Switch course ▾
Session 1. What to Scale: Core, Safe Adaptation and Readiness
Objective. Write the core of a proven pilot on one line, separate fixed from flexible parts, and judge with evidence whether it is ready to be scaled to more classrooms or schools.
Every school leader has seen a good practice travel badly: the name stays but the point is gone. Before you send anything to more classrooms, you need to know what made it work and whether it is ready. In this session you learn practical ways to find the core of a practice, to tell evidence from enthusiasm, and to decide honestly whether to scale, test more, fix or stop. A careful 'not yet' costs little now and saves a great deal of trust later.
Key ideas
Scale is more than numbers
- Cynthia Coburn (2003) described four dimensions of scale: SPREAD (more classrooms and schools), DEPTH (real change in how teachers think and teach, not only in forms), SUSTAINABILITY (it lasts after the first champion or the first funding ends) and SHIFT IN OWNERSHIP (schools and teachers, not the originator, carry it).
- Course 14.5 (Leading Educational Innovation) taught you to read adoption inside one school, where you know the people. Across a network you meet teachers you have never seen. A form can reach 12 schools in a month and change nothing in lessons: that is spread without depth.
- Before any launch ask four questions: how many classes (spread), what will teachers do differently on Monday (depth), what happens when I leave (sustainability), and who will own it in a year (ownership)? A plan that answers only the first is a distribution plan, not a scaling plan.
Tight on the core, loose on the rest
- FIDELITY means how closely teachers use a practice as its designers intended: how often, with what training and with which key moves. The EEF guide to implementation advises being 'tight' on the active ingredients, the few parts that make the practice work, and 'loose' on everything else.
- The EEF guide also says local adaptation is almost inevitable and can help buy-in, but the evidence on it is not yet strong. Adding something new tends to help more than changing a core part, and adaptation is safer after the core is securely understood and in use.
- Write a RECIPE CARD for every practice you plan to scale: 3 to 5 core moves with one reason each, a list of what may change freely, and RED LINES, changes that destroy the practice (for example turning peer explanation into marks for the markbook). One page, so that a stranger can use it.
Is the pilot ready to travel?
- A pilot run by its inventor, in a friendly class, with extra time, shows only that the idea can work in the best case. Banerjee, Duflo and colleagues describe how early designs of a promising Indian programme, Teaching at the Right Level, did not produce effects inside the regular government school system until later versions were redesigned.
- Use six readiness questions: (1) Better than the baseline in more than one class? (2) Run by teachers who are not the inventor? (3) Cost and teacher hours counted honestly, with the champion's extra time? (4) Core written clearly enough for a stranger? (5) Tried in a harder setting, such as a big class or few resources? (6) Does a receiving school want it and have the capacity?
- Give one of four verdicts: SCALE to the next ring, TEST MORE in the same ring, FIX THEN RETEST, or DO NOT SCALE. 'Test more' is not failure; a honest 'not yet' protects pupils and the idea's reputation. World Health Organization guidance on scaling up also assumes the innovation has first been tested for feasibility and effect in the local context.
Scale in rings, not in one jump
- One sensible pattern is rings: Ring 0, the originator with 1 to 3 classes; Ring 1, other teachers in the same school; Ring 2, three or four different schools, including a harder setting; Ring 3, the district network. Each ring ends with a written gate, the tool you met in course 14.5.
- The question changes from ring to ring. Ring 1 asks 'does it work without the inventor in the room?' Ring 2 asks 'does it work where conditions differ?' Ring 3 asks 'does it work without us visiting, and who pays?' Do not skip a ring because the previous one went well.
- Number of schools per ring is a judgement, not a rule. In a district with 20 schools, two or three in Ring 2 may be enough to learn; the point is variety (urban, village, large, small) rather than count. Keep a record of what you learned in each ring so the next one starts wiser.
Scaling in rings
Finding the core: ask what happens if it is removed
The core of a practice is rarely what its inventor says first. A simple method is to watch several lessons of the same practice, including some where it works less well, and write down what the stronger lessons have that the weaker ones lack. Then test each candidate with one question: if this step were removed, would the effect probably disappear? Steps that fail this test are flexible; steps that pass are candidates for the core.
Also ask pupils and colleagues. A Grade 6 pupil who says 'I have to explain, not just copy' is pointing at an active ingredient. Keep the core short, three to five moves, with a reason for each. A core of twelve items is a script, and scripts do not travel: teachers will follow them badly or drop them.
What counts as evidence of readiness
Good evidence for scaling is modest but specific: counts against a baseline in more than one class, a teacher who is not the inventor getting a similar pattern, an honest cost in hours, and a note of what did not work. Enthusiasm, a happy visit and a handful of praised pupils are encouraging, but they are not evidence that it will work with a tired teacher in a class of 40.
Beware of the champion effect: an inventor brings energy, time and goodwill that cannot be copied. Ask 'what would an ordinary teacher need to get the same result?' and test that with an ordinary teacher before spreading it. If the answer is 'a lot of extra hours', the practice may need to be simplified first.
Examples from the classroom
Finding the core by watching
Scene. Malika, a Grade 4 teacher in Samarkand, has a 'reading buddies' routine. The deputy head watches six lessons in three classes. In the best lessons pupils retell the text in their own words; in the weakest they only read aloud together. Malika had named the pairing as the key idea, but when the retell is removed in a trial lesson the discussion goes flat.
Insight. The core is found by comparing stronger and weaker lessons and by removal tests, not only by the inventor's account.
A readiness check that says 'not yet'
Scene. Aziz, a science teacher in Fergana, runs a practical lesson with pupils predicting results on a card. Eleven schools ask for it. The deputy head checks six questions: results exist for 3 classes, but only Aziz has taught it, he spent about 5 hours preparing cards and one school has no lab table. She decides on 'test more' in Ring 1 with two colleagues first.
Insight. Several 'no' answers on readiness are a reason for one more ring, not for refusal.
Adding, not changing
Scene. After a term, a Tashkent teacher asks to add a drawing step to the Pair-Explain Routine in geometry. The network lead checks the recipe card: explanation aloud, partner checklist, listening to five pairs and starting from the common error are all in place in her class. She approves the drawing as an addition and records it in the change log.
Insight. Adding something new after the core is secure is safer than changing a core move.
Common pitfalls
- Writing a core of ten or twelve items, which turns a practice into a script that nobody follows.
- Treating a glowing visit or testimonials as evidence of readiness.
- Skipping Ring 1 because the inventor's own classes did so well.
- Refusing every adaptation, or accepting every adaptation, instead of judging each against the core.
Core, Adaptation and Readiness File
- Pairs (15 min): in Table A, sort the eight features of a sample pilot (the Pair-Explain Routine in Grade 7 mathematics) into CORE, FLEXIBLE or RED LINE and give a reason of a few words.
- Pairs (15 min): in Table B, decide for five adaptation requests from sample schools whether to ACCEPT, MODIFY or REFUSE, and write your one-sentence reply to each school.
- Trios (15 min): in Table C, score the six readiness questions for the sample pilot as YES, PARTLY or NO using the facts given, and agree one of the four verdicts. Be ready to defend it with one fact.
- Part B, alone (20 min): in Table D, write the recipe card and the readiness check for a REAL pilot or practice from your school or cluster (core moves, flexible list, red lines, six answers, verdict, next ring).
- Pairs (10 min): swap cards. Your partner reads only the card and says in three minutes what they would do in a lesson. Note every place where their version differs from your core and fix the card.
- Whole group (10 min): hear three cards and their verdicts, and compare with the answer note. Ask each presenter which red line was hardest to name.
Your facilitator will guide you through this activity in the session.
Questions to discuss
- Which feature of the sample pilot was hardest to classify, and what did you need to know to decide?
- When your partner read only your card, what did they get wrong, and what does that say about the card?
- Which readiness question does your real pilot fail, and what is the cheapest way to collect the missing evidence?
Reflect
- Think of an idea in your school that changed so much while spreading that it lost its point. What was the core?
- Who in your network could run your best practice tomorrow from a one-page card alone?
- Have you ever scaled something because it felt successful, without evidence from anyone but its inventor?
Try it tomorrow
- Watch two lessons of a practice you want to scale and write down one thing the stronger lesson had that the weaker lacked.
- Write the recipe card for that practice and ask one colleague who has never used it to say back what they would do.
- Answer the six readiness questions in writing and mark the one you have the least evidence for.
Further reading and links
Before you scale, write the core on one card, say what may change and what may not, and prove the pilot works without its inventor. Scale in rings and keep an honest 'not yet' available.