Gradual Release of Responsibility — The 4 Phases, Examples and Classroom Application

Gradual release of responsibility is an instructional framework where the cognitive load of a task shifts from the teacher to the student in four deliberate phases — focused instruction (I do), guided instruction (we do), collaborative learning (you do it together), and independent practice (you do). It is the canonical scaffold for direct instruction lessons that introduce new skills or content. The aim is for every student to leave the lesson capable of doing the skill alone, with the teacher having released support only as fast as evidence allowed.

The original three-phase model — teacher responsibility shifting to student responsibility — was published by Pearson and Gallagher in 1983. The contemporary four-phase version, which adds collaborative learning between guided and independent practice, is most closely associated with Douglas Fisher and Nancy Frey, whose work over the last two decades made GRR the dominant scaffold for direct instruction in K–12 classrooms worldwide.

The four phases are: 1) Focused instruction (I do) — the teacher models the skill and thinks aloud while students observe. 2) Guided instruction (we do) — the teacher and students work a parallel problem together, with the teacher releasing decisions as students show readiness. 3) Collaborative learning (you do it together) — students apply the skill in pairs or small groups while the teacher monitors. 4) Independent practice (you do) — each student demonstrates the skill alone, giving the teacher formative evidence for the next lesson.

Almost. The three-step shorthand I do / we do / you do captures the original Pearson and Gallagher model. The modern four-phase GRR, as taught by Fisher and Frey, inserts a collaborative phase between we do and you do — sometimes called you do it together. The point of the extra phase is that students often fail at the independent step not because the modelling was weak, but because they have not yet articulated the steps in their own words to a peer. Collaborative learning surfaces the gap before the student is alone with the task.

Rough rule of thumb for a 50-minute lesson on new content: 10–15 minutes focused instruction, 10–15 minutes guided, 15–20 minutes collaborative, 10–15 minutes independent. The phases do not need to be equal — what matters is that students get genuine independent practice before the lesson ends. A lesson that runs out of time before phase four leaves you with no formative evidence and a class that has only ever done the task with support.

GRR is built for teaching new skills and procedural knowledge — decoding, an algorithm, a writing structure, a lab technique. It is less appropriate for inquiry, project-based learning, discovery tasks, or open-ended creative work where the point is for students to construct understanding themselves. If the learning intention is application of an already-known skill, GRR adds overhead. If the learning intention is a new skill where the wrong technique now is hard to unlearn later, GRR is among the strongest scaffolds available.

Direct instruction and explicit instruction are broader categories of teacher-led, structured lesson design. GRR is one specific scaffold inside that family — the one that names the four phases and the explicit handoff from teacher to student. You can run a direct-instruction lesson without using GRR phases (for example, a Madeline Hunter seven-step lesson), and you can use GRR phases inside a wider explicit instruction routine. Most contemporary direct instruction templates assume GRR as the default arc.

Skipping the collaborative phase. Teachers commonly model strongly (I do), do one guided example (we do), and then go straight to independent practice (you do). Students who have not yet articulated the steps to a peer often fail at independent and the teacher reads it as a content gap rather than a scaffolding gap. The fix is to plan a 10-minute collaborative task with a tangible product so students rehearse the reasoning out loud before they are alone with the page.

Yes — comprehension strategies are one of the strongest use cases for GRR. Phase one: model the strategy (asking questions, visualising, summarising) on a shared text while thinking aloud. Phase two: read a second passage together, prompting students to apply the strategy at chosen stopping points. Phase three: pairs apply the strategy to a third passage and discuss. Phase four: each student applies the strategy to a fresh passage and produces a written response. The same arc works for inference, synthesis, and main-idea instruction.

Both. The four phases scale from a 20-minute small-group reading lesson to a 60-minute whole-class maths lesson. In small groups the cycle can run faster (a single sitting), while in whole-class lessons the collaborative and independent phases usually need more time because variation across the room is wider. The key constraint is the same at any size: the teacher must release support only as fast as evidence permits, and every student must reach an independent attempt before the lesson ends.

Start at the end. Write the independent task first — exactly what each student should be able to do alone by the end of the lesson. Back-design the collaborative task so pairs rehearse the same reasoning the independent task will demand. Then plan the guided example as a parallel problem to the collaborative task. Finally, plan the focused instruction as the worked example that exposes the thinking the guided phase will surface. Planning back-to-front keeps the four phases aligned to a single skill rather than drifting into related-but-different content.

The independent phase is the primary formative checkpoint — what students produce alone is the evidence of whether the scaffolding worked. Inside the lesson, the guided phase produces continuous low-stakes evidence through mini-whiteboards, cold-call, and hinge questions. The collaborative phase produces evidence by listening as pairs talk through the reasoning. Treat the independent work as a diagnostic, not a grade — its purpose is to tell you what to reteach tomorrow, not to score the lesson.