You want three reading levels doing the same group project. Your fastest group needs a real challenge or they'll finish in ten minutes and start distracting everyone else. Your struggling group needs the same core skill without three extra layers of vocabulary standing in the way. And you have one planning period, not three.
The instinct is to write three completely separate assignments. Don't. What you actually need is one task with tiered entry points and one rubric that scores the same underlying skill regardless of which tier a group worked from.
The trap of three separate rubrics
If group A gets a different rubric than group B, you've created a fairness problem before the project even starts. Students compare notes. Someone always notices their rubric "looks easier" or "looks harder," and now you're managing a grievance instead of a lesson.
The fix is one rubric built around the transferable skill, with the tiering happening in the *task itself* — the text complexity, the scaffolding provided, the amount of the process that's already modeled for them — not in what's being graded.
Building the tiered task
Start with the skill you actually care about. Not "read this article," but "identify an author's claim and evaluate the evidence supporting it." That skill can live at three different entry points without changing what you're assessing.
Ask an AI to help you build the tiers fast:
"I'm building a tiered group project on [skill/standard] for [grade level]. I want three versions of the same task: one with a shorter, more scaffolded text and a partially-filled graphic organizer; one at grade level with a blank organizer; and one with a longer, denser source text and an open-ended prompt instead of an organizer. All three should be assessed by the same rubric measuring [the actual skill]. Draft the three task versions and one shared rubric."
Read what comes back critically. AI is decent at generating the scaffolding differences but sometimes drifts the actual skill being measured between tiers without meaning to — check that a paragraph in the top tier and a paragraph in the bottom tier are genuinely assessing the same thing, just with different support around it.
Assigning groups without making the tiering obvious
A few things that keep this from feeling like tracking in disguise:
- Don't name the tiers by ability. "Group 1, 2, 3" beats "on-level, advanced, foundational" printed anywhere a student might see it.
- Rotate which students get which entry point across the year. The same three kids getting the scaffolded version every time turns a differentiation tool into a label.
- Let the task, not the roster, decide. Sometimes group placement should follow interest or reading stamina that week, not a fixed tier assignment from September.
Where our tools fit
Our [differentiation plan generator](/differentiation-plan) is the faster starting point if you want the full tiered structure — text complexity, scaffolding, and grouping logic — built out for a specific unit rather than assembled piece by piece. And once you've got the tiered task itself sorted, our [rubric generator](/grading-rubric) helps you lock in the single shared rubric so the assessment stays consistent across all three versions, which is really the part that keeps the whole thing fair.
What still needs your judgment
AI can draft three task versions in the time it takes you to describe them. It can't sit in the room and notice that the "on-level" group is actually struggling more than the scaffolded group this particular week, or that a student you tiered up needs to move back down for this one project. Treat the tiers as a starting draft for Monday, not a fixed assignment for the semester.
