Differentiating a Single Lesson Five Ways in Twenty Minutes

Differentiation fails in practice for a boring reason: there is not enough time on a Sunday evening to make five real versions of a task, so most of us make one and improvise. This is a method for getting to five genuine versions in about twenty minutes, and for noticing when a version looks differentiated but is not.

The important idea first. Differentiation is not making a task easier. It is changing what gets in the pupil's way while keeping the thinking intact. A watered-down worksheet removes the thinking, which is why pupils who always receive the easy sheet never catch up.

Step 1: Name the thinking you refuse to remove

Before anything else, write one sentence: every version of this task must require pupils to ___.

For a Year 9 geography lesson on flooding, mine was: every version must require pupils to explain how one human decision made flood damage worse. That sentence is the constant. Reading level, sentence length, scaffolding, output format - all of that can move. The causal explanation cannot.

Write it down before you generate anything, or you will drift.

Step 2: Describe your actual pupils, not labels

Generic requests produce generic output. Instead of make an EAL version, describe the pupil:

Recently arrived, Arabic first language, strong subject knowledge from previous schooling, reads English at roughly two years below chronological age, understands far more than they can currently write.

That produces a version with subject vocabulary preserved and a glossary added, rather than a version with the content stripped out - which is what the label alone tends to give you.

Keep five short profiles in a document. Reuse them all year. They take fifteen minutes to write once.

Step 3: One generation, five versions

I request all five at once so they stay genuinely parallel:

Core task below. The non-negotiable thinking: every pupil must explain how one
human decision made flood damage worse.

Produce five versions of this task for these pupils: [paste your five profiles].

For each version:
- keep the non-negotiable thinking intact
- state what you changed and why in one line
- keep all subject vocabulary; add a glossary where reading load is the barrier
- do not reduce the number of reasoning steps for any version

Do not produce a version that only asks pupils to identify or list.
Core task: [...]

That final prohibition is load-bearing. Left to itself, a model turns the lowest version into recall questions, which is exactly the failure mode you are trying to escape.

Step 4: Check each version against four tests

Fifteen seconds per version, and it catches nearly everything.

1. Thinking test. Does this version still require the non-negotiable reasoning? If a pupil could complete it by copying, reject it.
2. Dignity test. Would a fourteen-year-old be embarrassed to be seen with this sheet? Cartoon fonts and baby images do more damage than a hard text.
3. Convergence test. Can all five groups contribute to the same plenary discussion? If the stretch group has been given different content rather than deeper demands, the class splits permanently.
4. Accuracy test. Simplified explanations are where factual errors hide. Read every simplified causal claim yourself.

A worked example

Core task: Using the source pack, explain why the 2007 floods caused so much damage in this town.

Version	What changed	What stayed
EAL, strong prior knowledge	Sources rewritten at shorter sentence length, bilingual glossary of eight terms, sentence stems for the explanation	Full source pack, same causal question
Reads two years below	One source removed, key data presented as an annotated map, paragraph frame given	Same question, same required causal link
On level	Unchanged	Everything
Stretch	Added a conflicting source and a counter-argument requirement	Same question plus evaluation
Reading-support plan	Sources provided as audio with a printed transcript, response may be recorded verbally	Same question, same reasoning, different output mode

Every row ends at the same plenary: whose decision mattered most, and what should the council do differently? That is what makes it one lesson rather than five.

Step 5: Keep the versions

Save all five with the lesson, named by profile rather than by pupil. Next year the twenty minutes becomes five, and the versions improve each time you use them because you have watched real pupils struggle with specific lines.

What this does not solve

It does not tell you which pupil needs which version - that is professional judgement built from knowing your class, and it changes week to week. It does not replace the SEND provision in a formal plan; if a plan specifies a modality, that is a requirement and not a suggestion. And it cannot tell you whether the reading level is genuinely appropriate. Readability scores are estimates; a pupil reading aloud for ninety seconds tells you more than any number.

FAQ

Will pupils notice they have different sheets? Yes, and it matters far less than we fear when the sheets look alike, carry the same title, and end in the same discussion. Pupils object to being visibly ranked, not to variety.

How do I manage five versions in one room? Print the versions but hand them out by table, set the same success criteria on the board for everyone, and run one shared plenary. You are managing one lesson, not five.

Is the on-level version really unchanged? Usually, and that is the point. Your core task was designed for the majority. Differentiation means adjusting around it, not rebuilding it.

How long before this actually saves time? The profiles cost fifteen minutes once, and the first lesson takes closer to thirty minutes than twenty. From the third lesson onwards it is genuinely faster than making one version used to be, because the structure is now automatic.