tend to feel more specific and grounded to the teacher receiving them — they reference what actually happened in the room rather than rubric-shaped fragments. That specificity tends to make post-observation conversations more useful and less performative.