How Lesson Starters Work | Why the First Ten Minutes Need a Job, Not Just an Activity

The first ten minutes of a lesson are unusually expensive.

They are short enough to be dismissed as “just the starter”, yet important enough to change the state of the room. Students arrive carrying different things: the last lesson, the corridor, a difficult morning, an unfinished conversation, a forgotten prerequisite, a strong memory of yesterday’s method, no memory of yesterday’s method, or simply the inertia of not yet thinking about this subject.

A lesson starter is the bridge between that incoming state and the intellectual work the lesson is actually for.

That makes the central question surprisingly difficult:

What should the first ten minutes do?

Should students retrieve previous learning? Discuss the new topic? Answer a prequestion? Correct yesterday’s errors? Read an example? Complete a short fluency task? Inspect an image? Write silently? Listen to an explanation?

There is no universal answer because these activities solve different problems.

A retrieval quiz is useful when the lesson needs older knowledge to become available. A short discussion may be useful when the new topic needs language, prediction or a problem space. A diagnostic check is useful when the teacher genuinely needs to know whether a prerequisite is secure before proceeding. A settling routine may be useful when transition cost is high. A prequestion can direct attention toward the relationship the explanation is about to resolve.

The starter therefore should not be selected because it is fashionable, familiar or easy to photocopy.

It should be selected because it has a job.

A strong lesson opening usually performs a small number of functions: secure transition, orient attention, activate or sample useful knowledge, establish the intellectual destination, and hand students into the core learning without consuming the lesson it was meant to launch.

The best starter is not the busiest first ten minutes.

It is the one that makes the next thirty, forty or fifty minutes more learnable.

The 50-second route

If you only have a minute, keep this:

• A lesson starter is a transition mechanism, not a decorative warm-up.
• Decide its job before choosing its activity: settle, retrieve, diagnose, foreground, rehearse, repair or orient.
• Retrieval is valuable when prior learning must be available, but a starter does not have to be a retrieval quiz every day.
• Discussion can foreground new learning, but it needs a tight intellectual object rather than vague “talk to your partner”.
• Keep the opening bounded. If the starter routinely becomes twenty-five minutes, it is no longer serving the lesson; it is replacing part of it.
• Use high-participation formats where possible. One volunteer answering does not tell you whether the class is ready.
• If the starter produces diagnostic evidence, the teacher should be willing to change the next move. Otherwise it was not really diagnostic.
• Make the handoff explicit: students should know when the starter has finished and what its result has to do with today’s learning.
• Judge starters by downstream value: Did the opening improve access to the main lesson?

The recent research question is more interesting than “retrieval or discussion?”

The Education Endowment Foundation’s A Winning Start trial, completed in September 2026, is useful precisely because it refuses a simplistic answer.

The feasibility trial compared two ten-minute science lesson starters: a retrieval quiz focused on previous learning and a discussion starter focused on foregrounding new learning. The headline finding was not that one starter type clearly defeated the other. The evaluation was mainly designed to test the feasibility of the Teacher Choices methodology, and the attainment analysis did not establish a reliable difference between the two approaches.

That result should not be turned into “retrieval does not work” or “discussion is equally good for everything”.

Retrieval practice has a much broader evidence base than this one starter comparison. The trial asked a narrower classroom-design question under particular implementation and measurement conditions.

The more useful lesson is that starter choice should follow starter purpose.

A previous-learning quiz and a new-learning discussion are not interchangeable versions of the same mechanism. One mainly asks, “Can students bring relevant knowledge back?” The other may ask, “Can students enter the problem space the next explanation will develop?”

Both can be sensible.

Both can be badly designed.

Both can overrun.

The first ten minutes need a job before they need a format.

1. A lesson starter solves a state-change problem

Students do not enter a lesson as blank processors.

At 10:03 a.m., a learner may still be emotionally in the previous lesson. At 2:15 p.m., another may be physically tired. One student has the prerequisite available. Another learned it months ago but cannot retrieve it. Another has never secured it. A fourth already knows today’s material and is waiting to see whether the lesson will challenge them.

The teacher also needs a state change.

They need the room to move from arrival to common intellectual direction.

Think of the opening as:

incoming state → attention → usable prior knowledge → shared purpose → first real learning move

A starter can help each transition.

But if the teacher tries to make it do everything—settle behaviour, review five old topics, diagnose misconceptions, teach vocabulary, preview the lesson, collect homework and launch a discussion—the ten minutes become an overloaded control panel.

Good starters are selective.

2. First decide whether the main lesson depends on older knowledge

Suppose today’s Mathematics lesson introduces solving simultaneous equations.

Students need:

• confidence with linear equations;
• interpretation of equality;
• basic substitution or elimination-related manipulation, depending on the chosen route.

A starter that samples one or two of those prerequisites can be high leverage.

Now suppose today’s English lesson begins a new poem whose central challenge is interpreting a strange speaker.

A five-question retrieval quiz on last month’s punctuation unit may be educationally useful in the long run, but it may not be the best immediate launch into this lesson.

The question is not:

“Is retrieval good?”

The question is:

“What knowledge needs to be available for the next learning move?”

That is a dependency question.

Cumulative review may still deserve a separate place in the curriculum. But a lesson starter should not become a ritual that ignores the lesson’s architecture.

3. Retrieval starters work when retrieval has a defined target

A weak retrieval starter is random memory housekeeping:

1. Define osmosis.
2. Name a planet.
3. State the formula for density.
4. What is a variable?
5. Label the heart.

All of those may be valid science knowledge.

Together, they may do little to prepare a lesson on diffusion across membranes.

A stronger starter selects knowledge with a reason:

• What is the difference between concentration and amount?
• Sketch particles in a region of high and low concentration.
• What does “net movement” mean?
• Predict what happens if the concentration difference becomes smaller.

Now retrieval is not just recall.

It is assembling the intellectual components the lesson is about to use.

The opening is functioning as prerequisite activation.

4. Discussion starters work when there is something worth discussing

“Talk to your partner about gravity.”

That is not a strong discussion starter.

The topic is too wide. Students may repeat fragments, share misconceptions or wait for someone else to speak.

A stronger foregrounding prompt creates a precise object:

“Two balls have the same size but different masses. They are dropped from the same height. Predict what happens and give one reason.”

Now the class has:

• a claim to make;
• prior ideas to activate;
• a disagreement space;
• a prediction that later explanation can revisit.

The starter does not need to settle the science.

Its job may be to make the question cognitively active.

That is different from retrieval.

5. Prequestions can make the lesson’s problem visible

A good prequestion gives students something to listen for.

Before teaching thermal expansion:

“Why do long bridges contain expansion joints?”

Before teaching persuasive technique:

“Why might the same fact persuade one audience and fail with another?”

Before teaching percentage reversal:

“If a price falls 20% and then rises 20%, is it back where it started?”

Students do not need to know the answer.

The attempt creates a problem space.

The later explanation now answers something the learner has already tried to resolve.

But use prequestions selectively.

If every lesson begins with six questions students cannot answer, the starter can feel like daily public failure.

One or two high-leverage questions are often enough.

6. Some starters should simply settle and orient

Not every opening needs clever cognition.

A class moving from a noisy transition may need:

• materials out;
• a short written task;
• clear first action;
• teacher circulation;
• quiet transition into subject mode.

This can be educationally valuable because the starter protects attention.

A stable opening routine reduces the executive cost of beginning.

Students know: enter, sit, open book, complete the first prompt, teacher begins at the agreed time.

Routine is not the enemy of thought.

It can create the conditions in which thought begins faster.

The danger is when the routine becomes intellectually empty.

Copying the date and title for six minutes may produce order while consuming prime learning time.

The starter should settle into learning, not merely settle.

7. The starter should expose readiness only when the teacher can respond

A quick check can be diagnostic.

Before teaching quadratic factorisation, the teacher asks students to expand:

[ (x+3)(x+5) ]

Half the class cannot.

That matters.

If the planned lesson assumes expansion is secure, the evidence should alter the route.

Perhaps the teacher repairs expansion briefly. Perhaps they split support. Perhaps they change the first worked example.

If the teacher sees the result and teaches the original lesson unchanged, the check was informational theatre.

The principle is simple:

evidence earns the name “diagnostic” only when it can change the next decision.

This is why not every starter needs to produce data.

Sometimes the job is orientation, not diagnosis.

8. High participation matters more in ten minutes

Ten minutes disappear quickly.

A starter built around: “Who remembers…?”

may be carried by the same three students.

The rest of the class can remain cognitively absent while the teacher experiences a lively beginning.

Prefer formats that make more thinking visible:

• mini-whiteboards;
• one-sentence written responses;
• simultaneous multiple choice;
• short individual solve-then-share;
• quick annotation;
• everyone marks a prediction;
• brief silent retrieval before discussion.

The aim is not forced public performance from every student.

It is to reduce the gap between “the class answered” and “a student answered”.

9. The first question should be easy enough to enter

A starter can fail because its first demand is too high.

Students arrive and immediately face an unfamiliar multi-step problem.

Some engage. Others stall. The teacher begins helping individuals. The opening fragments.

A strong first minute often has a low activation cost.

Examples:

“Write everything you remember about the term ‘gradient’ for thirty seconds.”

“Which of these three graphs could show direct proportion?”

“Read these two sentences. Underline the difference.”

“Estimate first—no calculation yet.”

The task should not be trivial forever.

It should make beginning cheap enough that the room enters the subject together.

10. The starter can become harder after entry

A bounded starter can have a short progression.

Mathematics:

1. retrieve a fact;
2. use it;
3. choose between methods.

Science:

1. label a diagram;
2. predict a change;
3. explain the prediction.

English:

1. identify a phrase;
2. infer its effect;
3. compare two interpretations.

This progression avoids the false choice between “easy warm-up” and “deep thinking”.

The opening can begin with access and move toward the lesson’s intellectual demand.

11. A starter should not become a miniature examination every day

Daily quizzes can provide useful retrieval.

But constant scoring can change the emotional meaning of the opening.

Students may begin each lesson feeling evaluated before they feel taught.

Low-stakes retrieval should actually be low stakes.

That can mean:

• self-checking;
• no grade;
• quick correction;
• normalising error;
• using results to guide teaching rather than rank students.

If the purpose is memory strengthening and readiness, high-stakes scoring may interfere with the mechanism.

12. The correction phase is part of the starter

Students answer five retrieval questions.

The teacher displays the answers.

Everyone ticks.

Starter finished?

Not necessarily.

The educational value often sits in the correction.

Why was the wrong answer tempting? Which distinction was forgotten? Which item matters for today’s lesson? What should be repaired now and what can wait?

A two-minute correction that reveals structure can be more useful than adding five more questions.

The goal is not answer-count throughput.

It is readiness.

13. Link the starter explicitly to the lesson

A common failure is the orphan starter.

Students complete it. Teacher marks it. Then:

“Right, today we’re doing something different.”

That teaches students that starters are administrative rituals.

Instead, make the bridge visible.

“You just retrieved the idea that equal operations preserve an equation. We need that in today’s simultaneous-equation method.”

“The prediction you made about the bridge is the problem today’s explanation will answer.”

“You noticed the writer keeps changing sentence length. Today we are asking what those shifts do to pace.”

The handoff converts the first ten minutes into part of one lesson rather than a separate mini-lesson.

14. The starter must know when to stop

A discussion becomes interesting.

A misconception appears.

The teacher follows it.

Ten minutes become twenty-five.

Sometimes that is the right decision.

But if it happens every day, the curriculum has a capacity problem.

Set a stopping rule.

Examples:

• five questions, eight minutes;
• one prediction, two partner exchanges, whole-class synthesis;
• three prerequisite checks, then proceed or trigger a defined repair;
• one error to analyse deeply, not six.

A starter earns its name partly by being bounded.

15. When should a starter expand?

Sometimes the opening reveals a structural problem.

Example:

A class is supposed to start solving equations with fractions.

The starter shows that most students cannot add simple fractions.

Proceeding would create failure downstream.

The teacher should expand the repair.

This is not “the starter overran”.

It is the starter doing diagnostic work that correctly changed the lesson.

The difference is deliberate adaptation.

Overrun without decision is drift. Expansion because evidence changed the route is responsive teaching.

16. When should the teacher ignore an error and proceed?

Not every wrong answer deserves immediate reteaching.

Suppose one student misremembers a low-leverage vocabulary term unrelated to today’s learning.

Stopping the class may not be worth it.

The teacher can note the issue for later cumulative review.

Strong starter use requires triage:

• blocks today → repair now;
• important but not blocking → schedule later;
• isolated slip → quick correction;
• broad misconception → investigate.

Ten minutes forces prioritisation.

17. Starter design should vary by subject

Mathematics

Good jobs:

• activate prerequisite procedures;
• compare methods;
• retrieve definitions;
• estimate;
• identify a misconception;
• set up a problem structure.

Avoid making every starter ten decontextualised calculations.

Science

Good jobs:

• predict;
• retrieve a model;
• interpret a small dataset;
• distinguish observation from explanation;
• recall a prerequisite concept;
• surface an existing model.

English

Good jobs:

• sentence-level rehearsal;
• vocabulary in context;
• short close reading;
• oral or written response to an interpretive question;
• retrieve a structural feature;
• foreground audience or purpose.

History

Good jobs:

• chronology;
• source orientation;
• causal recall;
• claim evaluation;
• retrieval of prior context.

The format should follow the epistemic work of the subject.

18. The same starter can serve different jobs—and that can be dangerous

A four-question quiz might be used for:

• retrieval practice;
• diagnosis;
• accountability;
• settling;
• grading;
• review.

Those purposes have different design requirements.

If the teacher does not know which purpose is primary, decisions become muddled.

For retrieval, difficulty should allow successful effort and correction.

For diagnosis, item choice should discriminate among likely knowledge states.

For grading, reliability and fairness matter more.

For settling, predictability and low start-up friction matter.

The visible activity is the same.

The mechanism is not.

19. Starter consistency and starter variety should coexist

Students benefit from predictable routines.

They also benefit from different cognitive jobs.

The solution is often to keep the container stable and vary the payload.

Example:

Every lesson begins with eight silent minutes and one visible slide.

But the slide may contain:

• three retrieval questions;
• one graph to interpret;
• one prequestion;
• one short error analysis;
• one sentence to improve.

Students know how to begin.

The intellectual task changes with the lesson.

This reduces procedural load without turning cognition into routine.

20. A starter should not become teacher paperwork

If every starter generates a score that must be recorded, analysed, entered and tracked, a useful ten-minute routine can create an unsustainable data burden.

Most starter evidence can remain ephemeral.

The teacher needs enough information to make the next decision.

Not every response needs a spreadsheet.

Record patterns when they matter:

• persistent prerequisite weakness;
• repeated misconception;
• intervention monitoring;
• progress across a planned sequence.

Otherwise, use the evidence and let it go.

21. A current trial gives an important implementation warning

The EEF A Winning Start feasibility work did more than compare starter types.

It also showed the difficulty of studying ordinary classroom routines cleanly.

Teachers did not always deliver the assigned starter within the intended ten-minute window. Teacher-developed attainment measures also raised validity and reliability challenges.

Those implementation details matter for schools too.

A starter policy can say “ten-minute retrieval” while classrooms implement:

• five minutes;
• fifteen minutes;
• discussion of answers for another ten;
• different numbers of questions;
• varying levels of teacher explanation.

The label can be consistent while the instructional dose is not.

If a school wants to evaluate starters, it should observe the actual routine, not just the policy name.

22. Why a universal starter mandate can go wrong

A school decides:

“Every lesson will begin with five retrieval questions.”

The aim is coherent.

The rule may produce useful consistency.

But it can also create:

• random questions unrelated to dependencies;
• retrieval squeezed into practical lessons where another launch is needed;
• duplicated review;
• teacher compliance instead of judgement;
• lost opportunities for prediction, discussion or diagnosis.

A better school-wide principle might be:

Every lesson should begin with a purposeful, bounded entry routine that prepares students for the learning ahead. Retrieval should be used frequently where previous knowledge needs to remain accessible.

That preserves evidence-informed practice without pretending one format solves every subject problem.

23. The learner route: use the starter as information, not wallpaper

Students can learn to read the opening.

Ask:

“What is this trying to make available?”

If it is retrieval: try before looking at notes.

If it is a prequestion: commit to a prediction.

If it is a diagnostic item: show what you actually know rather than copying.

If it is error analysis: explain why the wrong route is tempting.

If it is discussion: bring one idea and one reason.

The starter becomes more useful when the learner understands its job.

24. The parent route: do not judge a lesson by how impressive the first worksheet looks

Parents rarely see lesson starters directly, but they may see homework, revision routines or school policies built around them.

A useful question is:

“What does the school use the opening for?”

If the answer is retrieval, ask whether it is cumulative and corrected.

If the answer is diagnosis, ask whether teachers respond to the results.

If the answer is settling, ask whether students transition quickly into substantive learning.

Avoid assuming that more starter questions mean more learning.

The starter should increase the value of the lesson that follows.

25. The teacher route: a five-question design check

Before choosing tomorrow’s starter, ask:

1. What state are students arriving from?
2. What must be mentally available for today’s learning?
3. What do I need to know about readiness before proceeding?
4. Which opening activity does that job at the lowest useful cost?
5. What is my stopping or adaptation rule?

Then design.

This sequence is stronger than: “I need a starter. What worksheet do I have?”

26. A simple starter decision tree

Does today’s lesson rely heavily on previous knowledge?

Yes → retrieve or check the prerequisite.

No → ask the next question.

Would a prediction/problem make the new learning more intelligible?

Yes → use a prequestion, example, phenomenon or discussion.

No → ask the next question.

Is transition into learning itself the main problem?

Yes → use a stable low-friction entry routine.

Do you need evidence that could change the lesson?

Yes → use a genuine diagnostic check with a response plan.

Is there a recent error that will contaminate today’s work?

Yes → repair it.

The tree does not choose for the teacher.

It makes the purpose explicit.

27. Worked case: the algebra starter that was too broad

A teacher begins every Mathematics lesson with ten mixed arithmetic questions.

Students improve at the routine.

Today’s lesson is completing the square.

Many students then fail because expanding ((x+a)^2) is unstable.

The starter included arithmetic but did not sample the dependency.

The teacher changes the system:

• two cumulative items;
• two questions linked to current prerequisites;
• one short reasoning item.

The starter becomes smaller and more useful.

28. Worked case: the science discussion that became guessing

Before a lesson on convection, the teacher asks:

“What do you think convection is?”

Students guess definitions.

Few have a basis.

The discussion produces noise.

Next class, the teacher shows a beaker diagram with warmer and cooler regions:

“Predict which water will move upward and explain what might cause the movement.”

Now the discussion has an object.

Students can use existing ideas about heating and density.

The later explanation can refine them.

29. Worked case: the English starter that repaired entry, not yesterday

A class begins a lesson on argumentative writing after lunch.

The teacher previously used five grammar retrieval questions.

Students completed them slowly and then had to switch mental modes into argument.

The teacher changes the opening.

Students read two short claims and choose which is easier to defend, then write one reason.

Within four minutes, the class is already operating inside argument.

Grammar is still maintained through separate cumulative review during the week.

The starter now serves the live lesson.

30. Worked case: the starter that correctly took twenty minutes

A class is about to learn percentage increase.

The starter asks: “What is 10% of 80? What is 25% of 80? Explain how you know.”

Half the class cannot find 10%.

The teacher stops and repairs percentage-as-operator using simple examples.

The planned lesson shifts.

This is not failure.

The starter prevented new teaching from being built on a missing prerequisite.

The cost of twenty minutes saved a larger downstream failure.

31. Common failure mode: starter as compliance ritual

Students learn: enter, copy, wait.

The intellectual work is minimal.

Repair: make the first action cognitively meaningful and short.

32. Common failure mode: starter as random retrieval

Questions are selected because they are available.

Repair: choose cumulative or prerequisite knowledge deliberately.

33. Common failure mode: starter as surprise test

Students are scored before instruction begins.

Repair: separate low-stakes retrieval/diagnosis from summative judgement.

34. Common failure mode: starter as teacher monologue

The teacher spends ten minutes explaining answers before students have attempted anything.

Repair: protect the learner attempt where retrieval or diagnosis is the job.

35. Common failure mode: starter as permanent overrun

Interesting discussion expands daily.

Repair: set a time boundary and decide which issues deserve immediate expansion.

36. Common failure mode: starter data with no decision

Mini-whiteboards reveal a misconception.

Teacher proceeds anyway.

Repair: define in advance what evidence would trigger reteaching, support or continuation.

37. Common failure mode: one starter for every subject

School policy equates consistency with identical cognition.

Repair: standardise the purposeful opening routine if useful; allow subject-appropriate payload.

38. Common failure mode: too much novelty

Every day introduces a different game, platform or procedure.

Students spend the first minutes learning how to do the starter.

Repair: keep procedural architecture predictable.

39. Common failure mode: too much ease

Students complete familiar recall flawlessly.

They feel successful.

The starter never exposes readiness for the harder learning.

Repair: include at least one item that reaches toward the dependency or decision the lesson needs.

40. Common failure mode: too much challenge too early

The first item is a demanding unfamiliar problem.

Many students freeze.

Repair: create an accessible entry followed by increasing cognitive demand.

41. Frequently asked questions

Should every lesson have a starter?

Every lesson needs a workable entry into learning. That does not require a separate worksheet or named starter. Sometimes the first explanation, demonstration, practical setup or reading activity is already the right entry.

Is retrieval practice the best lesson starter?

Retrieval practice is strongly supported for strengthening memory, but starter design is a purpose question. Retrieval is especially useful when older knowledge should become available or remain durable. Other lessons may benefit more from a prediction, phenomenon, diagnostic check or direct entry.

How long should a starter last?

Five to ten minutes is common because it protects the main lesson, but there is no universal magic number. The essential question is whether the opening earns its time and has a clear stopping rule.

Should the starter always review the previous lesson?

No. Cumulative review often matters more than simply reviewing yesterday. Some starters should activate older prerequisites; others should foreground new learning.

What if students get most questions wrong?

Diagnose the pattern. If the missing knowledge blocks today’s lesson, repair it. If it is important but not immediately blocking, schedule later review. Do not merely display answers and proceed.

Should starter work be marked?

Usually it does not need formal grading. Rapid self-checking, whole-class correction or teacher scanning may be enough. Formal marks change the stakes and create administrative cost.

Are discussion starters too unstructured?

They can be. Give discussion a precise question, object, time and output. “Discuss the topic” is weak; “choose which claim is better supported and give one reason” is stronger.

Can starters help behaviour?

A predictable entry routine can reduce transition ambiguity and get students working quickly. But starter design is not a substitute for a wider classroom-management system.

Should schools mandate one starter format?

Caution is sensible. A school can establish principles—purposeful, bounded, high participation, connected to learning—without requiring one identical cognitive activity in every lesson.

How do I know whether a starter works?

Look downstream. Does the class enter the main task faster? Are prerequisites available? Did the diagnostic evidence prevent avoidable failure? Does the later explanation connect to the initial question? If the starter vanished tomorrow, what valuable function would disappear?

The deeper lesson: the first ten minutes are an allocation decision

Time at the beginning of a lesson feels cheap because the major explanation has not started.

It is not cheap.

Those minutes decide whether students arrive intellectually, whether useful knowledge is available, whether a hidden gap is discovered, whether the lesson has a problem worth solving, and whether the transition consumes energy or releases it.

The best starter is therefore not the most entertaining, the most rigorous-looking or the most fashionable.

It is the one with the clearest job.

Sometimes that job is retrieval.

Sometimes it is prediction.

Sometimes it is diagnosis.

Sometimes it is settling.

Sometimes the strongest opening is simply one excellent question that makes the whole lesson necessary.

The design rule is:

Start with the state the learner needs to reach, then choose the shortest opening that gets them there.

When the first ten minutes have a job, the rest of the lesson inherits a better starting position.

Sources

• Education Endowment Foundation, Teacher Choices trial: A Winning Start (completed September 2026): https://educationendowmentfoundation.org.uk/projects-and-evaluation/projects/teacher-choices-trial-a-winning-start
• Education Endowment Foundation, Metacognition and self-regulation — Developing Independent Learners (24 August 2026): https://educationendowmentfoundation.org.uk/16-19/developing-independent-learners/metacognition-and-self-regulation
• OECD, Unlocking High-Quality Teaching — Ensuring cognitive engagement (2025): https://www.oecd.org/en/publications/unlocking-high-quality-teaching_f5b82176-en/full-report/ensuring-cognitive-engagement_998c3147.html

Surgical internal links

• How Prequestions Work | Why Trying Before Teaching Can Focus Attention Without Becoming a Test
• How Cumulative Review Works | Why Old Learning Must Reappear Before Students Need It Again
• Why Good Directions Start Before the First Word | How Instructional Directions Work
• Why Should Students Know the Destination Before the Lesson? | How Learning Goals Activate Prior Knowledge and Self-Monitoring
• How X Works | eduKateSG