Programme / 15.6 Scaling Innovation Across a School Network

Switch course ▾

5.1 Understanding How Students Learn 5.2 Lesson Planning Fundamentals 5.3 Classroom Management 5.4 Student Engagement 5.5 Assessment Fundamentals 5.6 Differentiated Instruction 5.7 Communication with Students & Parents 5.8 Digital Teaching Fundamentals 5.9 AI Literacy for Teachers 5.10 Reflective Teaching Practice 6.1 Multilingual Classrooms: Uzbek, Russian & English Instruction 6.2 Teaching Large Classes Effectively 6.3 Early Literacy & Numeracy Foundations 6.4 Classroom Technology Basics 6.5 Working with Parents in a Multicultural Community 6.6 Wellbeing & Stress Management for New Teachers 8.1 Advanced Assessment & Feedback 8.2 Learning Analytics 8.3 Inquiry-Based Learning 8.4 Project-Based Learning 8.5 Critical Thinking 8.6 Differentiation & Inclusion 8.7 AI-Enhanced Teaching 8.8 Curriculum Design 8.9 Student Motivation 8.10 Action Research 9.1 Teaching STEM through Inquiry 9.2 Arts & Creativity Integration 9.3 Environmental & Sustainability Education 9.4 Blended & Flipped Classroom Design 9.5 Supporting Multilingual Learners 9.6 Gamification & Game-Based Learning 11.1 Instructional Coaching 11.2 Mentoring Beginning Teachers 11.3 Advanced Pedagogical Design 11.4 Evidence-Based Teaching 11.5 Classroom Research 11.6 Learning Analytics for Improvement 11.7 AI Innovation in Education 11.8 Designing Professional Learning 11.9 Peer Observation 11.10 Leading Classroom Innovation 12.1 Coaching Colleagues Through Instructional Rounds 12.2 Data-Informed School Improvement 12.3 Designing Assessment for National/International Benchmarks 12.4 Leading a Professional Learning Community 12.5 AI Tools for Personalized Learning at Scale 12.6 Special Educational Needs & Inclusive Pedagogy 14.1 Educational Leadership 14.2 Instructional Leadership 14.3 Teacher Development Systems 14.4 School Improvement 14.5 Leading Educational Innovation 14.6 Strategic Use of AI 14.7 Education Research & Policy 14.8 Coaching & Mentoring Systems 14.9 Professional Learning Communities 14.10 Change Management in Education 15.1 Education Policy & Systems Thinking 15.2 Building a Teacher Training Curriculum 15.3 Grant Writing & Resourcing School Innovation 15.4 International Benchmarking: AITSL, NIE & PISA Frameworks 15.5 AI Governance & Ethics in Schools 15.6 Scaling Innovation Across a School Network Programme (64) →

Session 2. Scaling Models: Replicate, Cascade, Network

Objective. Compare four scaling models, calculate how much of a practice survives a cascade, and choose and sequence models for your own network.

Why this matters

A model is only a channel; what travels depends on how you run it. Many networks choose a cascade because it is cheap and then wonder why classrooms look different from the training room. Others start a teachers' group that meets happily and changes nothing. In this session you learn how to run a cascade that keeps the core, how to make a community of practice do real work, and how to combine models so that each covers the weakness of another.

Key ideas

Four ways to carry a practice

REPLICATE: a central team brings the package and delivers or coaches it in each school. TRAIN THE TRAINERS (cascade): the team trains a group, who train others. COMMUNITY OF PRACTICE: teachers who share an interest meet regularly and build a shared repertoire. SCHOOL-TO-SCHOOL SUPPORT: a leading school helps others.
Each model trades speed, cost, control of the core, ownership and depth. Replication keeps control but is slow and costly. A cascade is fast and cheap but dilutes. A network builds ownership and depth but takes time to start. Models can be combined.
Match the model to the practice. A simple move with clear materials, such as a five-minute routine, can travel by a short cascade. A practice that needs judgement, such as giving feedback on writing, needs watching real lessons and coaching, so a network or school-to-school support fits better.

The cascade and its dilution problem

David Hayes (2000) studied cascade training for teachers and found that content weakens as it moves down the levels: the further down, the less is understood. He located the problem not in the idea of a cascade but in how it is run: expertise stays at the top and training is transmission of information at every level.
Simple arithmetic shows the danger. ASSUME each handover keeps 80 percent of the core (an assumption, not a measurement). After one handover 80 percent remains, after two about 64, after three about 51. Every extra level multiplies the loss, so fewer levels are safer than many.
Reduce dilution: keep to two levels; give every level the same recipe card and materials, not a summary; train by doing and reflecting, not by telling; let trainers help write the materials; and watch real lessons at the bottom of the cascade, because that is where you learn what arrived.

Communities of practice and school-to-school support

Etienne Wenger and Beverly Wenger-Trayner describe a community of practice by three elements: a shared DOMAIN of interest, a COMMUNITY that learns together through joint activity and mutual help, and a shared PRACTICE, a growing repertoire of stories, tools and ways of solving recurring problems.
In a district this can be a monthly meeting of Grade 5 mathematics teachers from six schools, rotating between hosts, with a shared folder of recipe cards and lesson notes on paper. It works when someone has time to convene it, when the meeting does real work (watch a lesson, compare pupils' work) and when it produces something that outlasts the meeting.
David Hopkins, writing about taking reform to scale, argues that systems must balance top-down direction with bottom-up professional ownership, and that 'it takes capacity to build capacity': strong schools supporting weaker ones is one route. A network without a clear core can drift, so keep the recipe card as its anchor.

Choosing and sequencing

A common sequence is: Ring 2 by replication, with a small core team who visit and learn what really matters; Ring 3 by a short cascade with observation and coaching; then a community or network that keeps the practice alive after the project team steps back. Do not start with the model that is merely the cheapest.
Five questions guide the choice: How much judgement does the practice need? How many schools and how far apart? Who can train and coach, and how many hours have they? How fast must it spread? Who should own it in two years? Write the answers before you pick a model.
Whatever the model, protect one thing: a way to SEE what arrives in classrooms. A visit, a pupil-work sample or a five-minute observation note at the far end tells you more about dilution than any attendance list of trainings.

Three models side by side

Replicate

Central team goes to schools

Keeps control of the core
Slow and costly
Little local ownership

Cascade

Trainers train trainers

Fast and cheap
Dilutes with each level
Needs observation at the end

Community

Teachers learn together

Builds depth and ownership
Slow to start
Needs a convenor and the card

Strengths and weaknesses are typical, not fixed; combine models to cover weak points.

A launch session that trains by doing

A launch of about three hours can work if teachers act more than they listen. A workable plan: 20 minutes to show the recipe card and one filmed or demonstrated lesson; 60 minutes in which teachers practise the core moves in pairs, with the trainer circulating and correcting; 40 minutes to plan their own first lesson on paper; 30 minutes to share plans and name one worry each; 10 minutes to agree the date of the first coaching visit.

Give every trainer the same kit: the card, the same task sets, a script of what to say and what to show, and a short observation guide. Trainers who improvise from their own notes create new versions. Ask each trainer to teach the launch once to a colleague and receive feedback before they teach a group.

A community of practice that does real work

A good meeting has a fixed shape, for example 90 minutes: 10 minutes of check-in, 30 minutes watching a short lesson or looking at pupils' work, 30 minutes comparing it with the recipe card, 15 minutes planning what each member will try, and 5 minutes recording what was decided. Name roles: a convenor, a host school and a note-taker, and rotate the host so that no school carries all the travel.

Communities drift when they talk about ideas but never look at lessons or pupils' work. Keep the card as the anchor: each meeting ends with a note of what was seen against the core and what was learned about the flexible parts. Share the notes on paper with the steering group so that learning travels beyond the room.

Examples from the classroom

Two levels and an observation

Scene. In Namangan, the district trains 12 school trainers for one full day, with 90 minutes of practice in pairs. Each trainer then runs a three-hour launch for their staff using the same card and kit. In the sixth week, three lead teachers observe five minutes in each of 20 classes. They see four of five core moves in 13 classes and ask the trainers of the other schools to repeat the practice part.

Insight. A short cascade plus observation at the bottom lets you see dilution and repair it.

Six schools, one monthly meeting

Scene. Grade 5 mathematics teachers from six schools around Bukhara meet on the first Saturday of each month, rotating hosts. Dilnoza convenes with 2 release hours a month. They watch a ten-minute video of a lesson that was filmed with the teacher's permission, compare it with the card, and each member writes one thing to try. After four months they have a paper folder of 16 improved task sets.

Insight. A community lasts when it does real work, has a convenor with hours and produces something.

A lead school with limits

Scene. A well-regarded school in Karshi agrees to support two small village schools. Its head, Jasur, finds that the two lead teachers also have full timetables. The network lead agrees four visits a term, each of two hours, funds the travel for them and asks the village schools to host the second visit, so that support goes both ways.

Insight. School-to-school support needs time and travel to be planned, and it should not overload the strongest school.

Common pitfalls

Letting each trainer prepare their own version of the launch from personal notes.
Starting a community with a nice name but no convenor, no fixed rhythm and no task.
Choosing a cascade because it is cheap without planning any way to see the last classroom.
Using the strongest school as the helper for everyone until its own teachers burn out.

Hands-on activity · ~85 min

Model Choice Workshop

Pairs (15 min): in Table A, rate each of the four models for speed, cost, dilution risk and ownership (high, medium or low) for a sample district of 14 schools, and write one condition under which each model fails.
Pairs (15 min): in Table B, calculate how much of the core reaches the classroom teacher in four sample cascades, then write one fix for the worst case.
Trios (15 min): in Table C, write a one-page charter for a community of practice of Grade 5 mathematics teachers from six schools: domain, members, rhythm, convenor, product and a rule for review.
Part B, alone (25 min): in Table D, choose and sequence models for YOUR network and practice from session 1: the five answers, the models by ring, who trains, and how you will see what arrives.
Pairs (10 min): exchange tables. Your partner plays a sceptical district head and asks: 'What if your two best trainers leave?' Note the answer and add a fix.
Whole group (5 min): hear two plans and compare with the answer note.

Your facilitator will guide you through this activity in the session.

Questions to discuss

Which model did most of the group choose first, and what did that choice assume about the practice?
In your own network, how many handovers are there between the people who know the core and the pupils?
What is the cheapest way for you to see what actually arrives in the last classroom?

Reflect

Think of a training you attended that was passed down two or three levels. What was left by the time it reached you?
Is there a group of teachers in your area who already meet regularly and could become a community of practice?
Which of your trainers or lead teachers is already overloaded?

Try it tomorrow

Draw your own cascade with the number of handovers, and mark the point where nobody checks what arrives.
Plan a 90-minute meeting with the five-part shape for a group of teachers you already work with.
Ask one trainer to teach your launch to a colleague, and note what changed in the retelling.