Collecting data without acting on it
is worse than not collecting it.
Exit tickets and formative checks are widely used in schools. Acting on them the same day is not. The typical pattern: teacher collects responses, scans them overnight, notes the gap — and then either does nothing about it, adds it to a list of things to address “later in the unit,” or plans a full re-teach that takes more time than is available. None of these responses is agile teaching.
Agile teaching is the practice of closing the loop — making a specific adjustment based on today's data, for tomorrow's lesson, before the gap has time to compound. The longer the gap between data collection and instructional response, the less useful the data becomes. A gap identified on Monday is actionable on Tuesday. The same gap identified on Monday but addressed the following Tuesday is three times harder to close because the class has moved on and the misconception has had time to solidify.
The barrier to same-day action is almost never motivation — it's the absence of a fast, reliable decision process. Without a protocol, a teacher staring at a stack of exit tickets is facing an open-ended planning problem at 8pm. With a protocol, the same stack takes 5 minutes and produces a single actionable decision: what to do with the first 5–8 minutes of tomorrow's lesson.
The decision framework:
pattern → response → prompt.
After scanning exit tickets or 3-question checks, you will find one of three patterns in the class data. Each pattern has a corresponding response type. The framework maps pattern to response — so the planning decision is reduced to: which pattern am I looking at?
Example — Science, osmosis: 14/25 students described the concentrated solution as “pulling” water across. Original explanation used a diagram with arrows. Tomorrow's approach: remove the arrows and use a ratio demonstration — show students that if 8 in 10 molecules are water (dilute) vs 3 in 10 (concentrated), net movement toward the concentrated side is the only outcome that doesn't require magic. Time to plan: 3–4 minutes. You're not rebuilding the lesson — you're replacing one explanation with another.
Example — History, WWI causes: Students can explain the alliance system in 1914 Europe but struggled to predict what would have happened if Austria-Hungary had stayed neutral. Tomorrow's opening: present the Moroccan Crises (1905, 1911) — a test of the alliance system before 1914 that didn't escalate to war. “Why didn't these crises produce a world war? What was different?” This forces students to apply the alliance-system logic comparatively rather than just to 1914.
Peer teaching works for this pattern because the students who got it right are close enough to the misconception to explain the correction in accessible language. A teacher's explanation is filtered through expert knowledge — a peer's explanation is filtered through recent confusion, which is often more relevant.
Peer teaching setup: “Person A — explain your Q3 answer to Person B. Don't just give them the answer. Ask them where their reasoning went and help them find the error themselves. Person B — tell Person A exactly where you lost the thread. You have 6 minutes.”
From scan to tomorrow's plan
in five minutes.
The full protocol — scan to decision — takes 5 minutes when the framework is familiar. The output is a single sentence and a single action. Nothing more is needed tonight.
Don't read fully on the first pass — skim for the key indicator. You're looking for the most common pattern, not individual exceptions. 30 responses in 90 seconds is achievable with practice.
Apply the A2 diagnostic matrix. Which Q1–Q2–Q3 pattern describes the majority of your 'partial' pile? Write the pattern on the first response you picked up — that's your reference for tomorrow. Most nights it's immediately obvious.
Write: "Tomorrow: [Pattern A/B/C response] — [specific content]." Stick it to your lesson notes or type it into your planning document. That's the entire planning output for tonight's exit ticket work.
For Pattern A responses — where you need a different explanation or analogy — AI can generate the replacement in under a minute. For Patterns B and C, the intervention is usually a restructuring of existing content, so AI generation isn't always necessary.
How C3 feeds C4 and what
the compounding looks like.
The data-to-action workflow in C3 doesn't operate in isolation — it is the input to the C4 lesson iteration workflow. Once same-day action is a reliable daily habit, the question shifts: how do you track which interventions worked? How do you know whether the Pattern A response you used on Tuesday actually closed the gap identified on Monday?
C4 covers the iteration workflow at the lesson level: the 20-minute process that converts a week of exit ticket data into a set of targeted improvements to the lesson plan itself, so the gap doesn't recur with the next cohort. C3 is the daily practice; C4 is the weekly learning loop. Together they close the agile teaching cycle.
Tuesday — C3 response delivered: Ratio demonstration instead of arrow diagram. New exit ticket: Q3 improvement. 18/25 now get it. 7 still using “pulling” language.
Tuesday evening — C3 loop continues: Pattern still present in 7 students. Tomorrow: peer teaching moment for those 7, while the 18 work on extension.
Wednesday — gap closed: All 25 students now show correct mechanism in Q3. Data documented in lesson iteration log.
C4 input — end of week: Lesson iteration: replace arrow diagram with ratio demonstration permanently. The next cohort won't hit the same wall at the same point. This is what compounding looks like in agile teaching.
Same-day action as a
5-minute daily habit.
The most common failure mode for formative assessment practice is unsustainability. A teacher implements exit tickets with full commitment, spends 45 minutes reading and planning responses for the first week, realises it's not manageable at that intensity, and gradually stops. The practice disappears.
The goal of the protocol in this article is to make same-day action compatible with the actual working conditions of a teacher. Five minutes is achievable every night. Forty-five minutes is not. The protocol is calibrated to 5 minutes because that is the threshold below which a practice can become habitual — above it, it requires a specific decision to invest time, which creates friction, which eventually wins.
The AI generation step keeps it sustainable for Pattern A responses — the ones that require new content. Without AI, a Pattern A response requires writing a new explanation from scratch, which is genuinely time-consuming. With AI, the explanation is generated in 60 seconds and reviewed in another 60 seconds. The total time investment stays below the 5-minute threshold even on the most demanding nights.
Where to go next in
the agile teaching loop.
C3 has covered the full feedback loop: designing exit tickets that produce actionable data (A1), using the 3-question check for structured multi-level assessment (A2), and converting tonight's scan into tomorrow's specific adjustment (A3). The daily practice is now complete.
C4 — Lesson iteration picks up where A3 ends: the weekly workflow that turns a series of same-day adjustments into a permanently improved lesson plan. C4 is where individual tactical responses become strategic improvements to the curriculum itself. C8 — AI as the agile tool covers how AI compresses each of these stages further, so the practice scales across a full teaching week without the time cost expanding accordingly.